Nothing Special   »   [go: up one dir, main page]

WO2023028774A1 - 一种激光雷达的标定方法、装置及存储介质 - Google Patents

一种激光雷达的标定方法、装置及存储介质 Download PDF

Info

Publication number
WO2023028774A1
WO2023028774A1 PCT/CN2021/115418 CN2021115418W WO2023028774A1 WO 2023028774 A1 WO2023028774 A1 WO 2023028774A1 CN 2021115418 W CN2021115418 W CN 2021115418W WO 2023028774 A1 WO2023028774 A1 WO 2023028774A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
lidar
marker
laser radar
vehicle
Prior art date
Application number
PCT/CN2021/115418
Other languages
English (en)
French (fr)
Inventor
冯超
李帅君
何启盛
文坤
张建军
霍梦晨
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to CN202180006105.6A priority Critical patent/CN114829971A/zh
Priority to PCT/CN2021/115418 priority patent/WO2023028774A1/zh
Publication of WO2023028774A1 publication Critical patent/WO2023028774A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
    • G01S7/497—Means for monitoring or calibrating

Definitions

  • the present application relates to the technical field of intelligent driving, and in particular to a laser radar calibration method, device and storage medium.
  • lidar In the field of intelligent driving, lidar is an indispensable part of realizing high-level automatic driving functions.
  • the calibration accuracy of the external parameters of lidar plays an important role in realizing functions such as perception, positioning or fusion, and ensuring vehicle safety.
  • the existing methods of calibrating LiDAR have low calibration accuracy, high requirements on the site, and high calibration cost; in the calibration process, they rely on manual operation or map construction, etc., and the calibration efficiency is low.
  • the embodiment of the present application provides a laser radar calibration method, the method comprising: acquiring the point cloud collected by the laser radar when the vehicle passes the target road, at least one side of the target road is provided with markers; according to The preset threshold value is used to preliminarily screen the collected point cloud; the preset threshold value is determined by the installation height of the laser radar; multiple fitting processes are performed on the preliminarily screened point cloud to obtain a ground point cloud; Extracting the marker point cloud in the collected point cloud; calibrating the external parameters of the lidar according to the marker point cloud and the ground point cloud.
  • the markers can be the curbs and railings of the road, and there are no special requirements for the site, and no additional calibration boards, targets, reflective stickers, etc. are required, which reduces the calibration cost and can be used on open roads (such as urban streets) , expressway, etc.), the online dynamic calibration of lidar can be completed by using the natural scene of the road.
  • the external parameters of the lidar are calibrated, so as to realize fully automatic online dynamic calibration without manual operation, and improve calibrated efficiency.
  • the collected point cloud is preliminarily screened according to the preset threshold; the preliminarily screened point cloud is subjected to multiple fitting processing, so that based on threshold filtering and multiple fitting processing, adaptive High-precision ground point cloud can be extracted accurately; it can also further automatically extract high-precision marker point cloud through point cloud slicing, thereby improving the external parameter accuracy of the calibrated lidar.
  • the method further includes: according to the calibrated external parameters of a plurality of laser radars, obtaining the points of the marker point clouds corresponding to each laser radar Position information and the position information of the ground point cloud; according to the position information of the marker point cloud corresponding to each lidar and the position information of the ground point cloud, the intersection feature point or the intersection domain is obtained, wherein the intersection domain represents parallel to The traveling direction of the vehicle, the area centered on the intersection feature point; according to the intersection feature point or intersection area, optimize the calibrated extrinsic parameters of any one of the plurality of lidars.
  • the position information of the marker point cloud corresponding to each laser radar and the position information of the ground point cloud are obtained, and then the external parameters of any laser radar are optimized by extracting cross feature points and cross domain features, which further improves the performance of the laser radar. Extrinsic precision.
  • the multiple laser radars include a master laser radar and a slave laser radar, wherein the master laser radar The radar is used to scan the environment in front of the vehicle, and the secondary laser radar is used to scan the side and/or rear environment of the vehicle; according to the calibrated external parameters of multiple laser radars, each laser radar is obtained
  • the position information of the corresponding marker point cloud and the position information of the ground point cloud including: according to the calibrated external parameters of the main lidar, the marker point cloud and the ground point cloud corresponding to the main lidar are converted to the car body coordinate system , obtain the position information of the marker point cloud corresponding to the main laser radar and the position information of the ground point cloud; according to the calibrated external parameters of the slave laser radar, the marker point cloud corresponding to the slave laser radar and the ground
  • the point cloud is converted into the vehicle body coordinate system, and the position information of the marker point cloud corresponding to the laser radar and the position information of the ground point cloud are obtained; the position information
  • the intersection feature point or the intersection domain including: according to the position information of the ground point cloud corresponding to the main laser radar and the position information of the ground point cloud corresponding to the slave laser radar, obtain the first Intersection feature point or first intersection domain; according to the position information of the marker point cloud corresponding to the main lidar and the position information of the marker point cloud corresponding to the slave lidar, the second intersection feature point or the second intersection feature point is obtained Domain; according to the intersection feature point or intersection domain, optimizing the calibrated extrinsic parameters of any lidar in the plurality of laser radars includes: according to the first intersection feature point or first intersection domain, Optimizing the pitch angle and roll angle calibrated from the lidar; optimizing the yaw angle calibrated from the lidar according to the second intersection feature point or the second intersection domain.
  • both the main laser point cloud and the side laser point cloud are converted to the vehicle body coordinate system through the corresponding calibrated external parameters, and then on the basis of the marker point cloud, the intersection feature points and the cross domain optimization compensation side are extracted.
  • the joint optimization of the system makes the external parameters of the lidar to the car body coordinate system more accurate.
  • the external parameters include at least one of pitch angle, roll angle, and yaw angle
  • the calibration of the external parameters of the laser radar according to the point cloud of the marker and the point cloud of the ground includes: calibration of the pitch angle and roll angle of the laser radar according to the point cloud of the ground ; According to the marker point cloud, calibrate the yaw angle of the lidar.
  • the extracting the marker point cloud in the collected point cloud includes: Filter out the ground point cloud in the collected point cloud; divide the filtered point cloud into a plurality of slices along the direction perpendicular to the vehicle's travel; extract the marker point cloud, and the marker point cloud includes feature points in a slice set that meets a preset condition, wherein the slice set includes one or more adjacent target slices, and the number of feature points in the target slice exceeds a threshold.
  • the method further includes: acquiring the first wire harness information of the ground point cloud, according to Perform downsampling processing on the ground point cloud for the first wire harness information; and/or, acquire second wire harness information of the marker point cloud, and perform downsampling on the marker point cloud according to the second wire harness information Sampling processing: according to the marker point cloud and the ground point cloud, the external parameters of the lidar are calibrated, including: according to the marker point cloud and the ground point cloud after the down-sampling process, the The external parameters of the lidar are calibrated.
  • the down-sampling process extracts accurate ground points, while improving the processing efficiency, the texture structure of the ground is fully preserved, thereby ensuring the accuracy of the ground point cloud.
  • the harness information of each feature point down-sampling is used to extract accurate feature point clouds; while improving processing efficiency, the texture structure of markers is fully preserved, thereby ensuring the accuracy of marker point clouds.
  • the acquired point cloud is a vehicle in a state of driving along a straight line, collected by a laser radar point cloud.
  • the markers on both sides of the road are parallel to the direction of the vehicle, and the point cloud collected by the lidar in this state is used for calibration, thereby improving the accuracy of external parameters such as the yaw angle of the lidar .
  • the markers include at least one of roadsides, guardrails, and buildings .
  • the markers can be roadsides, guardrails, buildings, etc. There is no special requirement for the site, and no additional calibration boards, targets, reflective stickers, etc. are required, which reduces the cost of calibration. On open roads (such as city streets, highways, etc.), the online calibration can be completed.
  • the embodiment of the present application provides a laser radar calibration method, the method comprising: obtaining the point cloud collected by the laser radar when the vehicle passes through the target area; at least one side of the target area is vertically provided with markers ; extract the point cloud of the marker in the collected point cloud; according to the point cloud of the marker, obtain the fitting line information of the marker; the fitting line information includes the position and direction information of the fitting line; according to the fitting Line information to obtain the value of the lidar extrinsic parameters.
  • the markers are easily set up, the requirements on the site are reduced, and the construction cost is low.
  • the fitting line information of the marker is obtained.
  • the fitting line based on the vertical marker needs to meet the vertical constraint, and the value of the external parameter of the lidar is obtained.
  • the laser can be calculated The value of the radar extrinsic parameters.
  • the calibration process does not depend on the ground point cloud, it can be applied to scenarios with insufficient ground information (for example, the lidar with a small vertical field of view cannot collect nearby ground point clouds, and the side lidar due to the limited size of the site) Unable to collect effective ground point clouds, installation of lidar with too large pitch angle upwards resulting in missing or less ground point clouds, etc.), realized high-precision calibration of single lidar in scenes with insufficient ground information.
  • the entire calibration process can be automatically performed, which improves the efficiency of single lidar calibration.
  • the external parameter includes a pitch angle
  • the method further includes: when the included angle between the lidar orientation and the vertical upward direction is less than In the case of the first preset threshold, and the value of the pitch angle is greater than the second preset threshold, the external parameters of the lidar are calibrated according to the marker point cloud, wherein the second preset threshold Determined by the vertical field of view of the lidar.
  • the external parameter includes a yaw angle
  • the method further includes: acquiring the position of the vehicle information and the position information of the laser radar; according to the position information of the vehicle and the position information of the laser radar, determine the heading angle of the vehicle; according to the heading angle, optimize the deflection of the laser radar after calibration flight angle.
  • the calibrated yaw angle can be optimized by combining the vehicle motion information without limiting the vehicle's driving deflection angle.
  • the vehicle is equipped with a master laser radar and a slave laser radar, wherein the master laser radar uses For scanning the front environment of the vehicle, the slave laser radar is used to scan the side and/or rear environment of the vehicle; the method also includes: according to the calibrated external parameters of the master laser radar and the master laser Determining the location information of the plurality of markers from the point cloud of the plurality of markers collected by the radar; obtaining the predicted position of the first marker according to the position information of the plurality of markers; Referring to the point cloud of the first marker collected from the laser radar, the measurement position of the first marker is obtained; by comparing the predicted position and the measurement position, optimizing the external parameters of the slave laser radar .
  • the method further includes: extracting the ground point cloud from the collected point cloud;
  • the marker point cloud, calibrating the extrinsic parameters of the lidar further includes: calibrating the extrinsic parameters of the lidar according to the marker point cloud and the ground point cloud.
  • the ground point cloud when there is an effective ground point cloud, the ground point cloud can be fully utilized, thereby further improving the calibration accuracy and stability.
  • the fitting line information of the marker is obtained according to the point cloud of the marker, Including: determining the initial value of the rotation angle according to the point cloud of the marker, the initial value of the rotation angle makes the projection area of the horizontal plane in the laser radar coordinate system the smallest after the initial value of the rotation angle rotates the point cloud of the marker; according to the initial value of the rotation angle , performing rotation processing on the marker point cloud; using the rotated marker point cloud to obtain the fitting line information of the marker.
  • the method further includes: when the angle between the lidar orientation and the vertical upward direction is less than In the case of the first preset threshold and the value of the pitch angle is not greater than the second preset threshold, the external parameters of the lidar are calibrated according to the marker point cloud and the ground point cloud.
  • the ground point cloud when there is an effective ground point cloud, the ground point cloud can be fully utilized, thereby further improving the calibration accuracy and stability.
  • the external parameters include at least one of a pitch angle, a roll angle, and a yaw angle.
  • the method further includes: when the angle between the lidar orientation and the vertical upward direction is smaller than a first preset threshold, and the value of the pitch angle is not greater than the second preset threshold, According to the ground point cloud, the pitch angle and roll angle of the lidar are calibrated; according to the position information of the fitting line, the yaw angle of the lidar is calibrated.
  • the pitch angle and roll angle of the laser radar can be calibrated by using the ground point cloud, which improves the accuracy and stability of the calibrated pitch angle and roll angle; using the position information of the fitting line, the laser radar The yaw angle is calibrated to improve the accuracy of the calibrated yaw angle.
  • At least one side of the target area is vertically provided with a plurality of markers, and The intersections of the multiple markers and the ground are on the same straight line.
  • the embodiment of the present application provides a laser radar calibration device, the device includes: an acquisition module, used to acquire the point cloud collected by the laser radar when the vehicle passes the target road, at least one side of the target road is set There are markers; the screening module is used for preliminary screening of the collected point cloud according to a preset threshold; the preset threshold is determined by the installation height of the laser radar; the first extraction module is used for the preliminary screening The screened point cloud is subjected to multiple fitting processing to obtain the ground point cloud; the second extraction module is used to extract the marker point cloud in the collected point cloud; the calibration module is used to extract the marker point cloud according to the marker point cloud and The ground point cloud calibrates the external parameters of the lidar.
  • the device further includes: a conversion module, configured to obtain the corresponding The position information of the marker point cloud and the position information of the ground point cloud; the third extraction module is used to obtain the intersection feature point or An intersection area, wherein the intersection area represents an area parallel to the traveling direction of the vehicle and centered on the intersection feature point; an optimization module is configured to optimize the multiple The external parameters of any laser radar in the laser radar are optimized after calibration.
  • a conversion module configured to obtain the corresponding The position information of the marker point cloud and the position information of the ground point cloud
  • the third extraction module is used to obtain the intersection feature point or An intersection area, wherein the intersection area represents an area parallel to the traveling direction of the vehicle and centered on the intersection feature point
  • an optimization module is configured to optimize the multiple The external parameters of any laser radar in the laser radar are optimized after calibration.
  • the multiple laser radars include a master laser radar and a slave laser radar, wherein the master laser radar The radar is used to scan the environment in front of the vehicle, and the slave laser radar is used to scan the side and/or rear environment of the vehicle; the conversion module is also used to: according to the calibrated exterior of the main laser radar Parameters, transform the marker point cloud and ground point cloud corresponding to the main laser radar into the vehicle body coordinate system, and obtain the position information of the marker point cloud and the ground point cloud corresponding to the main laser radar; The calibrated external parameters of the radar are used to convert the marker point cloud and the ground point cloud corresponding to the laser radar into the vehicle body coordinate system, and obtain the position information and the ground point cloud of the marker point cloud corresponding to the laser radar The position information of the cloud; the third extraction module is also used to obtain the first intersection feature according to the position information of the ground point cloud corresponding to the main lidar and the position information of the ground point cloud corresponding to the
  • the optimization module is further configured to: optimize the pitch angle and roll angle calibrated from the laser radar according to the first intersection feature point or the first intersection area; Two intersection domains, optimize the yaw angle calibrated from the lidar.
  • the external parameters include at least one of pitch angle, roll angle, and yaw angle
  • the calibration module is further configured to: calibrate the pitch angle and roll angle of the lidar according to the ground point cloud; and calibrate the yaw angle of the lidar according to the marker point cloud.
  • the second extraction module is further configured to: in the collected point cloud Filter out the ground point cloud; divide the filtered point cloud into multiple slices along the direction perpendicular to the vehicle's travel; extract the marker point cloud, and the marker point cloud includes slices that meet the preset conditions feature points in the set, wherein the set of slices includes one or more adjacent target slices, and the number of feature points in the target slices exceeds a threshold.
  • the device further includes a downsampling module, configured to: acquire ground point cloud The first wire harness information, according to the first wire harness information, perform down-sampling processing on the ground point cloud; and/or, acquire the second wire harness information of the marker point cloud, and according to the second wire harness information, perform the down-sampling process on the ground point cloud;
  • the marker point cloud is subjected to downsampling processing;
  • the calibration module is further configured to calibrate the external parameters of the lidar according to the downsampled marker point cloud and the ground point cloud.
  • the acquired point cloud is a vehicle that is collected by a laser radar when the vehicle is driving in a straight line. point cloud.
  • the markers include at least one of roadsides, guardrails, and buildings .
  • an embodiment of the present application provides a laser radar calibration device, the device includes: an acquisition module, configured to acquire a point cloud collected by the laser radar when a vehicle passes through a target area; at least one side of the target area is vertically Markers are directly set; the extraction module is used to extract the marker point cloud in the collected point cloud; the fitting module is used to obtain the fitting line information of the marker according to the marker point cloud; the fitting line The information includes the position and direction information of the fitting line; the calculation module is used to obtain the value of the external parameter of the lidar according to the information of the fitting line.
  • the external parameter includes a pitch angle
  • the device further includes: a calibration module, configured to adjust the orientation and vertical direction of the lidar
  • the external parameters of the lidar are calibrated according to the marker point cloud, wherein the first The two preset thresholds are determined by the vertical field of view of the lidar.
  • the external parameter includes a yaw angle
  • the device further includes: an optimization module, configured to obtain The position information of the vehicle and the position information of the laser radar; according to the position information of the vehicle and the position information of the laser radar, determine the heading angle of the vehicle; according to the heading angle, optimize the calibrated The yaw angle of the lidar.
  • the vehicle is equipped with a master laser radar and a slave laser radar, wherein the master laser radar uses For scanning the front environment of the vehicle, the slave laser radar is used to scan the side and/or rear environment of the vehicle; the device also includes: a determination module, used for and the multiple marker point clouds collected by the main lidar, determining the position information of the multiple markers; the prediction module is used to obtain the predicted position of the first marker according to the position information of the multiple markers The measurement module is used to obtain the measurement position of the first marker according to the calibrated external reference from the laser radar and the first marker point cloud collected from the laser radar; the matching module is used to pass For the predicted position and the measured position, optimize the extrinsic parameters from the laser radar.
  • the master laser radar uses For scanning the front environment of the vehicle, the slave laser radar is used to scan the side and/or rear environment of the vehicle
  • the device also includes: a determination module, used for and the multiple marker point clouds collected by the main lidar, determining the position information of the multiple markers; the prediction module is used to obtain
  • the extraction module is further configured to: extract the ground point cloud in the collected point cloud;
  • the calibration module is further configured to: calibrate the external parameters of the lidar according to the marker point cloud and the ground point cloud.
  • the fitting module is further configured to: determine The initial value of the rotation angle, after the initial value of the rotation angle rotates the marker point cloud, the projection area of the horizontal plane in the lidar coordinate system is the smallest; according to the initial value of the rotation angle, the marker point cloud is rotated ; Using the rotated marker point cloud to obtain the fitting line information of the marker.
  • the calibration module is further configured to: When the included angle in the vertical direction is less than the first preset threshold and the value of the pitch angle is not greater than the second preset threshold, according to the marker point cloud and the ground point cloud, the outer surface of the lidar Refer to calibration.
  • the external parameters include at least one of pitch angle, roll angle, and yaw angle.
  • the calibration module is also used for: the angle between the lidar orientation and the vertical upward direction is less than a first preset threshold, and the value of the pitch angle is not greater than the second preset threshold In this case, the pitch angle and roll angle of the lidar are calibrated according to the ground point cloud; the yaw angle of the lidar is calibrated according to the position information of the fitting line.
  • At least one side of the target area is vertically provided with a plurality of markers, and The intersections of the multiple markers and the ground are on the same straight line.
  • an embodiment of the present application provides a laser radar calibration device, including: at least one processor; a memory for storing processor-executable instructions; wherein the at least one processor is configured to execute the
  • the instructions are mentioned above, the first aspect or one or more lidar calibration methods of the first aspect are realized, or the second aspect or one or more lidar calibration methods of the second aspect are realized.
  • the embodiments of the present application provide a non-volatile computer-readable storage medium, on which computer program instructions are stored, wherein, when the computer program instructions are executed by a processor, the first aspect or One or more laser radar calibration methods in the first aspect, or, realize the above-mentioned second aspect or one or more laser radar calibration methods in the second aspect.
  • the embodiments of the present application provide a computer program product, including computer readable code, or a non-volatile computer readable storage medium bearing computer readable code, when the computer readable code is stored in an electronic
  • the processor in the electronic device implements the first aspect or one or more lidar calibration methods of the first aspect, or implements the second aspect or one or more of the second aspects LiDAR calibration method.
  • FIG. 1 shows a schematic diagram of an application scenario of a lidar calibration method according to an embodiment of the present application
  • Figure 2(a)- Figure 2(b) shows a schematic diagram of the road where the vehicle is located according to an embodiment of the present application
  • Fig. 3 shows a schematic diagram of an application scenario of a lidar calibration method according to an embodiment of the present application
  • Figure 4(a)- Figure 4(e) shows a schematic diagram of the road where the vehicle is located according to an embodiment of the present application
  • Fig. 5 (a)-Fig. 5 (d) have shown the schematic diagram of several kinds of coordinate systems according to an embodiment of the present application;
  • FIG. 6 shows a flow chart of a lidar calibration method according to an embodiment of the present application
  • Fig. 7 shows a schematic diagram of a point cloud slice according to an embodiment of the present application.
  • FIG. 8 shows a flow chart of a lidar calibration method according to an embodiment of the present application.
  • FIG. 9 shows a schematic diagram of time synchronization according to an embodiment of the present application.
  • Fig. 10 shows a schematic diagram of an intersection feature point according to an embodiment of the present application.
  • Fig. 11 shows a schematic diagram of a cross domain according to an embodiment of the present application.
  • FIG. 12 shows a flow chart of another lidar calibration method according to an embodiment of the present application.
  • Fig. 13 shows a schematic diagram of a single lidar calibration according to an embodiment of the present application
  • Fig. 14 shows a schematic diagram of a lidar scanning scene according to an embodiment of the present application
  • Figure 15(a)- Figure 15(b) shows a schematic diagram of a production line environment according to an embodiment of the present application
  • Fig. 16 shows a comparative schematic diagram of pitch angle calibration according to an embodiment of the present application
  • FIG. 17 shows a flowchart of a lidar calibration method according to an embodiment of the present application.
  • Fig. 19 shows a structural diagram of a lidar calibration device according to an embodiment of the present application.
  • Fig. 20 shows a structural diagram of a lidar calibration device according to an embodiment of the present application
  • Fig. 21 shows a schematic structural diagram of a lidar calibration device according to an embodiment of the present application.
  • Fig. 1 shows a schematic diagram of an application scenario of a lidar calibration method according to an embodiment of the present application.
  • the application scenario may include: a vehicle 101, a road 102, a laser radar 103, and a marker 104; wherein, the laser radar 103 is installed on the vehicle 101, and the marker 104 may be a roadside, a road fence, a building facade etc.
  • the road 103 may be an open road.
  • road 103 can be an urban road, and this urban road is comparatively smooth, and both sides have curb or guardrail of certain height;
  • road 103 can be an expressway, and the road surface of this expressway is relatively smooth, The curb is clearer.
  • the ego vehicle 101 can run on the road shown in Fig. 2(a) or Fig. 2(b) above, the speed of the ego vehicle can be greater than 40km/h, and keep running straight for a period of time.
  • the radar 103 scans the external environment of the self-vehicle to complete the point cloud collection work.
  • the point cloud collected by the laser radar 103 is processed by the laser radar calibration method in the embodiment of the present application (see the following for a detailed description), so as to realize the external parameter calibration of the laser radar 103 .
  • Fig. 3 shows a schematic diagram of an application scenario of a lidar calibration method according to an embodiment of the present application.
  • the application scenario may include: a vehicle 301, a road 302, a laser radar 303, and a marker 304; wherein, the laser radar 303 is installed on the vehicle 301, and the marker 304 may be vertically arranged on the road 303,
  • the road 303 may be a dedicated road with a length of 30-100m and a width of 3-8m;
  • the marker 304 may be made of a metal material, or a reflective sticker may be attached on the surface of the marker 304 to improve laser Reflectivity;
  • the cross-section of the marker 304 can be circular, square, triangular, etc., and the geometric characteristics of the cross-section are known.
  • the cross-section type of the vertical rod can be a circular cross-section, and the radius is known, and the surface of the vertical rod can be pasted with reflective stickers;
  • the number of markers 304 can be multiple, Marker 304 can be arranged on one side or both sides of road 302, when a plurality of markers 304 are arranged on any side of road 302, the intersection point of each marker 304 and ground is on the same straight line, and the spacing between each marker 304 can be Same or different.
  • Figure 4(a)- Figure 4(e) shows a schematic diagram of the road where the vehicle is located according to an embodiment of the present application, as shown in Figure 4(a), a row of markers can be set on one side of the road (In the figure, the markers are set on the left side of the road as an example), as shown in Figure 4(b), a row of markers is set on both sides of the road, and the two rows of markers are symmetrically distributed along the center line of the road, and any row
  • the spacing between adjacent markers in the markers is the same; as shown in Figure 4(c), a row of markers is set on both sides of the road, and the two rows of markers are distributed in parallel, and the distance between adjacent markers in any row of markers The spacing is the same; as shown in Figure 4(d), a row of markers is set on both sides of the road, and the two rows of markers are distributed in parallel, and the adjacent markers in any row of markers are not exactly the same; Figure 4( As shown in e), the road surface of the road may be uneven, and the road may
  • the ego vehicle 301 can drive in at one end of the road shown in any of the above-mentioned Fig. 4(a)-Fig. h, it can maintain a constant speed, and can also maintain a straight-line driving state.
  • the laser radar 303 scans the external environment of the vehicle to complete the point cloud collection work.
  • the point cloud collected by the laser radar 303 is processed by the laser radar calibration method in the embodiment of the present application (see below for a detailed description), so as to realize the external parameter calibration of the laser radar 303 .
  • the embodiment of the present application does not limit the number and type of lidars installed on the vehicle.
  • the number of laser radar 103 or laser radar 303 can be one or more, and the above-mentioned figures 1 and 3 all take two laser radars as an example, and more laser radars can be installed on the vehicle 101 according to actual needs 103, or install more laser radars 303 on the ego vehicle 301.
  • the above-mentioned laser radar 103 or laser radar 303 may include a main laser radar, which is used to detect the environment in front of the vehicle, or to detect the environment behind the vehicle, or to detect the environment around the vehicle; it may also include a slave laser radar, a slave laser radar Radar can be used to detect the side environment of the vehicle (also known as side lidar), or to detect the environment behind the vehicle (also known as rear lidar); compared with the slave lidar, the main lidar can detect obstacles in front of the vehicle .
  • a main laser radar which is used to detect the environment in front of the vehicle, or to detect the environment behind the vehicle, or to detect the environment around the vehicle
  • a slave laser radar Radar can be used to detect the side environment of the vehicle (also known as side lidar), or to detect the environment behind the vehicle (also known as rear lidar); compared with the slave lidar, the main lidar can detect obstacles in front of the vehicle .
  • the aforementioned self-vehicle 101 or self-vehicle 301 may also include a positioning device (not shown in the figure), which may include a wheel speedometer, a satellite navigation system (Global Navigation Satellite System, GNSS), an inertial navigation system (Inertial Navigation System, INS), etc., used to obtain the pose information of the vehicle.
  • a positioning device not shown in the figure
  • GNSS Global Navigation Satellite System
  • INS Inertial Navigation System
  • the application scenario shown in FIG. 1 or FIG. 3 may also include a laser radar calibration device (not shown in the figure), and the laser radar calibration method provided in the embodiment of the present application may be implemented by the laser radar calibration device, using It is used to perform efficient automatic data processing on the point cloud collected by the above-mentioned laser radar 103 or laser radar 303 , and the external parameter accuracy of the calibrated laser radar is relatively high.
  • the embodiment of the present application does not limit the type of the lidar calibration device.
  • the lidar calibration device can be the above-mentioned self-vehicle 101 (or self-vehicle 301), or other components with data processing functions in the self-vehicle 101 (or self-vehicle 301), such as: vehicle-mounted terminal, vehicle-mounted control Vehicle-mounted terminal, vehicle-mounted controller, vehicle-mounted module, vehicle-mounted module, vehicle-mounted component, vehicle-mounted chip, vehicle-mounted unit, vehicle-mounted sensor wait.
  • the lidar calibration device is integrated in the self-vehicle 101 or the self-vehicle 301 in an automatic driving system (Automated Driving System, ADS) or an advanced driver assistant system (Advanced Driver Assistant Systems, ADAS) or a vehicle-mounted computing platform.
  • ADS Automatic Driving System
  • ADAS Advanced Driver Assistant Systems
  • the lidar calibration device may also be other smart terminals with data processing capabilities other than the vehicle, or components or chips set in the smart terminals.
  • the smart terminal may be a device equipped with a laser radar, such as a smart transportation device, a smart wearable device, a smart home device, a smart auxiliary aircraft, a robot, or an unmanned aerial vehicle.
  • the lidar calibration device may be a general device or a special device.
  • the device can also be a desktop computer, a portable computer, a network server, a handheld computer (personal digital assistant, PDA), a mobile phone, a tablet computer, a wireless terminal device, an embedded device or other devices with data processing functions, Or components or chips within these devices.
  • PDA personal digital assistant
  • the laser radar calibration device may also be a chip or a processor with processing functions, and the laser radar calibration device may include multiple processors.
  • the processor may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor.
  • the chip or processor with processing functions may be set in the laser radar, or may not be set in the laser radar, but set at the receiving end of the output signal of the laser radar.
  • Fig. 5 (a)-Fig. 5 (d) shows the schematic diagram of several coordinate systems according to an embodiment of the present application
  • Fig. 5 (a)-Fig. 5 (b) shows the car body coordinate system
  • car body coordinates car body coordinates
  • the origin of the system can coincide with the center of mass of the vehicle.
  • the x-axis is parallel to the ground and points to the front of the vehicle
  • the z-axis points upward through the center of mass of the car
  • the y-axis points to the left of the driver.
  • the external parameters of the lidar may include: one or more of the pitch angle (pitch), the roll angle (roll), and the yaw angle (yaw), and may also include the installation height of the lidar, etc.; wherein, The pitch angle represents the angle of rotation around the y-axis, the yaw angle represents the angle of rotation around the z-axis, and the roll angle represents the angle of rotation around the x-axis.
  • Figure 5(c) shows the lidar coordinate system. The origin of the lidar coordinate system can coincide with the lidar centroid.
  • Figure 5(d) shows the world coordinate system. It can be seen from the X'Y' plane that the lidar is in The yaw angle of the car body coordinate system, the vehicle heading angle and the laser radar heading angle.
  • the lidar calibration method provided by the embodiment of the present application will be described in detail below based on the application scenario described in FIG. 1 above.
  • Fig. 6 shows a flow chart of a lidar calibration method according to an embodiment of the present application; the method may be performed by the above lidar calibration device; as shown in Fig. 6, the method may include the following steps:
  • Step 601. Obtain the point cloud collected by the lidar when the vehicle passes the target road, at least one side of the target road is provided with markers.
  • the laser radar may be any laser radar installed on the vehicle, for example, it may be a main laser radar or a side laser radar.
  • the target road can be a smooth open road
  • the marker can be a curb, a guardrail or a building on one or both sides of the open road
  • the vehicle can be the above-mentioned self-car 101
  • the target road can be the above-mentioned FIG. 2 ( a) or the road shown in Figure 2(b).
  • the acquired point cloud may be a point cloud collected by a lidar when the vehicle is driving in a straight line.
  • it may be a point cloud collected by the lidar 103 when the ego vehicle 101 is driving straight on the road shown in FIG. 2( a ) or FIG. 2( b ).
  • the driving state of the vehicle can be determined according to the pose information of the vehicle when it passes the target road; for example, the point cloud collected by the on-board lidar can be extracted when the vehicle is determined to be going straight through the target road through the inertial navigation system.
  • Step 602 Preliminarily screen the collected point cloud according to the preset threshold.
  • the preset threshold is determined by the installation height of the laser radar, wherein the installation height of the laser radar can be predetermined; for example, the preset threshold can be equal to the installation height of the laser radar.
  • the collected point cloud includes the coordinate information (X value, Y value, Z value) of each scanning point under the lidar coordinates; when the scanning point is a ground point, the corresponding height value (ie Z value) is usually is smaller and will not exceed the installation height of the lidar; therefore, by judging whether the height value of each scanning point in the collected point cloud exceeds the preset threshold, the scanning points whose height value does not exceed the preset threshold can be screened out. All scanned points can be regarded as coarse-grained ground point cloud.
  • any frame point cloud obtained in the above step 601 is preliminarily screened, and the coarse-grained ground point cloud is screened out, thereby reducing the number of point clouds and improving the data quality. Processing efficiency.
  • Step 603 performing multiple fitting processes on the preliminarily screened point cloud to obtain a ground point cloud.
  • a multi-step random sampling consistent fitting algorithm may be used to adaptively extract high-precision ground points.
  • the random sampling consensus algorithm is used to fit the plane; the scanning points in the coarse-grained ground point cloud whose distance from the fitting plane is greater than the preset distance threshold are filtered out to obtain Medium-grained ground point cloud, where the initial value of the preset distance threshold can be an empirical value; reduce the above-mentioned distance threshold, continue to perform the above-mentioned step of fitting the plane using the random sampling consensus algorithm, and further filter out the middle distance of the medium-grained ground point cloud The above-mentioned fitted plane distance is larger than the scanned point after the reduced distance threshold, thereby obtaining a fine-grained ground point cloud. It can be understood that the operation of narrowing the distance threshold and fitting the plane using the random sampling consensus algorithm can be further performed through multiple iterations as required, so as to obtain a more accurate ground point cloud.
  • the first wire harness information of the ground point cloud may also be acquired, and the ground point cloud may be down-sampled according to the first wire harness information.
  • the collected point cloud may include the wire harness information of each feature point, or, based on the configuration parameters of the lidar, the wire harness information of each feature point in the collected point cloud may be determined.
  • threshold filtering and multiple fitting processes are used to automatically extract accurate ground point clouds, and the final ground point cloud can also be obtained by down-sampling according to the wire harness information of the ground points.
  • Step 604 extracting the marker point cloud in the collected point cloud.
  • this step may include: filtering out ground point clouds from the collected point clouds.
  • the scanned points in the filtered point cloud can include feature points of landmarks, or non-ground points such as vehicles and pedestrians in the road; along the direction perpendicular to the vehicle's travel, the filtered point cloud is divided into multiple slices.
  • the marker point cloud is extracted, and the marker point cloud includes feature points in a slice set satisfying a preset condition, wherein the slice set includes one or more adjacent target slices, and the number of feature points in the target slice exceeds a preset threshold. In this way, by dividing the point cloud slices and extracting the marker point cloud, the automatic extraction of high-precision marker point cloud is realized.
  • the width and preset threshold of each slice can be set as required; the preset condition can include that along the traveling direction of the vehicle, the positions of all feature points in the slice set exceed the length range.
  • FIG. 7 shows a schematic diagram of a point cloud slice according to an embodiment of the present application; as shown in FIG. 7, taking the main laser radar installed on a vehicle as an example, the scanning points in FIG. 7 are: Filter out the ground point cloud from a frame of point cloud collected by the main lidar, and scan points in the filtered frame of point cloud.
  • the filtered point cloud is divided into multiple slices, that is, S1, S2...SN in the figure, N is a positive integer, where each The width of each slice can be 1m; starting from slice S1, for any slice Si, where i is a positive integer not greater than N, if the number of scanning points in Si is greater than the preset threshold, Si is marked as the target slice; for Target slice Si, continue to judge whether the next slice is the target slice, and so on, until the current slice is not the target slice or the number of target slices exceeds 5, for example, for the target slice Si, if Si+1 is still the target slice, Si+1 is not the target slice, then slices Si and Si+1 constitute a slice set.
  • one or more slice sets can be obtained according to the N slices in S1-SN; for any slice set, judge whether the X value range of all feature points in the slice set exceeds the length range, that is, judge whether
  • the feature points in the slice set that meet the conditions are used as the landmark point cloud to realize the extraction of high-precision landmark point cloud; as shown in Figure 7, slice S1 is the landmark point cloud on the left side of the road, and slice SN is Point cloud of landmarks on the right side of the road.
  • the second wire harness information of the marker point cloud may also be acquired, and the down-sampling process is performed on the marker point cloud according to the second wire harness information.
  • the method of taking points at intervals is down-sampled to extract an accurate feature point cloud; while improving the data processing efficiency, the texture structure of the marker is fully preserved , thus ensuring the accuracy of the landmark point cloud.
  • Step 605 Calibrate the external parameters of the lidar according to the marker point cloud and the ground point cloud.
  • the external parameters of the lidar are calibrated.
  • the extrinsic parameters of the lidar can be calibrated according to the above-mentioned down-sampled marker point cloud and ground point cloud, so as to improve processing efficiency.
  • the pitch angle and roll angle of the lidar can be calibrated according to the above-mentioned ground point cloud; thus, the high-precision ground point cloud extracted above can be used to make the calibrated pitch angle and roll angle more accurate .
  • the ground and the horizontal plane of the car body can be used as a constraint to establish an L1 loss function to solve the pitch angle and roll angle of the lidar.
  • L g represents the loss function corresponding to the solution pitch angle and roll angle
  • n indicates the number of ground points contained in the ground point cloud
  • n Indicates the mean of the z-values of the ground point cloud.
  • the yaw angle of the laser radar can be calibrated according to the point cloud of the above-mentioned markers; thus, the precision of the calibrated yaw angle is higher by using the above-mentioned extracted high-precision point cloud of the markers.
  • the parallelism between the marker and the forward direction of the car body can be used as a constraint, and the L1 loss function can be established to solve the yaw angle from the lidar coordinate system to the car body coordinate system.
  • L w represents the loss function corresponding to the solution yaw angle
  • m indicates the number of feature points contained in the marker point cloud
  • the formula (1) is used to calculate the yaw angle corresponding to the minimum L w , which is the yaw angle of the lidar.
  • the above-mentioned processing can be performed on the multi-frame point cloud accumulated for a period of time, and the obtained calibration results can be statistically analyzed, so as to optimize and obtain the final lidar The extrinsic parameters to the car body coordinate system.
  • the combination of multi-step fitting and threshold filtering can be used to extract high-precision ground point clouds, and then solve the pitch angle and roll angle of the lidar. Calculate the yaw angle of the laser radar; thus realize the high-precision online dynamic calibration of a single laser radar.
  • the landmarks can be the curbs and railings of the road, and there are no special requirements for the site, and no additional calibration boards, targets, reflective stickers, etc. are required, which reduces the calibration cost. Streets, highways, etc.), the online dynamic calibration of lidar can be completed by using the natural scene of the road. At the same time, by automatically extracting the marker point cloud in the collected point cloud; and based on the marker point cloud and the ground point cloud, the external parameters of the lidar are calibrated, so as to realize fully automatic online dynamic calibration without manual operation, and improve calibrated efficiency.
  • the collected point cloud is preliminarily screened according to the preset threshold; the preliminarily screened point cloud is subjected to multiple fitting processing, so that based on threshold filtering and multiple fitting processing, adaptive High-precision ground point cloud can be extracted accurately; it can also further automatically extract high-precision marker point cloud through point cloud slicing, thereby improving the external parameter accuracy of the calibrated lidar.
  • lidars with low total price, small viewing angle, and high wiring harness are more widely used.
  • Multiple lidars on the vehicle can achieve scene coverage and complementarity.
  • the external parameters of the slave laser radar can be optimized according to the external parameters of the master laser radar calibrated through the above steps 601 to 605 and the external parameters of the slave laser radar calibrated through the above steps 601 to 605 .
  • Fig. 8 shows a flowchart of a lidar calibration method according to an embodiment of the present application; as shown in Fig. 8, the method may include:
  • Step 801 according to the calibrated external parameters of multiple laser radars, the position information of the marker point cloud and the position information of the ground point cloud corresponding to each laser radar are obtained.
  • the plurality of lidars are lidars installed on the same vehicle.
  • the external parameters of any lidar can be calibrated through the above steps 601-605.
  • the marker point cloud corresponding to each laser radar may be the marker point cloud extracted in the above step 604
  • the ground point cloud corresponding to each laser radar may be the marker point cloud obtained in the above step 603 .
  • the position information of the marker point cloud represents the three-dimensional coordinates (x value, y value, z value) of the marker point cloud in the vehicle body coordinate system.
  • the plurality of lidars may include a master lidar and a slave lidar (eg, a side lidar).
  • the marker point cloud and the ground point cloud corresponding to the main lidar can be converted into the car body coordinate system to obtain the corresponding sign of the main lidar
  • the point cloud collected by the main laser radar can be converted into the vehicle body coordinate system to obtain the position information of the point cloud collected by the main laser radar, and then The position information of the ground point cloud corresponding to the main lidar can be obtained through the above extraction of the ground point cloud; the position information of the marker point cloud corresponding to the main lidar can be obtained through the above method of extracting the marker point cloud.
  • the point cloud collected by the side lidar can be converted into the vehicle body coordinate system to obtain the position information of the point cloud collected by the side lidar, and then the ground point can be extracted through the above
  • the position information of the ground point cloud corresponding to the side lidar can be obtained; the position information of the marker point cloud corresponding to the side lidar can be obtained through the above method of extracting the point cloud of the marker.
  • FIG. 9 shows a schematic diagram of a time synchronization according to an embodiment of the present application, as shown in FIG. 9 , where the arrows indicate The direction of the time axis, the point on the time axis represents the point cloud data packet, according to the timestamp of each point cloud data packet (that is, the corresponding position of the point cloud data packet on the time axis), the time axis of the main lidar The data packet is matched with the data packet on the time axis of the side lidar. If the time difference of the nearest adjacent data packets on the two time axes is less than the threshold, that is, it is located in the oval area shown in Figure 9, then the time Synchronization succeeded.
  • Step 802 according to the position information of the marker point cloud corresponding to each lidar and the position information of the ground point cloud, obtain the intersection feature point or intersection area.
  • intersection feature points represent the intersection points of two types of point clouds.
  • the intersection domain refers to the area parallel to the direction of vehicle travel and centered on the intersection feature point.
  • the first intersection feature point or the first intersection region may be obtained according to the position information of the ground point cloud corresponding to the main lidar and the position information of the ground point cloud corresponding to the side lidar.
  • the first intersection feature point represents the intersection point of the ground point cloud corresponding to the main lidar and the ground point cloud corresponding to the side lidar
  • the intersection point can be a ground point pair
  • the ground point pair includes the ground point cloud corresponding to the main lidar A ground point in the ground point, and a ground point in the ground point cloud corresponding to the side laser radar; That is, the intersection point of the ground point cloud corresponding to the main lidar and the ground point cloud corresponding to the side lidar.
  • the first intersection domain represents the area centered on the first intersection feature point, and the first intersection domain may include multiple ground point pairs, wherein each ground point pair includes a ground point in the ground point cloud corresponding to the main lidar , and a ground point in the ground point cloud corresponding to the side lidar; for example, FIG. 11 shows a schematic diagram of an intersection domain according to an embodiment of the present application. As shown in FIG. 11 , the area in the ellipse is the first intersection domain, and each ellipse includes multiple ground point pairs.
  • the position information of the ground point cloud corresponding to the main lidar and the position information of the ground point cloud corresponding to the side lidar are gridded with a 5m ⁇ 5m grid on the xy plane in the vehicle body coordinate system Processing; in any grid, based on any ground point in the ground point cloud corresponding to the side lidar, find the ground point in the ground point cloud corresponding to the main lidar with the smallest Manhattan distance between it, where the Manhattan distance represents The sum of the absolute axis distances of two ground points on the coordinate system, for example, the ground point a(x 1 ,y 1 ) in the ground point cloud corresponding to the side lidar, and the ground point b in the ground point cloud corresponding to the main lidar
  • the Manhattan distance of (x 2 ,y 2 ) is:
  • ground point pairs may be expanded to form the first intersection domain with the first intersection feature point as the center according to the wire harness information of the ground point, as shown in FIG. 11 above.
  • the second intersection feature point or the second intersection region may be obtained according to the position information of the marker point cloud corresponding to the main lidar and the position information of the marker point cloud corresponding to the side lidar.
  • the second intersection feature point represents the intersection point of the marker point cloud corresponding to the main lidar and the marker point cloud corresponding to the side lidar
  • the intersection point can be a feature point pair
  • the feature point pair includes the sign corresponding to the main lidar A feature point in the object point cloud, and a feature point in the marker point cloud corresponding to the side lidar.
  • the second intersection domain represents the area centered on the second intersection feature point, and the second intersection domain may include multiple feature point pairs, wherein each feature point pair includes a feature in the marker point cloud corresponding to the main lidar point, and a feature point in the landmark point cloud corresponding to the side lidar.
  • the position information of the marker point cloud corresponding to the main lidar and the position information of the marker point cloud corresponding to the side lidar, on the xz plane in the car body coordinate system refer to the above-mentioned extraction of the first intersection feature points or the first intersection domain to obtain the second intersection feature point or the second intersection domain, which will not be repeated here.
  • Step 803 optimize the calibrated extrinsic parameters of any lidar among the plurality of lidars according to the intersection feature points or intersection domains.
  • the yaw angle calibrated by the side lidar can be optimized according to the second intersection feature point or the second intersection domain.
  • an objective function can be constructed: min
  • y2 and y1 respectively represent the y values of the feature point pairs corresponding to the second intersection feature points.
  • y2 can represent a feature point pair, and the main lidar corresponds to The y value of the feature point in the car body coordinate system in the marker point cloud of the marker point cloud, y 1 can represent the y value of the feature point in the car body coordinate system of the feature point pair in the side lidar corresponding to the marker point cloud;
  • the objective function can be solved using the second intersection domain.
  • y2 represents all points in the point cloud corresponding to the main lidar in the second intersection domain.
  • the mean value of the y values of the feature points in the car body coordinate system, and y1 represents the y value of any feature point in the car body coordinate system in the second intersection domain, in the marker point cloud corresponding to the side lidar.
  • the calibrated pitch angle and roll angle of the side lidar may be optimized according to the first intersection feature point or the first intersection domain.
  • can be constructed to optimize the pitch angle and roll angle of the compensation side lidar, wherein, when the number of the first intersection feature points is not less than the preset threshold, it can be Use the first intersection feature point to solve the objective function.
  • z 2 and z 1 respectively represent the z value of the ground point pair corresponding to each first intersection feature point .
  • the z value of the ground point in the vehicle body coordinate system in the ground point cloud corresponding to the radar, z 1 can represent the z value of the ground point in the vehicle body coordinate system in the ground point cloud corresponding to the side laser radar;
  • the first intersection domain can be used to solve the objective function.
  • z 2 represents all ground points in the ground point cloud corresponding to the main lidar in the first intersection domain
  • z 1 represents the z value of any ground point in the vehicle body coordinate system in the ground point cloud corresponding to the side lidar in the first intersection domain.
  • the compensation amount of the rotation matrix R can be expressed as:
  • ⁇ represents the yaw angle
  • ⁇ represents the pitch angle
  • ⁇ represents the roll angle
  • the ground point pair corresponding to the first intersection feature point has the following relationship:
  • (x 1 , y 1 , z 1 ) represents the coordinate value of a ground point in the vehicle body coordinate system in the ground point cloud corresponding to the mid-side lidar
  • (x 2 , y 2 , z 2 ) represents the coordinate value of the ground point in the vehicle body coordinate system in the ground point cloud corresponding to the main lidar in the ground point pair.
  • z 2 represents the z value of the ground point in the vehicle body coordinate system in the ground point cloud corresponding to the main laser radar in the ground point pair
  • x 1 , y 1 , z 1 represent the ground point centering
  • the above-mentioned processing can be performed on the multi-frame point cloud accumulated for a period of time, and the obtained multiple sets of optimized calibration results can be statistically analyzed, In this way, the external parameters of the final optimized rear lidar to the car body coordinate system are obtained.
  • the calibration parameters of the lidar can be updated according to the optimized and compensated side laser yaw angle; thus, the intelligent driving function can be turned on or updated, based on the optimized and compensated high-precision external parameters of the lidar, the perception, positioning or The accuracy of functions such as fusion has been improved.
  • the point clouds collected by each laser radar are converted into the vehicle body coordinate system, and the position information of the marker point cloud and the position information of the ground point cloud corresponding to each laser radar are obtained, and then by extracting Intersection feature points and cross-domain features optimize the external parameters of the lidar; in some examples, both the main laser point cloud and the side laser point cloud can be converted to the car body coordinate system through the corresponding calibrated external parameters, and the main laser radar can be completed Time synchronization with the side lidar, and then on the basis of the marker point cloud, extract the intersection feature point and cross domain optimization to compensate the yaw angle from the side lidar to the vehicle body coordinate system; on the basis of the ground point, extract the intersection feature point And the cross-domain optimization compensates the pitch angle and roll angle of the side lidar to the car body coordinate system, so as to complete the joint optimization of the multi-lidar extrinsic parameters, so that the accuracy of the extrinsic parameters from the lidar to the car body coordinate system is higher
  • the calibration can be completed on daily road sections.
  • the road surface can also be uneven, convex, concave, uneven road surface or the surface of the markers can be used for calibration optimization; at the same time, the external parameters of the calibration are effectively improved. Accuracy and calibration efficiency.
  • the LiDAR calibration method shown in Figure 6 or Figure 8 above can be used in general scene calibration, service calibration, user self-calibration calibration, etc. in urban areas or elevated roads; in some examples, during the daily use of the vehicle, with Over time, due to uncertain factors such as object deformation, temperature, and small touches, the external parameters of the vehicle-mounted lidar will change. At this time, the vehicle does not need to return to the factory, and can keep driving in a straight line for a short period of time on open urban roads or expressways.
  • the calibration and optimization of the lidar external parameters can be completed, and the system can be updated
  • the internal and external parameters so that users can adjust and correct the external parameters of the lidar online in real time in daily life, ensure the safe use of intelligent driving functions, and improve intelligent driving performance.
  • the lidar calibration method provided in the embodiment of the present application will be described in detail below based on the application scenario described in FIG. 3 above.
  • Fig. 12 shows a flow chart of another lidar calibration method according to an embodiment of the present application; the method can be executed by the above lidar calibration device; as shown in Fig. 12, the method can include the following steps:
  • Step 1201. Obtain the point cloud collected by the lidar when the vehicle passes through the target area.
  • the laser radar may be any laser radar installed on the vehicle, for example, it may be a main laser radar or a side laser radar.
  • Markers are vertically arranged on at least one side of the target area.
  • multiple markers are vertically arranged on at least one side of the target area, and intersection points of the multiple markers and the ground are on the same straight line.
  • the vehicle may be the ego vehicle 301 mentioned above, and the target area may be the road shown in any one of Fig. 4(a)-Fig. 4(e) above.
  • the acquired point cloud may be a point cloud collected by a lidar when the vehicle is driving along a straight line at a constant speed.
  • it may be a point cloud collected by the lidar 303 when the ego vehicle 301 is traveling straight and at a constant speed on any of the roads shown in FIG. 4( a )- FIG. 4( e ).
  • the driving state of the vehicle can be determined according to the pose information of the vehicle passing the target area; for example, the point cloud collected by the on-board lidar can be extracted when the vehicle is determined to pass the target area at a constant speed through the inertial navigation system.
  • the target area can be a road as shown in Figure 4(a), a row of parallel vertical markers can be set on the left side (or right side) of the road, the markers are equidistant, and the markers can be reflective stickers cylindrical straight bar; this target area is easy to construct and saves cost.
  • the target area can be a road as shown in Figure 4(b).
  • Equidistant cylindrical straight bars are symmetrically distributed on both sides of the target area.
  • the two rows of cylindrical straight bars have a parallel relationship, and the cylindrical straight bars on the left and right sides are symmetrically distributed. ;
  • the target area can be a road as shown in Figure 4(c).
  • Equidistant cylindrical straight bars are distributed alternately on both sides of the target area.
  • the two rows of cylindrical straight bars have a parallel relationship, and the cylindrical straight bars on the left and right sides are staggered.
  • the staggered offset distance can be half of the distance between adjacent cylindrical straight rods on the same side.
  • the number of markers scanned by the forward main lidar in the target area will decrease in the time domain, and the distance to scan the nearest marker will be reduced by half. Before and after the scanning field of view disappears, there will be no problem that the frame distance jumps too much before and after scanning the markers, thereby further improving the accuracy, stability and calibration efficiency of external reference calibration.
  • the target area can be a road as shown in Figure 4(d). Cylindrical straight bars are distributed at any distance on both sides of the target area, and the two rows of cylindrical straight bars have a parallel relationship. In the target area, at least one row of cylindrical straight rods may be at unequal intervals, so that the construction of the target area is more convenient, without the need for high-precision measurement and precise construction, and the flexibility and versatility of use are improved. Wherein, the distance between adjacent markers can be estimated by means of marker point cloud fitting processing and other methods as required.
  • Step 1202 extracting the marker point cloud in the collected point cloud.
  • the marker point cloud can be extracted according to the material characteristics of the marker. Taking the landmark as a cylindrical vertical rod as an example, the cylindrical vertical rod point cloud can be screened out from the point cloud collected by the lidar according to the reflection intensity.
  • the marker point cloud in the point cloud collected in step 1201 may be extracted with reference to the manner of extracting the marker point cloud shown in FIG. 6 above.
  • accurate marker point cloud may be extracted.
  • Step 1203 according to the point cloud of the marker, obtain the fitting line information of the marker.
  • the fitting line information includes the position and direction information of the fitting line.
  • the fitting line of the marker may include a straight line perpendicular to the ground such as the center line of the marker, the generatrix of the marker, and the sideline of the marker, wherein the center line of the marker represents the distance between the upper and lower section centers of the marker.
  • the embodiment of the present application takes the fitting line of the marker as the center line of the marker as an example, and exemplifies the way to obtain the information of the fitting line of the marker;
  • the center line of the marker can be object vector Represents, wherein, a, b, c represent the three components of the marker vector l in the direction vector, that is, the direction information of the centerline; Indicates the position vector of the intersection point of the marker vector l and the XY plane, that is, the position information of the center line.
  • this step may include: determining the initial value of the rotation angle according to the point cloud of the marker, where the initial value of the rotation angle rotates the point cloud of the marker, and the projection of the horizontal plane in the lidar coordinate system The area is the smallest; according to the initial value of the rotation angle, the point cloud of the marker is rotated; the fitting line information of the marker is obtained by using the rotated point cloud of the marker.
  • the direction vector [a, b, c] and the intersection position of l and XY plane Thus the marker vector l is obtained.
  • FIG. 13 shows a schematic diagram of a single lidar calibration according to an embodiment of the present application.
  • the marker vector l that is, the centerline of the marker
  • the vertical landmark vector l' that is, the direction vector is the unit vector [0,0,1]
  • [0, 0, 1] T represents the transposition matrix of the unit vector [0, 0, 1]
  • [a, b, c] T represents the transposition matrix of the vector [a, b, c] .
  • p i represents the coordinates of the laser point i on the marker before rotation
  • R the coordinates of the laser point on the rotated marker
  • the marker vector l before rotation, after rotation R, in the laser radar coordinate system, the intersection point of the marker vector l′ and the XY plane obtained is
  • x l , y l are coordinate values of the intersection point.
  • r is the radius of the marker section; is the distance from the laser point p i to the landmark vector l; where, is a point on the marker vector l; express The angle with R[a, b, c] T vector, its value is: Right now The angle between the line segment and the vertical direction.
  • the quantities to be optimized are a, b, c
  • the optimized initial value is R init , and based on the known radius of the marker section, the marker vector l can be obtained by solving the above formula (8) through optimization.
  • Step 1204 according to the fitting line information, obtain the value of the external parameter of the lidar.
  • the value of at least one of the pitch angle, roll angle, and yaw angle of the lidar can be obtained according to the fitting line information.
  • the value of the pitch angle of the lidar can be obtained according to the direction information of the center line (such as a single marker vector) obtained above. Since each direction component of a single marker vector represents the projected length of the marker vector on each direction axis, the pitch angle can be estimated using the components of a single marker vector. For example, based on The a, c components in the direction vector [a, b, c] of the direction vector [a, b, c] calculate the value of the pitch angle through atan(a/c).
  • the value of the external parameters of the lidar can be obtained according to the point cloud of a landmark collected by the lidar.
  • the pitch angle of the lidar plays a vital role in whether there are enough laser points on the ground near the vehicle; at the same time, ground information can increase the accuracy and stability of calibration optimization to a certain extent. Therefore, when the lidar is facing upward , it can also be judged whether the obtained value of the pitch angle of the lidar is greater than the second preset threshold, that is, the value of the pitch angle is used as a decision-making reference value for whether to use ground information for calibration optimization; the laser radar installation deviation (for example, pitch The problem of unavailable ground information caused by angle) can also make full use of effective ground information.
  • the object point cloud is used to calibrate the external parameters of the lidar.
  • the first preset threshold may be 90°, and if the angle between the lidar orientation and the vertical upward direction is less than 90°, it indicates that the lidar orientation is upward.
  • the pitch angle, yaw angle, The roll angle is jointly calibrated. Since the centerline of the marker is vertically upward in the world coordinate system, the centerline information can be used to calibrate the pitch angle, yaw angle, and roll angle of the lidar at the same time.
  • marker vectors can be used The position vector of Based on the objective function shown in formula (9), jointly calibrate the pitch angle, yaw angle, and roll angle of the lidar.
  • ⁇ and ⁇ are the proportional coefficients of the residual item, r is the radius of the marker section, z is the height of the laser radar, j is the number of the scanned marker, and P′ ij is the rotated
  • the i-th laser point of the j-th marker, l j (0, 0, 1, d j ) represents the j-th marker vector, and d j is the intersection point of the j-th marker and the XY plane.
  • P′ ij R roll R pitch R yaw P ij +[0,0,x] T , R yaw represents the yaw angle, R pitch represents the pitch angle, R roll represents the roll angle; P ij represents the jth roll angle before rotation The i-th laser point of a marker.
  • the ground point cloud does not depend on the ground point cloud.
  • the high-precision calibration of the lidar extrinsic parameters can be realized by using the marker point cloud.
  • the installation pitch angle is too large, resulting in the inability to scan Ground point cloud or lidar with less scanned ground point cloud; lidar with higher installation height and farther distance of scanned ground point cloud, etc., can achieve high-precision external parameter calibration.
  • FIG. 14 shows a schematic diagram of a LiDAR scanning scene according to an embodiment of the present application.
  • the main LiDAR vertically The field angle is small, and the main lidar cannot scan the ground close to the vehicle, that is, it cannot obtain the nearby ground point cloud.
  • the calibration accuracy of the main lidar extrinsic parameter calibration based on ground information is adopted in related technologies, the calibration accuracy is usually low. Therefore, it is possible to set markers on both sides of the road with reference to scenes such as those shown in Figure 4(a)- Figure 4(e). In the case of a small field of view, the high-precision calibration of the external parameters of the main lidar is completed.
  • Figure 15(a)- Figure 15(b) shows a schematic diagram of a production line environment according to an embodiment of the present application, as shown in Figure 15(a)-
  • the side lidar of the vehicle cannot scan the ground, that is, the ground point cloud cannot be obtained; or the scanned ground area is small, that is, the number of ground point clouds obtained is small .
  • the solution in related technologies that relies on ground information to calibrate the external parameters of the side lidar is adopted, it usually cannot operate normally. Therefore, it is possible to set markers on the production line site with reference to the scenarios shown in Figure 4(a)- Figure 4(e). In the narrow production line site, the high-precision calibration of the external parameters of the side lidar is completed.
  • Fig. 16 shows a comparative schematic diagram of pitch angle calibration according to an embodiment of the present application. As shown in Fig.
  • the ground point cloud in the collected point cloud can also be extracted, and the ground point cloud and marker point cloud can be used to calibrate the lidar extrinsic parameters, so as to make full use of the ground information and marker information and improve the external parameters.
  • Parameter calibration accuracy for example, when the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, and the value of the pitch angle is greater than the second preset threshold, according to the marker point cloud and the ground point Cloud, to calibrate the external parameters of the lidar.
  • the ground point clouds can be fully utilized, thereby further improving the calibration accuracy and stability.
  • marker vectors can be used The position vector of And the ground point cloud, based on the objective function shown in formula (10), jointly calibrate the pitch angle, yaw angle, and roll angle of the lidar.
  • a frame of point cloud contains laser points corresponding to multiple landmarks
  • the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, and the value of the pitch angle of the lidar is not greater than the second preset threshold, according to the marker point cloud and the ground point cloud , to calibrate the external parameters of the lidar.
  • the lidar When the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, and the value of the pitch angle of the lidar is not greater than the second preset threshold, the lidar is heading upwards and can scan the ground at the same time , that is, there is a ground point cloud. Therefore, the external parameters of the lidar can be calibrated by using the ground point cloud and the marker point cloud, so as to make full use of the ground information and improve the calibration accuracy of the external parameters of the lidar.
  • the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, and the value of the pitch angle of the lidar is not greater than the second preset threshold, according to the ground point Cloud, to calibrate the pitch angle and roll angle of the lidar, and can also calibrate the height of the lidar; according to the position information of the fitting line, calibrate the yaw angle of the lidar.
  • marker vectors can be used The position vector of Calibrate the yaw angle of the lidar.
  • the lidar can scan the ground, and there is a ground point cloud, so , can use the ground point cloud to calibrate the pitch angle and roll angle of the lidar, which improves the calibration accuracy and stability.
  • the driving deflection angle of the vehicle is not limited, and the calibration result can be optimized by combining the vehicle motion information after the above calibration is completed.
  • the location information of the vehicle and the location information of the lidar can be obtained;
  • the heading angle of the vehicle can be determined according to the location information of the vehicle and the location information of the lidar;
  • the calibrated lidar can be optimized according to the heading angle yaw angle.
  • the lidar can locate the position information of the lidar; the vehicle can estimate the position information of the vehicle through its own chassis and other information; the vehicle's heading can be estimated by filtering using the position information of the vehicle and the position information of the lidar Angle; use the parallel constraint of the vehicle heading angle to compensate the dynamic change of the yaw angle.
  • the above-mentioned processing can be performed on the multi-frame point cloud accumulated for a period of time, and the obtained calibration results can be statistically analyzed, so as to optimize and obtain the final lidar external reference.
  • the fitting line using the vertical markers needs to comply with the principle of vertical constraints, resulting in The value of the external parameter of the lidar; in some examples, based on the value of the calculated pitch angle, when the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, whether the value of the pitch angle is greater than the second preset threshold Set the threshold to judge, so as to automatically judge the usability of the ground point cloud, improve the calibration efficiency and automation; and only use the marker point cloud to calibrate the pitch angle, yaw angle and roll angle of a single lidar at the same time.
  • the dynamic change of the yaw angle can be compensated by combining the vehicle motion information, thus realizing the high-precision dynamic calibration of a single lidar.
  • At least one side of the target area is vertically provided with markers, which are simple to set up, reduce requirements on the site, and have low construction costs.
  • the fitting line information of the marker is obtained.
  • the fitting line based on the vertical marker needs to meet the vertical constraint, and the value of the external parameter of the lidar is obtained.
  • the laser can be calculated The value of the radar external parameter; in some examples, when the angle between the lidar orientation and the vertical upward direction is less than the first preset threshold, and the value of the lidar pitch angle is greater than the second preset threshold, according to the sign
  • the object point cloud can complete the high-precision calibration of the lidar external parameters; at the same time, because the calibration process does not depend on the ground point cloud, it can be applied to the scene with insufficient ground information (for example, the laser with a small vertical field of view
  • the radar cannot collect nearby ground point clouds, the lidar on the limited side of the site cannot collect effective ground point clouds, and the installation of the lidar with an excessively large pitch angle leads to missing or less ground point clouds, etc.).
  • High Accuracy Calibration of Single LiDAR in Scenes with Insufficient Ground Information compared with methods such as mapping and calibration, the entire calibration process can be automatically performed, which improves the efficiency of single lidar calibration.
  • the calibration results of any laser radar can be further optimized according to the calibration results of any two laser radars to the vehicle body coordinate system to achieve multi-laser calibration.
  • the external parameters of the slave lidar can be optimized according to the external parameters of the master lidar calibrated through the above steps 1201-1204 and the external parameters of the slave lidar calibrated through the above steps 1201-1204.
  • the master lidar and the slave lidar may have no common view area, or the common view area may be small.
  • the main lidar can scan at least two landmarks; by predicting and matching the position of the landmarks, the external parameters of the slave lidar can be optimized.
  • Fig. 17 shows a flow chart of a lidar calibration method according to an embodiment of the present application; as shown in Fig. 17, the method may include the following steps:
  • Step 1701. Determine the position information of multiple markers according to the calibrated external parameters of the main lidar and the point clouds of multiple markers collected by the main lidar.
  • the external parameters of the main lidar may be the external parameters calibrated through the above steps 1201-1204.
  • the position information of the plurality of markers may include the position information of the fitting lines of the plurality of markers converted based on the above-mentioned calibration external parameters.
  • the point clouds of multiple markers collected by the main lidar can be used to fit the position vectors of multiple markers Then, using the calibrated external parameters of the main lidar, the position vector Converted to the vertical direction, the position vector obtained at this time is the position information of multiple markers.
  • the main laser radar continuously tracks the markers, and judges and records the unique number j of the markers by scanning the sequence.
  • the distance Wj between any two markers can be measured; let Wj represent the jth marker The distance from the object to the j+1th marker.
  • Figures 18(a)-18(b) show a schematic diagram of a multi-lidar joint optimization according to an embodiment of the present application; as shown in Figure 18(a), the distance between the markers can be Equal, the main lidar numbers the landmarks tracked continuously, that is, W0, W1...Wj; as shown in Figure 18(b), the lidar can currently track the landmarks W0, W1, and at the same time can be based on the determined landmarks The location information of W0 and W1 is used to obtain the distance between W0 and W1.
  • Step 1702. Obtain the predicted position of the first marker according to the position information of the multiple markers.
  • the first marker is a marker that can be tracked by the laser radar and is behind the above-mentioned multiple markers.
  • the predicted position of the first marker can be deduced by using the number of the first marker and the distance Wj between the markers.
  • the predicted position of the first marker may include position information of a fitted line of the first marker (eg, position information of a central line of the first marker).
  • the direction prediction can be performed first, and then the distance prediction can be performed to obtain the predicted position of the first marker; for example, since multiple vertical markers on the same side are parallel, based on two markers Based on the location information of the object, a straight line can be determined, and it can be seen that the rest of the markers are on this straight line, and then combined with the sequence of the two markers, the arrangement direction of the markers on the straight line can be obtained; then based on the straight line
  • the predicted position of the first marker is estimated through the distance Wj estimated by the main lidar tracking markers, such as W2 and W3.
  • W2 can be tracked from the lidar, that is, the first marker can be W2, as shown in Figure 18(b), the position information of W0 and W1 can be used to determine W0 and W1 The arrangement direction of the markers on the straight line, and then based on the direction, the predicted position of W2 is calculated by using the distance between W0 and W1 and the position information of W1.
  • Step 1703 Obtain the measurement position of the first marker according to the calibrated extrinsic parameters from the lidar and the point cloud of the first marker collected from the lidar.
  • the external parameters from the laser radar may be the external parameters calibrated through the above steps 1201-1204.
  • the position information of the first marker may include the position information of the centerline of the first marker converted based on the above-mentioned calibration extrinsic parameters.
  • the position vector of the first marker can be obtained by using the point cloud of the first marker collected from the lidar in the manner described above Then, using the calibrated external parameters from the lidar, the position vector Converting to the vertical direction, the position vector obtained at this time is the measurement position of the first marker.
  • Step 1704 optimize the extrinsic parameters from the lidar by predicting the position and measuring the position.
  • the external parameters of the side lidar are optimized so that the coincidence degree of the position vector of the first marker is the best.
  • the measurement position may be transformed by using the external parameters of the slave laser radar relative to the master laser radar, and then the transformed measurement position and the predicted position may be used for matching processing to complete the joint optimization of the external parameters. It can be understood that if there is a deviation in the external parameters of the slave lidar relative to the main lidar, the measured position obtained from the slave lidar is inconsistent with the predicted position after being converted by the biased relative external parameters.
  • the least squares form can be used to optimize the solution based on the optimization objective function: 0.5[( ⁇ a) 2 +( ⁇ b) 2 +( ⁇ c) 2 ], where ⁇ a, ⁇ b, and ⁇ c are the main lidar-slave The difference of the three components of the lidar's direction vector.
  • the lidar calibration method shown in Figure 12 or Figure 17 above can be used in scenarios such as end-of-production line calibration, service calibration, and online self-calibration with high-definition maps; Multi-sensor adaptation; fast calibration; low cost, simple site, and can be promoted and constructed in different factories; a large angle calibration tolerance range is required, and an abnormal alarm is required when there is a large deviation in the installation; all-weather work, the navigation system cannot It also needs to be able to work normally when used; at this time, the vehicle drives in from one end of any road in Figure 4 (a)- Figure 4 (e) provided by the embodiment of the present application and drives out from the other end, and the vehicle passes the Driving, and then based on the single lidar external parameter calibration and/or multi-lidar joint optimization method of the above-mentioned embodiment of the application, the vehicle-mounted lidar calibration can be completed to meet the needs of the calibration at the end of the production line; and the external parameters in the system can be updated. Ensure the safe use of intelligent driving functions and improve intelligent driving performance.
  • embodiments of the present application also provide a laser radar calibration device, which can be used to implement the technical solutions described in the above method embodiments.
  • Fig. 19 shows a structural diagram of a laser radar calibration device according to an embodiment of the present application.
  • the device includes: an acquisition module 1901, which is used to obtain the points collected by the laser radar when the vehicle passes the target road cloud, at least one side of the target road is provided with markers; the screening module 1902 is used to perform preliminary screening on the collected point cloud according to a preset threshold; the preset threshold is determined by the installation height of the lidar; The first extraction module 1903 is used to perform multiple fitting processing on the initially screened point cloud to obtain the ground point cloud; the second extraction module 1904 is used to extract the marker point cloud in the collected point cloud;
  • the calibration module 1905 is configured to calibrate the external parameters of the lidar according to the marker point cloud and the ground point cloud.
  • the device further includes: a conversion module, configured to obtain the position information of the marker point cloud corresponding to each laser radar and the position information of the ground point cloud according to the calibrated external parameters of multiple laser radars. Position information; the third extraction module is used to obtain intersection feature points or intersection domains according to the position information of the marker point clouds corresponding to the lidars and the position information of the ground point clouds, wherein the intersection domains represent parallel to The direction in which the vehicle is traveling is the area centered on the intersection feature point; an optimization module is configured to, according to the intersection feature point or the intersection domain, calibrate the extrinsic parameters of any one of the plurality of lidars optimize.
  • the multiple laser radars include a master laser radar and a slave laser radar, wherein the master laser radar is used to scan the environment in front of the vehicle, and the slave laser radar is used to scan The side and/or rear environment of the vehicle; the conversion module is also used to: convert the marker point cloud and the ground point cloud corresponding to the main laser radar to the vehicle body according to the calibrated external parameters of the main laser radar In the coordinate system, the position information of the marker point cloud corresponding to the main laser radar and the position information of the ground point cloud are obtained; according to the calibrated external parameters of the slave laser radar, the marker point cloud corresponding to the slave laser radar is And the ground point cloud is converted into the car body coordinate system to obtain the position information of the marker point cloud corresponding to the laser radar and the position information of the ground point cloud; the third extraction module is also used for: according to the main The position information of the ground point cloud corresponding to the laser radar and the position information of the ground point cloud corresponding to the slave laser radar obtain the first intersection feature point or the first intersection domain;
  • the external parameters include at least one of a pitch angle, a roll angle, and a yaw angle; the calibration module is further configured to: calibrate the lidar according to the ground point cloud The pitch angle and roll angle; according to the marker point cloud, the yaw angle of the lidar is calibrated.
  • the second extraction module is further configured to: filter out the ground point cloud from the collected point cloud; divide the filtered point cloud along a direction perpendicular to the traveling direction of the vehicle is a plurality of slices; extracting the marker point cloud, the marker point cloud includes feature points in a slice set that satisfies a preset condition, wherein the slice set includes one or more adjacent target slices, the The number of feature points in the target slice exceeds the threshold.
  • the device further includes a down-sampling module, configured to: acquire first wire harness information of the ground point cloud, and perform down-sampling processing on the ground point cloud according to the first wire harness information; And/or, acquire the second wire harness information of the marker point cloud, and perform down-sampling processing on the marker point cloud according to the second wire harness information; the calibration module is further configured to: according to the down-sampled The marker point cloud and the ground point cloud calibrate the external parameters of the lidar.
  • the acquired point cloud is a point cloud collected by a laser radar when the vehicle is driving in a straight line.
  • the marker includes at least one of a roadside, a guardrail, and a building.
  • Fig. 20 shows a structural diagram of a lidar calibration device according to an embodiment of the present application.
  • the device includes: an acquisition module 2001, which is used to acquire the points collected by the lidar when the vehicle passes through the target area cloud; at least one side of the target area is vertically provided with markers; the extraction module 2002 is used to extract the marker point cloud in the collected point cloud; the fitting module 2003 is used to obtain the marker according to the marker point cloud The fitting line information of the object; the fitting line information includes the position and direction information of the fitting line; the calculation module 2004 is used to obtain the value of the lidar external parameter according to the fitting line information;
  • the external parameter includes a pitch angle
  • the device further includes: a calibration module, configured to make an angle between the lidar orientation and the vertical upward direction smaller than a first preset threshold, and When the value of the pitch angle is greater than a second preset threshold, the external parameters of the lidar are calibrated according to the marker point cloud, wherein the second preset threshold is determined by the vertical Angle of view is fixed.
  • the external parameter includes a yaw angle
  • the device further includes: an optimization module, configured to acquire the position information of the vehicle and the position information of the laser radar; The position information and the position information of the lidar determine the heading angle of the vehicle; according to the heading angle, optimize the calibrated yaw angle of the lidar.
  • the vehicle is equipped with a master laser radar and a slave laser radar, wherein the master laser radar is used to scan the environment in front of the vehicle, and the slave laser radar is used to scan the vehicle The side and/or rear environment;
  • the device also includes: a determination module, configured to determine the plurality of marker point clouds collected by the main laser radar according to the calibrated external parameters of the main laser radar The position information of the markers;
  • the prediction module is used to obtain the predicted position of the first marker according to the position information of the plurality of markers;
  • the measurement module is used to obtain the predicted position of the first marker based on the calibrated external parameters of the laser radar and the slave
  • the point cloud of the first marker collected by the laser radar is used to obtain the measurement position of the first marker;
  • the matching module is used to optimize the external parameter from the laser radar by comparing the predicted position and the measured position .
  • the extraction module is further configured to: extract the ground point cloud in the collected point cloud;
  • the calibration module is further configured to: according to the marker point cloud and the ground point cloud, Calibrate the external parameters of the lidar.
  • the fitting module is further configured to: determine the initial value of the rotation angle according to the point cloud of the marker, and after the initial value of the rotation angle rotates the point cloud of the marker, the The projection area of the horizontal plane in the coordinate system is the smallest; according to the initial value of the rotation angle, the point cloud of the marker is rotated; and the fitting line information of the marker is obtained by using the point cloud of the rotated marker.
  • the calibration module is further configured to: when the angle between the lidar orientation and the vertical upward direction is smaller than a first preset threshold, and the value of the pitch angle is not greater than the In the case of the second preset threshold, the external parameters of the lidar are calibrated according to the marker point cloud and the ground point cloud.
  • the external parameters include at least one of a pitch angle, a roll angle, and a yaw angle; the calibration module is also used to: When the included angle is less than the first preset threshold and the value of the pitch angle is not greater than the second preset threshold, the pitch angle and roll angle of the lidar are calibrated according to the ground point cloud; The position information of the line is used to calibrate the yaw angle of the lidar.
  • At least one side of the target area is vertically provided with a plurality of markers, and intersection points of the plurality of markers and the ground are on the same straight line.
  • An embodiment of the present application provides a laser radar calibration device, including: a processor and a memory for storing processor-executable instructions; wherein, the processor is configured to implement the above laser radar calibration method when executing the instructions .
  • Fig. 21 shows a schematic structural diagram of a laser radar calibration device according to an embodiment of the present application.
  • the laser radar calibration device may include: at least one processor 2101, a communication line 2102, a memory 2103 and at least A communication interface 2104.
  • the processor 2101 can be a general-purpose central processing unit (central processing unit, CPU), a microprocessor, a specific application integrated circuit (application-specific integrated circuit, ASIC), or one or more devices used to control the program execution of the application program integrated circuit.
  • CPU central processing unit
  • ASIC application-specific integrated circuit
  • Communications link 2102 may include a pathway for passing information between the components described above.
  • the communication interface 2104 uses any device such as a transceiver for communicating with other devices or communication networks, such as Ethernet, RAN, wireless local area networks (wireless local area networks, WLAN) and so on.
  • a transceiver for communicating with other devices or communication networks, such as Ethernet, RAN, wireless local area networks (wireless local area networks, WLAN) and so on.
  • the memory 2103 may be a read-only memory (read-only memory, ROM) or other types of static storage devices that can store static information and instructions, random access memory (random access memory, RAM) or other types that can store information and instructions It can also be an electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or can be used to carry or store desired program code in the form of instructions or data structures and can be programmed by a computer Any other medium accessed, but not limited to.
  • the memory may exist independently and be connected to the processor through the communication line 2102.
  • Memory can also be integrated with the processor.
  • the memory provided by the embodiment of the present application may generally be non-volatile.
  • the memory 2103 is used to store computer-executed instructions for implementing the solution of the present application, and the execution is controlled by the processor 2101 .
  • the processor 2101 is configured to execute computer-executed instructions stored in the memory 2103, so as to implement the methods provided in the above-mentioned embodiments of the present application.
  • the computer-executed instructions in the embodiments of the present application may also be referred to as application program codes, which is not specifically limited in the embodiments of the present application.
  • the processor 2101 may include one or more CPUs, such as CPU0 and CPU1 in FIG. 21 .
  • the lidar calibration device may include multiple processors, such as processor 2101 and processor 2107 in FIG. 21 .
  • processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor.
  • a processor herein may refer to one or more devices, circuits, and/or processing cores for processing data (eg, computer program instructions).
  • the lidar calibration apparatus may further include an output device 2105 and an input device 2106 .
  • the output device 2105 is in communication with the processor 2101 and can display information in a variety of ways.
  • the output device 2105 may be a liquid crystal display (liquid crystal display, LCD), a light emitting diode (light emitting diode, LED) display device, a cathode ray tube (cathode ray tube, CRT) display device, or a projector (projector) wait.
  • the input device 2106 communicates with the processor 2101 and can receive user input in various ways.
  • the input device 2106 may be a mouse, a keyboard, a touch screen device, or a sensory device, among others.
  • An embodiment of the present application further provides a laser radar calibration system, and the laser radar calibration system includes at least one laser radar calibration device mentioned in the above-mentioned embodiments of the present application.
  • An embodiment of the present application further provides a vehicle, the vehicle includes at least one laser radar calibration device or laser radar calibration system mentioned in the above embodiments of the present application.
  • An embodiment of the present application provides a non-volatile computer-readable storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the foregoing method is realized.
  • An embodiment of the present application provides a computer program product, including computer-readable codes, or a non-volatile computer-readable storage medium bearing computer-readable codes, when the computer-readable codes are stored in a processor of an electronic device When running in the electronic device, the processor in the electronic device executes the above method.
  • a computer readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device.
  • a computer readable storage medium may be, for example, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing.
  • Non-exhaustive list of computer-readable storage media include: portable computer disk, hard disk, random access memory (Random Access Memory, RAM), read only memory (Read Only Memory, ROM), erasable Electrically Programmable Read-Only-Memory (EPROM or flash memory), Static Random-Access Memory (Static Random-Access Memory, SRAM), Portable Compression Disk Read-Only Memory (Compact Disc Read-Only Memory, CD -ROM), Digital Video Disc (DVD), memory sticks, floppy disks, mechanically encoded devices such as punched cards or raised structures in grooves with instructions stored thereon, and any suitable combination of the foregoing .
  • RAM Random Access Memory
  • ROM read only memory
  • EPROM or flash memory erasable Electrically Programmable Read-Only-Memory
  • Static Random-Access Memory SRAM
  • Portable Compression Disk Read-Only Memory Compact Disc Read-Only Memory
  • CD -ROM Compact Disc Read-Only Memory
  • DVD Digital Video Disc
  • Computer readable program instructions or codes described herein may be downloaded from a computer readable storage medium to a respective computing/processing device, or downloaded to an external computer or external storage device over a network, such as the Internet, local area network, wide area network, and/or wireless network.
  • the network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers.
  • a network adapter card or a network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing/processing device .
  • Computer program instructions for performing the operations of the present application may be assembly instructions, instruction set architecture (Instruction Set Architecture, ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or in one or more source or object code written in any combination of programming languages, including object-oriented programming languages—such as Smalltalk, C++, etc., and conventional procedural programming languages—such as the “C” language or similar programming languages.
  • Computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server implement.
  • the remote computer can be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or it can be connected to an external computer such as use an Internet service provider to connect via the Internet).
  • electronic circuits such as programmable logic circuits, field-programmable gate arrays (Field-Programmable Gate Array, FPGA) or programmable logic arrays (Programmable Logic Array, PLA), the electronic circuit can execute computer-readable program instructions, thereby realizing various aspects of the present application.
  • These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine such that when executed by the processor of the computer or other programmable data processing apparatus , producing an apparatus for realizing the functions/actions specified in one or more blocks in the flowchart and/or block diagram.
  • These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause computers, programmable data processing devices and/or other devices to work in a specific way, so that the computer-readable medium storing instructions includes An article of manufacture comprising instructions for implementing various aspects of the functions/acts specified in one or more blocks in flowcharts and/or block diagrams.
  • each block in a flowchart or block diagram may represent a module, a portion of a program segment, or an instruction that includes one or more Executable instructions.
  • the functions noted in the block may occur out of the order noted in the figures. For example, two blocks in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block in the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented with hardware (such as circuits or ASIC (Application Specific Integrated Circuit, application-specific integrated circuit)), or can be implemented with a combination of hardware and software, such as firmware.
  • hardware such as circuits or ASIC (Application Specific Integrated Circuit, application-specific integrated circuit)
  • firmware such as firmware

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Optical Radar Systems And Details Thereof (AREA)
  • Traffic Control Systems (AREA)

Abstract

一种激光雷达的标定方法、装置及存储介质。该方法包括：获取车辆通过目标道路时激光雷达采集的点云，目标道路的至少一边设置有标志物（601）；根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定（602）；对所述初步筛选出的点云进行多次拟合处理，得到地面点云（603）；提取所采集的点云中的标志物点云（604）；根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定（605）。该方法降低了对标定场地的要求，提高了标定效率及标定精度。

Description

一种激光雷达的标定方法、装置及存储介质 技术领域
本申请涉及智能驾驶技术领域，尤其涉及一种激光雷达的标定方法、装置及存储介质。
背景技术
在智能驾驶领域，激光雷达是实现高级别自动驾驶功能必不可少的组成部分。激光雷达的外参标定精度对于实现感知、定位或融合等功能，以及保障车辆安全性起着重要作用。
现有标定激光雷达的方式，标定精度较低，且对场地具有较高要求，标定成本高；标定过程中，依赖于人工操作或建图等，标定效率低。
发明内容
有鉴于此，提出了一种激光雷达的标定方法、装置及存储介质。
第一方面，本申请的实施例提供了一种激光雷达的标定方法，所述方法包括：获取车辆通过目标道路时激光雷达采集的点云，所述目标道路的至少一边设置有标志物；根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定；对所述初步筛选出的点云进行多次拟合处理，得到地面点云；提取所采集的点云中的标志物点云；根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
基于上述技术方案，标志物可以为道路的路沿、路栏等，对场地的要求无特殊要求，无需额外设置标定板，靶标，反光贴等，降低了标定成本，在开放道路(例如城市街道、高速公路等)，利用道路的自然场景即可完成激光雷达的在线动态标定。同时，通过自动提取所采集的点云中的标志物点云；并根据标志物点云及地面点云，对激光雷达的外参进行标定，从而实现全自动在线动态标定，无需人工操作，提高了标定效率。此外，在提取地面点云时，根据预设阈值对所采集的点云进行初步筛选；对初步筛选出的点云进行多次拟合处理，从而基于阈值过滤与多次拟合处理，自适应地提取高精度的地面点云；还可以进一步通过点云切片，自动提取高精度的标志物点云，从而提高所标定的激光雷达的外参精度。
根据第一方面，在所述第一方面的第一种可能的实现方式中，所述方法还包括：根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息；根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，其中，所述交叉域表示平行于所述车辆行进的方向，以所述交叉特征点为中心的区域；根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化。
基于上述技术方案，获取各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，进而通过提取交叉特征点及交叉域特征优化任一激光雷达的外参，进一步提升激光雷达的外参精度。
根据第一方面的第一种可能的实现方式，在所述第一方面的第二种可能的实现方式中，所述多个激光雷达包括主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方的环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，包括：根据主激光雷达的标定后的外参，将主激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述主激光雷达对应的标志物点云的位置信息及地面点云的位置信息；根据从激光雷达的标定后的外参，将所述从激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述从激光雷达对应的标志物点云的位置信息及地面点云的位置信息；所述根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，包括：根据所述主激光雷达对应的地面点云的位置信息及所述从激光雷达对应的地面点云的位置信息，得到第一交叉特征点或第一交叉域；根据所述主激光雷达对应的标志物点云的位置信息及所述从激光雷达对应的标志物点云的位置信息，得到第二交叉特征点或第二交叉域；所述根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化，包括：根据所述第一交叉特征点或第一交叉域，对所述从激光雷达标定后的俯仰角和翻滚角进行优化；根据所述第二交叉特征点或第二交叉域，对所述从激光雷达标定后的偏航角进行优化。
基于上述技术方案，将主激光点云和侧激光点云均通过对应标定后的外参转换到车体坐标系，进而在标志物点云的基础上，提取交叉特征点和交叉域优化补偿侧激光雷达到车体坐标系的偏航角；在地面点的基础上，提取交叉特征点和交叉域优化补偿侧激光雷达到车体坐标系的俯仰角和翻滚角，从而完成多激光雷达外参的联合优化，使得激光雷达到车体坐标系的外参精度更高。
根据第一方面或上述第一方面的各种可能的实现方式，在所述第一方面的第三种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项；所述根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定，包括：根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据所述标志物点云，标定所述激光雷达的偏航角。
基于上述技术方案，利用提取的高精度的地面点云，使得所标定的俯仰角和翻滚角精度更高；利用提取的高精度的标志物点云，使得所标定的偏航角的精度更高。
根据第一方面或上述第一方面的各种可能的实现方式，在所述第一方面的第四种可能的实现方式中，所述提取所采集的点云中的标志物点云，包括：在所采集的点云中过滤掉所述地面点云；沿垂直于车辆行进的方向，将过滤后的点云划分为多个切片；提取所述标志物点云，所述标志物点云包括满足预设条件的切片集合中的特征点，其中，所述切片集合包括相邻的一个或多个目标切片，所述目标切片中特征点数量超过阈值。
基于上述技术方案，通过划分点云切片，提取标志物点云，实现了自动提取高精度 的标志物点云。
根据第一方面或上述第一方面的各种可能的实现方式，在所述第一方面的第五种可能的实现方式中，所述方法还包括：获取地面点云的第一线束信息，根据所述第一线束信息，对所述地面点云进行降采样处理；和/或，获取标志物点云的第二线束信息，根据所述第二线束信息，对所述标志物点云进行降采样处理；根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定，包括：根据降采样处理后的所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
基于上述技术方案，依据各地面点的线束信息，降采样处理提取精确的地面点，在提升处理效率的同时，充分保留地面的纹理结构，从而保证了地面点云的精度。或者，依据各特征点的线束信息，降采样提取精确的特征物点云；在提升处理效率的同时，充分保留标志物的纹理结构，从而保证了标志物点云的精度。
根据第一方面或上述第一方面的各种可能的实现方式，在所述第一方面的第六种可能的实现方式中，所获取的点云为车辆在沿直线行驶状态下，激光雷达采集的点云。
基于上述技术方案，车辆在沿直线行驶状态下，道路两边的标志物与车辆前进方向平行，利用该状态下激光雷达采集的点云进行标定，从而提升了激光雷达的偏航角等外参精度。
根据第一方面或上述第一方面的各种可能的实现方式，在所述第一方面的第七种可能的实现方式中，所述标志物包括路沿、护栏、建筑物中的至少一项。
基于上述技术方案，标志物可以为道路的路沿、护栏、建筑物等，对场地的要求无特殊要求，无需额外设置标定板，靶标，反光贴等，降低了标定成本，在开放道路(例如城市街道、高速公路等)，即可完成在线标定。
第二方面，本申请的实施例提供了一种激光雷达的标定方法，所述方法包括：获取车辆通过目标区域时激光雷达采集的点云；所述目标区域的至少一边竖直设置有标志物；提取所采集的点云中的标志物点云；根据标志物点云，得到标志物的拟合线信息；所述拟合线信息包括拟合线的位置及方向信息；根据所述拟合线信息，得到所述激光雷达外参的数值。
基于上述技术方案，目标区域的至少一边竖直设置有标志物，标志物设置简单，降低了对场地的要求，施工成本低。根据标志物点云，得到标志物的拟合线信息，基于竖直标志物的拟合线需要符合竖直约束，得到激光雷达外参的数值，这样，根据标志物点云，即可计算激光雷达外参的数值。同时，由于标定过程中不依赖于地面点云，可以适用于地面信息不充分的场景(例如，小竖直视场角的激光雷达无法采集近处地面点云、由于场地尺寸受限侧激光雷达无法采集到有效的地面点云、安装俯仰角过大的激光雷达向上抬导致地面点云缺失或较少等等)，实现了地面信息不充分的场景中单激光雷达的高精度标定。此外，相对于建图标定等方式，可以自动执行整个标定过程，提高了单激光雷达标定的效率。
根据第二方面，在所述第二方面的第一种可能的实现方式中，所述外参包括俯仰角，所述方法还包括：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值大于第二大于预设阈值的情况下，根据标志物点云，对所述激光雷达的外参进行标定，其中，所述第二预设阈值由所述激光雷达的竖直视场角确定。
基于上述技术方案，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且俯仰角的数值大于第二大于预设阈值的情况下，表明激光雷达朝向偏上，地面点云可能不足，这样，根据标志物点云，即可完成对激光雷达外参的高精度标定；可以有效适用于地面信息不充分的场景。
根据第二方面的第一种可能的实现方式，在所述第二方面的第二种可能的实现方式中，所述外参包括偏航角；所述方法还包括：获取所述车辆的位置信息及所述激光雷达的位置信息；根据所述车辆的位置信息及所述激光雷达的位置信息，确定所述车辆的航向角；根据所述航向角，优化标定后的所述激光雷达的偏航角。
基于上述技术方案，考虑到车辆行驶过程中，较难实现直线行驶，可以不限制车辆的行驶偏角，结合车辆运动信息，优化标定的偏航角。
根据上述第二方面的各种可能的实现方式，在所述第二方面的第三种可能的实现方式中，所述车辆安装有主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述方法还包括：根据主激光雷达的标定后的外参及所述主激光雷达采集的多个标志物点云，确定所述多个标志物的位置信息；根据所述多个标志物的位置信息，得到第一标志物的预测位置；根据从激光雷达的标定后的外参及所述从激光雷达采集的所述第一标志物点云，得到所述第一标志物的测量位置；通过对所述预测位置与所述测量位置，优化所述从激光雷达的外参。
基于上述技术方案，基于单激光标定的结果，利用解算的标志物两两之间距离和/或朝向，预测其余标志物的位置，实现任意朝向多激光雷达间俯仰角、偏航角、翻滚角的联合优化。针对多激光雷达由于安装位置和角度偏差较大，点云投射到不同空间位置，不适合直接进行点云配准的场景，该方法可以有效提高标定的精度，从而实现了没有共视区域、或者共视区域较小的多激光雷达联合标定。此外，该方法不需要提前建图，显著提高了多激光雷达标定的效率。
根据上述第二方面的各种可能的实现方式，在所述第二方面的第四种可能的实现方式中，所述方法还包括：提取所采集的点云中的地面点云；所述根据标志物点云，对所述激光雷达的外参进行标定，还包括：根据标志物点云和所述地面点云，对所述激光雷达的外参进行标定。
基于上述技术方案，在存在有效地面点云的情况下，可以充分利用地面点云，从而进一步提高标定精度和稳定性。
根据第二方面或上述第二方面的各种可能的实现方式，在所述第二方面的第五种可能的实现方式中，所述根据标志物点云，得到标志物的拟合线信息，包括：根据标志物点云，确定旋转角初始值，所述旋转角初始值使标志物点云旋转后，在所述激光雷达坐标系中水平平面的投影区域最小；根据所述旋转角初始值，对标志物点云进行旋转处理；利用所述旋转后的标志物点云，得到标志物的拟合线信息。
根据上述第二方面的各种可能的实现方式，在所述第二方面的第六种可能的实现方式中，所述方法还包括：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据标志物点云及地面点云，对所述激光雷达的外参进行标定。
基于上述技术方案，在存在有效地面点云的情况下，可以充分利用地面点云，从而进一步提高标定精度和稳定性。
根据第二方面或上述第二方面的各种可能的实现方式，在所述第二方面的第七种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项，所述方法还包括：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据拟合线的位置信息，标定所述激光雷达的偏航角。
基于上述技术方案，可以利用地面点云，对激光雷达的俯仰角和翻滚角进行标定，提高了所标定的俯仰角和翻滚角精度和稳定性；利用拟合线的位置信息，对激光雷达的偏航角进行标定，提高了所标定的偏航角精度。
根据第二方面或上述第二方面的各种可能的实现方式，在所述第二方面的第八种可能的实现方式中，所述目标区域的至少一边竖直设置有多个标志物，且所述多个标志物与地面的交点在同一直线上。
第三方面，本申请的实施例提供了一种激光雷达的标定装置，所述装置包括：获取模块，用于获取车辆通过目标道路时激光雷达采集的点云，所述目标道路的至少一边设置有标志物；筛选模块，用于根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定；第一提取模块，用于对所述初步筛选出的点云进行多次拟合处理，得到地面点云；第二提取模块，用于提取所采集的点云中的标志物点云；标定模块，用于根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
根据第三方面，在所述第三方面的第一种可能的实现方式中，所述装置还包括：转换模块，用于根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息；第三提取模块，用于根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，其中，所述交叉域表示平行于所述车辆行进的方向，以所述交叉特征点为中心的区域；优化模块，用于根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化。
根据第三方面的第一种可能的实现方式，在所述第三方面的第二种可能的实现方式中，所述多个激光雷达包括主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方的环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述转换模块，还用于：根据主激光雷达的标定后的外参，将主激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述主激光雷达对应的标志物点云的位置信息及地面点云的位置信息；根据从激光雷达的标定后的外参，将所述从激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述从激光雷达对应的标志物点云的位置信息及地面点云的位置信息；所述第三提取模块，还用于：根据所述主激光雷达对应的地面点云的位置信息及所述从激光雷达对应的地面点云的位置信息，得到第一交叉特征点或第一交叉域；根据所述主激光雷达对应的标志物点云的位置信息及所述从激光雷达对应的标志物点云的位置信息，得到第二交叉特征点或第二交叉域。
所述优化模块，还用于：根据所述第一交叉特征点或第一交叉域，对所述从激光雷达标定后的俯仰角和翻滚角进行优化；根据所述第二交叉特征点或第二交叉域，对所述从激光雷达标定后的偏航角进行优化。
根据第三方面或上述第三方面的各种可能的实现方式，在所述第三方面的第三种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项；所述标定模块，还用于：根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据所述标志物点云，标定所述激光雷达的偏航角。
根据第三方面或上述第三方面的各种可能的实现方式，在所述第三方面的第四种可能的实现方式中，所述第二提取模块，还用于：在所采集的点云中过滤掉所述地面点云；沿垂直于车辆行进的方向，将过滤后的点云划分为多个切片；提取所述标志物点云，所述标志物点云包括满足预设条件的切片集合中的特征点，其中，所述切片集合包括相邻的一个或多个目标切片，所述目标切片中特征点数量超过阈值。
根据第三方面或上述第三方面的各种可能的实现方式，在所述第三方面的第五种可能的实现方式中，所述装置还包括降采样模块，用于：获取地面点云的第一线束信息，根据所述第一线束信息，对所述地面点云进行降采样处理；和/或，获取标志物点云的第二线束信息，根据所述第二线束信息，对所述标志物点云进行降采样处理；所述标定模块，还用于：根据降采样处理后的所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
根据第三方面或上述第三方面的各种可能的实现方式，在所述第三方面的第六种可能的实现方式中，所获取的点云为车辆在沿直线行驶状态下，激光雷达采集的点云。
根据第三方面或上述第三方面的各种可能的实现方式，在所述第三方面的第七种可能的实现方式中，所述标志物包括路沿、护栏、建筑物中的至少一项。
第四方面，本申请的实施例提供了一种激光雷达的标定装置，所述装置包括：获取模块，用于获取车辆通过目标区域时激光雷达采集的点云；所述目标区域的至少一边竖直设置有标志物；提取模块，用于提取所采集的点云中的标志物点云；拟合模块，用于根据标志物点云，得到标志物的拟合线信息；所述拟合线信息包括拟合线的位置及方向信息；计算模块，用于根据所述拟合线信息，得到所述激光雷达外参的数值。
根据第四方面，在所述第四方面的第一种可能的实现方式中，所述外参包括俯仰角；所述装置还包括：标定模块，用于在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，对所述激光雷达的外参进行标定，其中，所述第二预设阈值由所述激光雷达的竖直视场角确定。
根据第四方面的第一种可能的实现方式，在所述第四方面的第二种可能的实现方式中，所述外参包括偏航角；所述装置还包括：优化模块，用于获取所述车辆的位置信息及所述激光雷达的位置信息；根据所述车辆的位置信息及所述激光雷达的位置信息，确定所述车辆的航向角；根据所述航向角，优化标定后的所述激光雷达的偏航角。
根据上述第四方面的各种可能的实现方式，在所述第四方面的第三种可能的实现方式中，所述车辆安装有主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述 装置还包括：确定模块，用于根据主激光雷达的标定后的外参及所述主激光雷达采集的多个标志物点云，确定所述多个标志物的位置信息；预测模块，用于根据所述多个标志物的位置信息，得到第一标志物的预测位置；测量模块，用于根据从激光雷达的标定后的外参及所述从激光雷达采集的所述第一标志物点云，得到所述第一标志物的测量位置；匹配模块，用于通过对所述预测位置与所述测量位置，优化所述从激光雷达的外参。
根据上述第四方面的各种可能的实现方式，在所述第四方面的第四种可能的实现方式中，所述提取模块还用于：提取所采集的点云中的地面点云；所述标定模块，还用于：根据标志物点云和所述地面点云，对所述激光雷达的外参进行标定。
根据第四方面或上述第四方面的各种可能的实现方式，在所述第四方面的第五种可能的实现方式中，所述拟合模块，还用于：根据标志物点云，确定旋转角初始值，所述旋转角初始值使标志物点云旋转后，在所述激光雷达坐标系中水平平面的投影区域最小；根据所述旋转角初始值，对标志物点云进行旋转处理；利用所述旋转后的标志物点云，得到标志物的拟合线信息。
根据第四方面或上述第四方面的各种可能的实现方式，在所述第四方面的第六种可能的实现方式中，所述标定模块，还用于：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据标志物点云及地面点云，对所述激光雷达的外参进行标定。
根据第四方面或上述第四方面的各种可能的实现方式，在所述第四方面的第七种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项；所述标定模块，还用于：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据拟合线的位置信息，标定所述激光雷达的偏航角。
根据第四方面或上述第四方面的各种可能的实现方式，在所述第四方面的第八种可能的实现方式中，所述目标区域的至少一边竖直设置有多个标志物，且所述多个标志物与地面的交点在同一直线上。
第五方面，本申请的实施例提供了一种激光雷达的标定装置，包括：至少一个处理器；用于存储处理器可执行指令的存储器；其中，所述至少一个处理器被配置为执行所述指令时实现上述第一方面或者第一方面的一种或几种的激光雷达的标定方法，或者，实现上述第二方面或者第二方面的一种或几种的激光雷达的标定方法。
第六方面，本申请的实施例提供了一种非易失性计算机可读存储介质，其上存储有计算机程序指令，其特征在于，所述计算机程序指令被处理器执行时实现第一方面或者第一方面的一种或几种的激光雷达的标定方法，或者，实现上述第二方面或者第二方面的一种或几种的激光雷达的标定方法。
第七方面，本申请的实施例提供了一种计算机程序产品，包括计算机可读代码，或者承载有计算机可读代码的非易失性计算机可读存储介质，当所述计算机可读代码在电子设备中运行时，所述电子设备中的处理器实现第一方面或者第一方面的一种或几种的激光雷达的标定方法，或者，实现上述第二方面或者第二方面的一种或几种的激光雷达的标定方法。
上述第三方面至第七方面的各方面，及各方面的各种可能的实现方式的技术效果，参见上述第一方面或第二方面。
附图说明
图1示出了根据本申请一实施例的激光雷达的标定方法的一种应用场景的示意图；
图2(a)-图2(b)示出了根据本申请一实施例的车辆所在道路的示意图；
图3示出了根据本申请一实施例的激光雷达的标定方法的一种应用场景的示意图；
图4(a)-图4(e)示出了根据本申请一实施例的车辆所在道路的示意图；
图5(a)-图5(d)示出了根据本申请一实施例中几种坐标系的示意图；
图6示出了根据本申请一实施例中一种激光雷达标定方法的流程图；
图7示出了根据本申请一实施例中一种点云切片的示意图；
图8示出了根据本申请一实施例中一种激光雷达标定方法的流程图；
图9示出了根据本申请一实施例中一种时间同步的示意图；
图10示出了根据本申请一实施例中一种交叉特征点的示意图；
图11示出了根据本申请一实施例中一种交叉域的示意图；
图12示出了根据本申请一实施例中另一种激光雷达标定方法的流程图；
图13示出了根据本申请一实施例中一种单激光雷达标定的示意图；
图14示出了根据本申请一实施例中一种激光雷达扫描场景的示意图；
图15(a)-图15(b)示出了根据本申请一实施例中一种产线环境的示意图；
图16示出了根据本申请一实施例中标定俯仰角的对比示意图；
图17示出了根据本申请一实施例中一种激光雷达标定方法的流程图；
图18(a)-18(b)示出了根据本申请一实施例中一种多激光雷达联合优化的示意图；
图19示出了根据本申请一实施例中一种激光雷达标定装置的结构图；
图20示出了根据本申请一实施例中一种激光雷达标定装置的结构图；
图21示出根据本申请一实施例的一种激光雷达标定装置的结构示意图。
具体实施方式
以下将参考附图详细说明本申请的各种示例性实施例、特征和方面。附图中相同的附图标记表示功能相同或相似的元件。尽管在附图中示出了实施例的各种方面，但是除非特别指出，不必按比例绘制附图。
在这里专用的词“示例性”意为“用作例子、实施例或说明性”。这里作为“示例性”所说明的任何实施例不必解释为优于或好于其它实施例。
下面首先对本申请实施例提供的激光雷达的标定方法的应用场景进行举例说明。
图1示出了根据本申请一实施例的激光雷达的标定方法的一种应用场景的示意图。如图1所示，该应用场景中可以包括：车辆101、道路102、激光雷达103、标志物104；其中，激光雷达103安装在车辆101上，标志物104可以为路沿、路栏、建筑物立面等。
示例性地，道路103可以为开放道路，举例来说，图2(a)-图2(b)示出了根 据本申请一实施例的车辆所在道路的示意图，如图2(a)所示，道路103可以一条城市道路，该城市道路较为平整，且两边具有一定高度的路沿或护栏；如图2(b)所示，道路103可以为一条高速公路，该高速公路的路面较为平整、路沿较清晰。
在一些示例中，自车101可以在上述图2(a)或图2(b)所示的道路上行驶，自身车速可以大于40km/h，保持一段时间的直线行驶，在行驶过程中，激光雷达103扫描自车外部环境，完成点云采集工作。通过本申请实施例中激光雷达标定方法(具体描述参见下文)对激光雷达103采集的点云进行处理，从而实现激光雷达103的外参标定。
图3示出了根据本申请一实施例的激光雷达的标定方法的一种应用场景的示意图。如图3所示，该应用场景中可以包括：车辆301、道路302、激光雷达303、标志物304；其中，激光雷达303安装在车辆301上，标志物304可以竖直设置在道路303上，示例性地，道路303可以为长度为30-100m，宽度为3-8m的专用道路；示例性地，标志物304可以为金属材质，或者可以在标志物304表面贴有反光贴，从而提高激光反射率；标志物304的截面可以为圆形、方形、三角形等，且截面的几何特性已知。例如，可以为圆柱形竖直杆，竖直杆的截面类型可以为圆截面，且半径已知，竖直杆表面可以贴有反光贴；示例性地，标志物304的数量可以为多个，标志物304可以设置在道路302的一边或两边，在道路302的任一边设置有多个标志物304时，各标志物304与地面的交点在同一直线上，各标志物304之间的间距可以相同，也可以不同。
举例来说，图4(a)-图4(e)示出了根据本申请一实施例的车辆所在道路的示意图，如图4(a)所示，在道路的一边可以设置有一排标志物(图中以标志物设置在道路左边为例)，如图4(b)所示，在道路的两边分别设置有一排标志物，且两排标志物沿道路中心线对称分布，且任一排标志物中相邻标志物的间距相同；如图4(c)所示，在道路的两边分别设置有一排标志物，两排标志物平行分布，且任一排标志物中相邻标志物的间距相同；如图4(d)所示，在道路的两边分别设置有一排标志物，两排标志物平行分布，且任一排标志物中相邻标志物的不完全相同；如图4(e)所示，道路的路面可以不平整，道路可以设置上述图4(a)-图4(d)任一所示的标志物。
在一些示例中，自车301可以在上述图4(a)-图4(e)中任一所示的道路的一端驶入，并从道路的另一端驶出，车速可以为5-40km/h，可以保持匀速，也可以保持直线行驶状态，在行驶过程中，激光雷达303扫描自车外部环境，完成点云采集工作。通过本申请实施例中激光雷达标定方法(具体描述参见下文)对激光雷达303采集的点云进行处理，从而实现激光雷达303的外参标定。
需要说明的是，本申请实施例对车辆安装的激光雷达的数量及类型不作限定。示例性地，激光雷达103或激光雷达303的数量可以为一个或多个，上述图1及图3均以两个激光雷达作为示例，可以根据实际需要在自车101上安装更多的激光雷达103，或者在自车301上安装更多的激光雷达303。示例性地，上述激光雷达103或激光雷达303可以包括主激光雷达，主激光雷达用于探测车辆前方环境，或者探测车辆后方环境，或者可以探测车辆四周环境；还可以包括从激光雷达，从激光雷达可以用于探测车辆侧方环境(又称侧激光雷达)，或者用于探测车辆后方环境(又称后激光雷达)；主激光雷达相比从激光雷达，可以探测到车辆最前方的障碍物。
示例性地，上述自车101或自车301中还可以包括定位装置(图中未示出)，该定位装置可以包括轮速计、卫星导航系统(Global Navigation Satellite System，GNSS)、惯性导航系统(Inertial Navigation System，INS)等，用于获取车辆的位姿信息。
示例性地，图1或图3所示的应用场景中还可以包括激光雷达标定装置(图中未示出)，本申请实施例提供的激光雷达标定方法可以由该激光雷达标定装置实现，用于对上述激光雷达103或激光雷达303采集的点云进行高效率的自动化数据处理，且所标定的激光雷达的外参精度较高。
本申请实施例不限定该激光雷达标定装置的类型。
示例性地，该激光雷达标定装置可以为上述自车101(或自车301)，或者，自车101(或自车301)中的其他具有数据处理功能的部件，例如：车载终端、车载控制器、车载模块、车载模组、车载部件、车载芯片、车载单元、车载传感器，车辆可通过该车载终端、车载控制器、车载模块、车载模组、车载部件、车载芯片、车载单元、车载传感器等。
示例性地，该激光雷达标定装置集成在自车101或自车301的自动驾驶系统(Automated Driving System，ADS)或高级驾驶辅助系统(Advanced Driver Assistant Systems，ADAS)或车载计算平台等中。
示例性地，该激光雷达标定装置还可以为除了车辆之外的其他具有数据处理能力的智能终端，或设置在智能终端中的部件或者芯片。例如，该智能终端可以为智能运输设备、智能穿戴设备、智能家居设备、智能辅助飞机、机器人(robot)或无人机(unmanned aerial vehicle)等安装有激光雷达的设备。
示例性地，该激光雷达标定装置可以是一个通用设备或者是一个专用设备。在具体实现中，该装置还可以台式机、便携式电脑、网络服务器、掌上电脑(personal digital assistant，PDA)、移动手机、平板电脑、无线终端设备、嵌入式设备或其他具有数据处理功能的设备，或者为这些设备内的部件或者芯片。
示例性地，该激光雷达标定装置还可以是具有处理功能的芯片或处理器，该激光雷达标定装置可以包括多个处理器。处理器可以是一个单核(single-CPU)处理器，也可以是一个多核(multi-CPU)处理器。该具有处理功能的芯片或处理器可以设置在激光雷达中，也可以不设置在激光雷达中，而设置在激光雷达输出信号的接收端。
需要说明的是，本申请实施例描述的上述应用场景是为了更加清楚的说明本申请实施例的技术方案，并不构成对于本申请实施例提供的技术方案的限定，本领域普通技术人员可知，针对其他相似的或新的应用场景的出现，本申请实施例提供的技术方案对于类似的技术问题，同样适用。
图5(a)-图5(d)示出了根据本申请一实施例中几种坐标系的示意图，图5(a)-图5(b)所示为车体坐标系，车体坐标系的原点可以与车辆质心重合，当车辆在水平路面上处于静止状态时，x轴平行于地面指向车辆前方，z轴通过汽车质心指向上方，y轴指向驾驶员的左侧。示例性地，激光雷达的外参可以包括：俯仰角(pitch)、翻滚角(roll)、偏航角(yaw)中的一项或多项，还可以包括激光雷达的安装高度等；其中，俯仰角表示围绕y轴旋转的角度，偏航角表示围绕z轴旋转的角度，翻滚角表 示围绕x轴旋转的角度。图5(c)所示为激光雷达坐标系，激光雷达坐标系的原点可以与激光雷达质心重合，图5(d)所示为世界坐标系，在X’Y’平面可以看出激光雷达在车体坐标系的偏航角，车辆航向角及激光雷达航向角。
下面基于上述图1所述的应用场景，对本申请实施例提供的激光雷达标定方法进行详细说明。
图6示出了根据本申请一实施例中一种激光雷达标定方法的流程图；该方法可以由上述激光雷达标定装置执行；如图6所示，该方法可以包括以下步骤：
步骤601、获取车辆通过目标道路时激光雷达采集的点云，目标道路的至少一边设置有标志物。
其中，激光雷达可以为安装在车辆上的任一激光雷达，例如，可以为主激光雷达，也可以为侧激光雷达。
示例性地，目标道路可以为平整的开放道路，标志物可以为开放道路一边或两边的路沿、护栏或建筑物等；例如，车辆可以为上述自车101，目标道路可以为上述图2(a)或图2(b)中所示的道路。
示例性地，所获取的点云可以为车辆在沿直线行驶状态下激光雷达采集的点云。例如，可以为自车101在图2(a)或图2(b)所示的道路上沿直线行驶时，激光雷达103采集的点云。
示例性地，可以根据车辆通过目标道路时的位姿信息，确定车辆的行驶状态；例如，可以通过惯性导航系统，确定车辆通过目标道路处于直行状态时，提取车载激光雷达采集的点云。
步骤602、根据预设阈值，对所采集的点云进行初步筛选。
示例性地，预设阈值由激光雷达的安装高度确定，其中，激光雷达的安装高度可以预先确定；例如，预设阈值可以等于激光雷达的安装高度。
示例性地，所采集的点云中包括各扫描点在激光雷达坐标下的坐标信息(X值、Y值、Z值)；扫描点为地面点时，对应的高度值(即Z值)通常较小，且不会超过激光雷达的安装高度；因此，可以通过判断所采集的点云中各扫描点的高度值是否超过预设阈值，筛选出高度值不超过预设阈值的扫描点，所筛选出的所有扫描点可看作粗粒度地面点云。
该步骤中，根据由激光雷达的安装高度确定的预设阈值，对上述步骤601中所获取的任一帧点云进行初步筛选，筛选出粗粒度地面点云，从而减少点云数量，提高数据处理效率。
步骤603、对初步筛选出的点云进行多次拟合处理，得到地面点云。
示例性地，针对上述步骤602中初步筛选出的点云，可以运用多步随机采样一致性拟合算法，从而自适应地提取出高精度的地面点。
在一种可能的实现方式中，对于粗粒度地面点云，运用随机采样一致性算法拟合平面；过滤掉粗粒度地面点云中距离上述拟合平面距离大于预设距离阈值的扫描点，得到中粒度地面点云，其中，预设距离阈值的初值可以取经验值；缩小上述距离阈值，继续执行上述运用随机采样一致性算法拟合平面步骤，并进一步过滤掉中粒度地面点云中距离上述拟合平面距离大于缩小后的距离阈值的扫描点，从而得到细粒度的地面 点云。可以理解的，可以根据需要进一步通过多次迭代，执行缩小距离阈值和运用随机采样一致性算法拟合平面的操作，从而得到更加精确的地面点云。
在一种可能的实现方式中，还可以获取地面点云的第一线束信息，并根据第一线束信息，对地面点云进行降采样处理。其中，所采集的点云中可以包括各特征点的线束信息，或者，可以基于激光雷达的配置参数，确定所采集的点云中各特征点的线束信息。这样，对于上述得到的细粒度的地面点，依据各地面点的线束信息，采取间隔取点的方式降采样提取精确的地面点；在提升数据处理效率的同时，充分保留地面的纹理结构，从而保证了地面点云的精度。
通过上述步骤602及步骤603，无需人工操作，采用阈值过滤与多次拟合处理，自动提取精确的地面点云，还可以根据地面点的线束信息降采样得到最终的地面点云。
步骤604、提取所采集的点云中的标志物点云。
在一种可能的实现方式中，该步骤可以包括：在所采集的点云中过滤掉地面点云。过滤后的点云中扫描点可以包括标志物的特征点，或者道路中车辆、行人等非地面点；沿垂直于车辆行进的方向，将过滤后的点云划分为多个切片。提取标志物点云，标志物点云包括满足预设条件的切片集合中的特征点，其中，切片集合包括相邻的一个或多个目标切片，且目标切片中特征点数量超过预设阈值。这样，通过划分点云切片，提取标志物点云，实现了自动提取高精度的标志物点云。
示例性地，各切片的宽度、预设阈值可以根据需要进行设置；预设条件可以包括沿车辆行进的方向，切片集合中所有特征点的所在位置超过长度范围。
举例来说，图7示出了根据本申请一实施例中一种点云切片的示意图；如图7所示，以安装在车辆上的主激光雷达为例，图7中扫描点为：在主激光雷达所采集的一帧点云中过滤掉地面点云，过滤后的一帧点云中的扫描点。沿垂直于车辆行进的方向(如沿车体坐标系的y轴的方向)，将过滤后的点云划分为多个切片，即图中S1、S2…SN，N为正整数，其中，每个切片的宽度可以为1m；从切片S1开始，对于任一切片Si，其中，i为不大于N的正整数，若Si中扫描点数量大于预设阈值，则将Si标记为目标切片；对于目标切片Si，继续判断下一切片是否为目标切片，以此类推，直到当前切片不是目标切片或者目标切片的数量超多5个，例如，对于目标切片Si，若Si+1仍为目标切片，Si+1不是目标切片，则切片Si与Si+1即组成一个切片集合。这样，可以根据S1-SN中的N个切片得到一个或多个切片集合；对于任一切片集合，判断切片集合中所有特征点的X值范围是否超过长度范围，即判断是否满足|X max-X min|>L threshold，其中，X max表示切片集合中所有特征点的X值中的最大值，X min表示切片集合中所有特征点的X值中的最小值，L threshold表示长度范围；进而将满足条件的切片集合中的特征点作为标志物点云，从而实现提取高精度的标志物点云；如图7所示，切片S1即为道路左侧的标志物点云，切片SN即为道路右侧的标志物点云。
在一种可能的实现方式中，还可以获取标志物点云的第二线束信息，根据第二线束信息，对标志物点云进行降采样处理。这样，对于上述得到的标志物点云，依据各特征点的线束信息，采取间隔取点的方式降采样提取精确的特征物点云；在提升数据处理效率的同时，充分保留标志物的纹理结构，从而保证了标志物点云的精度。
步骤605、根据标志物点云及地面点云，对激光雷达的外参进行标定。
利用上述步骤603及步骤604提取的高精度的地面点云和标志物点云，对激光雷达的外参进行标定。
示例性地，可以根据上述降采样处理后的标志物点云及地面点云，对激光雷达的外参进行标定，从而提高处理效率。
在一种可能的实现方式中，可以根据上述地面点云，标定激光雷达的俯仰角和翻滚角；从而利用上述提取的高精度的地面点云，使得所标定的俯仰角和翻滚角精度更高。
示例性地，可以将地面与车体水平面平行作为约束，建立L1损失函数，解算激光雷达的俯仰角和翻滚角。
Figure PCTCN2021115418-appb-000001
在上述公式(1)中，L g表示解算俯仰角和翻滚角对应的损失函数，
Figure PCTCN2021115418-appb-000002
表示第i个地面点在车体坐标系中的z值，n表示地面点云中所包含的地面点的数量，
Figure PCTCN2021115418-appb-000003
表示地面点云z值的均值。利用公式(1)解算出，使得L g最小时对应的俯仰角和翻滚角，即为激光雷达的俯仰角和翻滚角。
进一步地，可以根据上述标志物点云，标定激光雷达的偏航角；从而利用上述提取的高精度的标志物点云，使得所标定的偏航角的精度更高。
示例性地，可以将标志物与车体前进方向平行作为约束，建立L1损失函数，解算激光雷达坐标系到车体坐标系的偏航角。
Figure PCTCN2021115418-appb-000004
在上述公式(2)中，L w表示解算偏航角对应的损失函数，
Figure PCTCN2021115418-appb-000005
表示第i个特征点在车体坐标系中的y值，m表示标志物点云中所包含的特征点的数量，
Figure PCTCN2021115418-appb-000006
表示标志物点云y值的均值。利用公式(1)解算出，使得L w最小时对应的偏航角，即为激光雷达的偏航角。
进一步地，对于步骤601中提取的一帧点云进行上述处理后，可以对积累一段时间的多帧点云，进行上述处理，并将得到的标定结果进行统计分析，从而优化得到最终的激光雷达到车体坐标系的外参。
这样，通过上述步骤601-605，利用开放道路中的场景信息，提取单个激光雷达采集的点云的地面点云及标志物点云，求解激光雷达到车体坐标系的外参，在一些示例中，可以通过多步拟合与阈值过滤相结合，提取高精度的地面点云，进而解算激光雷达的俯仰角和翻滚角，可以通过点云切片得提取高精度标志物点云，进而解算激光雷达的偏航角；从而实现单激光雷达的高精度在线动态标定。
本申请实施例中，标志物可以为道路的路沿、路栏等，对场地的要求无特殊要求，无需额外设置标定板，靶标，反光贴等，降低了标定成本，在开放道路(例如城市街道、高速公路等)，利用道路的自然场景即可完成激光雷达的在线动态标定。同时，通过自动提取所采集的点云中的标志物点云；并根据标志物点云及地面点云，对激光雷达的外参进行标定，从而实现全自动在线动态标定，无需人工操作，提高了标定效 率。此外，在提取地面点云时，根据预设阈值对所采集的点云进行初步筛选；对初步筛选出的点云进行多次拟合处理，从而基于阈值过滤与多次拟合处理，自适应地提取高精度的地面点云；还可以进一步通过点云切片，自动提取高精度的标志物点云，从而提高所标定的激光雷达的外参精度。
进一步地，随着智能驾驶领域的发展，低总价，小视角，高线束等激光雷达应用更加广泛，车载多个激光雷达可以实现场景的覆盖与互补，在车辆安装有多个激光雷达时，还可以根据任意两个激光雷达到车体坐标系的标定结果，对任一激光雷达的标定结果进行进一步优化。例如，可以根据通过上述步骤601-步骤605标定的主激光雷达的外参和通过上述步骤601-步骤605标定的从激光雷达的外参，对该从激光雷达的外参进行优化。
图8示出了根据本申请一实施例中一种激光雷达标定方法的流程图；如图8所示，该方法可以包括：
步骤801、根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息。
其中，多个激光雷达为安装在同一车辆上的激光雷达，示例性地，任一激光雷达的外参可以通过上述步骤601-步骤605进行标定。各激光雷达对应的标志物点云可以为上述步骤604中所提取的标志物点云，各激光雷达对应的地面点云可以为上述步骤603中得到的标志物点云。标志物点云的位置信息表示标志物点云在车体坐标系中的三维坐标(x值、y值、z值)。
示例性地，多个激光雷达可以包括主激光雷达和从激光雷达(例如，侧激光雷达)。
在一种可能的实现方式中，可以根据主激光雷达的标定后的外参，将主激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到主激光雷达对应的标志物点云的位置信息及地面点云的位置信息；根据侧激光雷达的标定后的外参，将侧激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到侧激光雷达对应的标志物点云的位置信息及地面点云的位置信息。
在一种可能的实现方式中，可以根据主激光雷达的标定后的外参，将主激光雷达采集的点云转换到车体坐标系中，得到主激光雷达所采集点云的位置信息，进而可以通过上述提取地面点云，得到主激光雷达对应的地面点云的位置信息；可以通过上述提取标志物点云的方式，得到主激光雷达对应的标志物点云的位置信息。相似地，可以根据侧激光雷达的标定后的外参，将侧激光雷达采集的点云转换到车体坐标系中，得到侧激光雷达所采集点云的位置信息，进而可以通过上述提取地面点云，得到侧激光雷达对应的地面点云的位置信息；可以通过上述提取标志物点云的方式，得到侧激光雷达对应的标志物点云的位置信息。
示例性地，在进行上述转换之前，可以对多个激光雷达之间进行时间同步，图9示出了根据本申请一实施例中一种时间同步的示意图，如图9所示，其中箭头指示时间轴的方向，时间轴上的点表示点云数据包，可以根据每一点云数据包的时间戳(即点云数据包在时间轴上对应的位置)，将主激光雷达的时间轴上的数据包与侧激光雷达的时间轴上的数据包进行匹配，若两个时间轴上相邻最近的数据包的时间差值小于阈值，即位于图9中所示的椭圆形区域内，则时间同步成功。
步骤802、根据各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域。
其中，交叉特征点表示两类点云的交点。交叉域表示平行于车辆行进的方向，以交叉特征点为中心的区域。
在一种可能的实现方式中，可以根据主激光雷达对应的地面点云的位置信息及侧激光雷达对应的地面点云的位置信息，得到第一交叉特征点或第一交叉域。
其中，第一交叉特征点表示主激光雷达对应的地面点云与侧激光雷达对应的地面点云的交点，该交点可以为一个地面点对，该地面点对包括主激光雷达对应的地面点云中一个地面点，及侧激光雷达对应的地面点云中一个地面点；例如，图10示出了根据本申请一实施例中一种交叉特征点的示意图，图10中所示的交叉特征点即为主激光雷达对应的地面点云与侧激光雷达对应的地面点云的交点。第一交叉域表示以第一交叉特征点为中心的区域，第一交叉域中可以包括多个地面点对，其中，每一地面点对中包括主激光雷达对应的地面点云中一个地面点，及侧激光雷达对应的地面点云中一个地面点；例如，图11示出了根据本申请一实施例中一种交叉域的示意图，如图11所示，椭圆中区域即为第一交叉域，每一椭圆中均包括多个地面点对。
示例性地，将主激光雷达对应的地面点云的位置信息及侧激光雷达对应的地面点云的位置信息，在车体坐标系中的xy平面上，以5m×5m网格进行网格化处理；在任一网格内，基于侧激光雷达对应的地面点云中任一地面点，寻找与其之间的曼哈顿距离最小的主激光雷达对应的地面点云中的地面点，其中，曼哈顿距离表示两个地面点在坐标系上的绝对轴距总和，例如，侧激光雷达对应的地面点云中地面点a(x 1,y 1)，与主激光雷达对应的地面点云中的地面点b(x 2,y 2)的曼哈顿距离为：|x 1-x 2|+|y 1-y 2|。进而根据预设距离阈值对得到的多对曼哈顿距离最小的地面点进行过滤，曼哈顿距离小于预设距离阈值的一对地面点，即为第一交叉特征点，如上述图10所示。进而，可以依据地面点的线束信息，以第一交叉特征点为中心，扩展一定数量的地面点对构成第一交叉域，如上述图11所示。
在一种可能的实现方式中，可以根据主激光雷达对应的标志物点云的位置信息及侧激光雷达对应的标志物点云的位置信息，得到第二交叉特征点或第二交叉域。
其中，第二交叉特征点表示主激光雷达对应的标志物点云与侧激光雷达对应的标志物点云的交点，该交点可以为一个特征点对，该特征点对包括主激光雷达对应的标志物点云中一个特征点，及侧激光雷达对应的标志物点云中一个特征点。第二交叉域表示以第二交叉特征点为中心的区域，第二交叉域中可以包括多个特征点对，其中，每一特征点对中包括主激光雷达对应的标志物点云中一个特征点，及侧激光雷达对应的标志物点云中一个特征点。
示例性地，将根据主激光雷达对应的标志物点云的位置信息及侧激光雷达对应的标志物点云的位置信息，在车体坐标系中的xz平面上，参照上述提取第一交叉特征点或第一交叉域的方式，得到第二交叉特征点或第二交叉域，此处不再赘述。
步骤803、根据交叉特征点或交叉域，对多个激光雷达中任一激光雷达的标定后外参进行优化。
在一种可能的实现方式中，可以根据第二交叉特征点或第二交叉域，对侧激光雷达 标定后的偏航角进行优化。
示例性地，可以构建目标函数：min|y 2-y 1|，优化补偿侧激光雷达的偏航角；其中，在第二交叉特征点的数量不小于预设阈值的情况下，可以采用第二交叉特征点求解该目标函数，此时，y 2与y 1分别表示各第二交叉特征点对应的特征点对的y值，例如，y 2可以表示一特征点对中，主激光雷达对应的标志物点云中特征点在车体坐标系中的y值，y 1可以表示该特征点对中，侧激光雷达对应的标志物点云中特征点在车体坐标系中的y值；在第二交叉特征点的数量小于预设阈值的情况下，可以采用第二交叉域求解该目标函数，此时，y 2表示第二交叉域中，主激光雷达对应的标志物点云中所有特征点在车体坐标系中的y值的均值，y 1表示该第二交叉域中，侧激光雷达对应的标志物点云中任一特征点在车体坐标系中的y值。
在一种可能的实现方式中，可以根据第一交叉特征点或第一交叉域，对侧激光雷达标定后的俯仰角和翻滚角进行优化。
示例性地，可以构建目标函数：min|z 2-z 1|，优化补偿侧激光雷达的俯仰角和翻滚角，其中，在第一交叉特征点的数量不小于预设阈值的情况下，可以采用第一交叉特征点求解该目标函数，此时，z 2与z 1分别表示各第一交叉特征点对应的地面点对的z值，例如，z 2可以表示一地面点对中，主激光雷达对应的地面点云中地面点在车体坐标系中的z值，z 1可以表示该地面点对中，侧激光雷达对应的地面点云中地面点在车体坐标系中的z值；在第一交叉特征点的数量小于预设阈值的情况下，可以采用第一交叉域求解该目标函数，此时，z 2表示第一交叉域中，主激光雷达对应的地面点云中所有地面点在车体坐标系中的z值的均值，z 1表示该第一交叉域中，侧激光雷达对应的地面点云中任一地面点在车体坐标系中的z值。
为了便于理解，对优化侧激光雷达的俯仰角和翻滚角的原理进行说明：旋转矩阵R补偿量可以表示为：
Figure PCTCN2021115418-appb-000007
在公式(3)中，α表示偏航角，β表示俯仰角，γ表示翻滚角；
Figure PCTCN2021115418-appb-000008
假设平移矩阵不补偿，在第一交叉特征点对应的地面点对存在如下关系：
Figure PCTCN2021115418-appb-000009
在公式(4)中，(x 1,y 1,z 1)表示一地面点对中侧激光雷达对应的地面点云中地面点在车体坐标系中的坐标值，(x 2,y 2,z 2)表示该地面点对中主激光雷达对应的地面点云中 地面点在车体坐标系中的坐标值。
基于上述优化补偿的偏航角，上述公式(4)展开可得：
z 2＝-sinβx 1+cosβsinγy 1+cosβcosγz 1……………………….(5)
在公式(5)中，z 2表示地面点对中主激光雷达对应的地面点云中地面点在车体坐标系中的z值，x 1、y 1、z 1分别表示该地面点对中侧激光雷达对应的地面点云中地面点在车体坐标系中x值、y值和z值。
将上公式(5)三角函数泰勒展开，可得：
z 2-z 1≈-βx 1+γy 1………………………………….(6)
由公式(6)可得，可以构建目标函数min|z 2-z 1|，采用蚁群算法等在一定范围内搜索使得|z 2-z 1|最小时，俯仰角和翻滚角的补偿量，从而利用该补偿量优化补偿上述标定的侧激光雷达的俯仰角和翻滚角。
进一步地，对于主激光雷达及侧激光雷达采集的一帧点云进行上述处理后，可以对积累一段时间的多帧点云，进行上述处理，并将得到的多组优化标定结果进行统计分析，从而得到最终的优化后侧激光雷达到车体坐标系的外参。
进一步地，可以根据优化补偿后的侧激光偏航角，更新激光雷达的标定参数；从而可以开启或更新智能驾驶功能，基于激光雷达优化补偿后的高精度外参，智能驾驶的感知、定位或融合等功能的精度得到提升。
这样，通过上述步骤801-步骤803，将各激光雷达采集的点云转换到车体坐标系中，获取各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，进而通过提取交叉特征点及交叉域特征优化激光雷达的外参；在一些示例中，可以将主激光点云和侧激光点云均通过对应标定后的外参转换到车体坐标系，并完成主激光雷达与侧激光雷达时间同步，进而在标志物点云的基础上，提取交叉特征点和交叉域优化补偿侧激光雷达到车体坐标系的偏航角；在地面点的基础上，提取交叉特征点和交叉域优化补偿侧激光雷达到车体坐标系的俯仰角和翻滚角，从而完成多激光雷达外参的联合优化，使得激光雷达到车体坐标系的外参精度更高。此外，对车辆行驶场地无特殊要求，可以在日常路段均可完成标定，路面也可以不平坦，凸凹、高低不平的路面或者标志物表面均可用于标定优化；同时，有效提高了标定外参的精度及标定效率。
上述图6或图8所示的激光雷达标定方法可以运用在城区或高架等通用场景标定、服务标定、用户自校准的标定等等场景；在一些示例中，在车辆日常使用过程中，随着时间推移，由于物体形变，温度，微小触碰等不确定性因素影响，会造成车载激光雷达外参变化；此时，车辆无需返厂，可以在开放城市道路上或高速路上保持直线行驶一小段距离，基于上述本申请实施例的基于开放道路的单激光雷达外参标定和/或多激光雷达联合优化的全自动在线标定方法，即可完成激光雷达外参的标定和优化，并且可以更新系统中外参，从而可以方便用户在日常实时在线调整和修正激光雷达的外参，保障智能驾驶功能的安全使用，提升智能驾驶性能。
下面基于上述图3所述的应用场景，对本申请实施例提供的激光雷达标定方法进行详细说明。
图12示出了根据本申请一实施例中另一种激光雷达标定方法的流程图；该方法可以由上述激光雷达标定装置执行；如图12所示，该方法可以包括以下步骤：
步骤1201、获取车辆通过目标区域时激光雷达采集的点云。
其中，激光雷达可以为安装在车辆上的任一激光雷达，例如，可以为主激光雷达，也可以为侧激光雷达。
目标区域的至少一边竖直设置有标志物。示例性地，目标区域的至少一边竖直设置有多个标志物，且多个标志物与地面的交点在同一直线上。例如，车辆可以为上述自车301，目标区域可以为上述图4(a)-图4(e)中任一所示的道路。
示例性地，所获取的点云可以为车辆沿直线匀速行驶状态下激光雷达采集的点云。例如，可以为自车301在图4(a)-图4(e)中任一所示的道路上匀速直线行驶时，激光雷达303采集的点云。
示例性地，可以根据车辆通过目标区域时的位姿信息，确定车辆的行驶状态；例如，可以通过惯性导航系统，确定车辆通过目标区域处于匀速直行状态时，提取车载激光雷达采集的点云。
举例来说，目标区域可以为如图4(a)所示道路，可以在道路左侧(或右侧)设置一排平行竖直标志物，标志物间等距，标志物可以为贴反光贴的圆柱形直杆；该目标区域易于施工，节约了成本。
目标区域可以为如图4(b)所示道路，目标区域的两侧对称地分布等距圆柱形直杆，两排圆柱形直杆具有平行关系，且左右两侧的圆柱形直杆对称分布；该目标区域的两侧均具有标志物，安装在车辆两侧的激光雷达均能够扫描到标志物，从而可以利用该目标区域标定安装在车辆两侧的从激光雷达；同时，对于前向主激光雷达，可以扫描到的标志物数量更多，从而提高外参标定的精度和稳定性。
目标区域可以为如图4(c)所示道路，目标区域的两侧交错地分布等距圆柱形直杆，两排圆柱形直杆具有平行关系，且左右两侧的圆柱形直杆交错分布，交错偏移的距离可以为同一侧相邻圆柱形直杆间距的一半。相对于左右两侧对称分布的方式，前向主激光雷达在该目标区域中扫描到的标志物数量在时域上变化减少，扫描最近标志物的距离会减少一半，因此，在最近标志物从扫描视野消失前后，不会出现扫描标志物前后帧距离跳动过大的问题，从而进一步提高外参标定的精度、稳定性及标定效率。
目标区域可以为如图4(d)所示道路，目标区域的两侧任意间距分布圆柱形直杆，两排圆柱形直杆具有平行关系。在该目标区域中，至少一排的圆柱形直杆可以不等间距，这样，对于目标区域的施工更加便捷，无需高精度测量和精准的施工，提高了使用灵活性和通用性。其中，相邻标志物的间距，可以根据需要，采用标志物点云拟合处理等方式估算得到。
步骤1202、提取所采集的点云中的标志物点云。
示例性地，可以根据标志物的材料特征，提取标志物点云。以标志物为圆柱形竖直杆为例，可以根据反射强度大小，在激光雷达所采集的点云中，筛选出圆柱形竖直杆点云。
示例性地，可以参照上述图6中所示的提取标志物点云的方式，提取步骤1201中所采集的点云中的标志物点云。
示例性地，还可以对所采集的点云进行运动去畸变处理后，提取精确的标志物点云。
步骤1203、根据标志物点云，得到标志物的拟合线信息。
其中，拟合线信息包括拟合线的位置及方向信息。示例性地，标志物的拟合线可以包括标志物的中心线、标志物的母线、标志物的边线等与地面垂直的直线，其中，标志物的中心线表示过标志物的上下截面中心的直线；本申请实施例以标志物的拟合线为标志物的中心线为例，对得到标志物的拟合线信息的方式进行示例性说明；示例性地，标志物的中心线可以用标志物矢量
Figure PCTCN2021115418-appb-000010
表示，其中，a，b，c表示标志物矢量l在方向矢量的三个分量，即中心线的方向信息；
Figure PCTCN2021115418-appb-000011
表示标志物矢量l与XY平面的交点的位置矢量，即中心线的位置信息。
在一种可能的实现方式中，该步骤可以包括：根据标志物点云，确定旋转角初始值，其中，旋转角初始值使标志物点云旋转后，在激光雷达坐标系中水平平面的投影区域最小；根据旋转角初始值，对标志物点云进行旋转处理；利用旋转后的标志物点云，得到标志物的拟合线信息。
示例性地，以标志物的拟合线为标志物的中心线为例，针对单组标志物点云，可以通过调整激光雷达旋转角R，使得标志物点云在激光雷达坐标系的Z＝0平面上，投影区域最小，得到旋转角初始值R init；进而可以根据R init，对标志物点云进行旋转处理，并利用旋转后的标志物点云，进行优化处理，得到标志物矢量l的方向矢量[a，b，c]及l与XY平面的交点位置
Figure PCTCN2021115418-appb-000012
从而得到标志物矢量l。
举例来说，图13示出了根据本申请一实施例中一种单激光雷达标定的示意图，如图13所示，旋转前的标志物矢量l(即标志物中心线)，旋转R后，得到竖直的标志物矢量l′(即方向矢量为单位矢量[0,0,1])，l与l′存在以下关系：
[0，0，1] T＝R[a，b，c]T……………………………….(7)
公式(7)中，[0，0，1] T表示单位矢量[0，0，1]的转置矩阵，[a，b，c] T表示矢量[a，b，c]的转置矩阵。
图13中，p i表示旋转前的标志物上的激光点i的坐标，旋转R后，得到旋转后的标志物上的激光点坐标为P i＝Rp i。
可以通过联立：θ＝<[a，b，c]，[0，0，1]>、ω＝[a，b，c]×[0，0，1]、
Figure PCTCN2021115418-appb-000013
及罗德里格(Rodrigues)旋转公式计算旋转矩阵
Figure PCTCN2021115418-appb-000014
解算出标志物矢量l的方向矢量，即l在单位矢量的三个分量a，b，c：其中，θ表示旋转角度，ω表示旋转轴信息，
Figure PCTCN2021115418-appb-000015
表示归一化旋转轴信息，<[a，b，c]，[0，0，1]>表示[a，b，c]与[0,0,1]的夹角。
进一步地，估算标志物矢量的位置矢量：
旋转前的标志物矢量l，旋转R后，在激光雷达坐标系中，得到的标志物矢量l′与XY平面的交点为
Figure PCTCN2021115418-appb-000016
其中，x l，y l为该交点的坐标值。
构建下述优化函数：
argmin0.5∑(f(p i，l)-r) 2........................................(8)
在公式(8)中，r为标志物截面的半径；
Figure PCTCN2021115418-appb-000017
Figure PCTCN2021115418-appb-000018
为激光点p i到标志物矢量l的距离；其中，
Figure PCTCN2021115418-appb-000019
为标志物矢量l上一点；
Figure PCTCN2021115418-appb-000020
表示
Figure PCTCN2021115418-appb-000021
与R[a，b，c] T矢量夹角，其数值为：
Figure PCTCN2021115418-appb-000022
即
Figure PCTCN2021115418-appb-000023
线段和竖直方向的夹角。
待优化量为a，b，c
Figure PCTCN2021115418-appb-000024
优化初值为R init，基于标志物截面的已知半径，通过优化求解上述公式(8)，即可得到标志物矢量l。
步骤1204、根据拟合线信息，得到激光雷达外参的数值。
其中，可以根据拟合线信息，得到激光雷达的俯仰角、翻滚角、偏航角中的至少一项的数值。
示例性地，可以根据上述得到的中心线的方向信息(如单根标志物矢量)，得到激光雷达的俯仰角的数值。由于单根标志物矢量的每个方向分量代表了标志物矢量在每个方向轴的投影长度，因此，可以使用单根标志物矢量的分量估算出俯仰角。例如，可以根据
Figure PCTCN2021115418-appb-000025
的方向矢量[a，b，c]中的a、c分量，通过atan(a/c)计算俯仰角的数值。
这样，可以根据激光雷达采集的一个标志物的点云，即可得到激光雷达外参的数值。
由于激光雷达的俯仰角对车辆的近处地面是否有足量激光点具有至关重要的作用；同时地面信息可以一定程度增加标定优化的精度和稳定性，因此，在激光雷达的朝向偏上时，还可以判断得到的激光雷达的俯仰角的数值是否大于第二预设阈值，即将俯仰角的数值作为是否使用地面信息进行标定优化的决策参考值；即可避免激光雷达安装偏差(例如，俯仰角)造成的地面信息不可用的问题，也可以充分利用有效的地面信息。
在一种可能的实现方式说，在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达的俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，对激光雷达的外参进行标定。
其中，第一预设阈值可以为90°，如果激光雷达朝向与竖直向上方向的夹角小于90°，则表明激光雷达的朝向偏上。第二预设阈值TH可以由激光雷达的竖直视场角(vertical-Field of View，V-FOV)确定；例如，TH＝0.5*(V-FOV)，即第二预设阈值可以为竖直视场角的一半。
示例性地，在激光雷达的朝向偏上，且激光雷达的俯仰角的数值大于第二预设阈值的情况下，可以根据标志物的中心线信息，对激光雷达的俯仰角、偏航角、翻滚角进行联合标定。由于标志物的中心线在世界坐标系中是竖直向上的，因此，可以利用中心线信息，同时标定激光雷达的俯仰角、偏航角、翻滚角。
示例性地，可以利用标志物矢量
Figure PCTCN2021115418-appb-000026
的位置矢量
Figure PCTCN2021115418-appb-000027
基于公式(9)所示的目标函数，联合标定激光雷达的俯仰角、偏航角、翻滚角。
argmin 0.5α∑∑(f(P′ ij，l j(0，0，1，d j))-r) 2+0.5β∑[z]..........(9)
在公式(9)中，α和β是残差项比例系数，r为标志物截面的半径，z为激光雷达的高度，j为所扫描到的标志物的编号，P′ ij为旋转后的第j个标志物的第i个激光点，l j(0，0，1，d j)表示第j个标志物矢量，d j为第j个标志物与XY平面的交点。其中，P′ ij＝R rollR pitchR yawP ij+[0，0，x] T，R yaw表示偏航角，R pitch表示俯仰角，R roll表示翻滚角；P ij为旋转前第j个标志物的第i个激光点。
该步骤中，不依赖于地面点云，在地面点云不足的情况下，利用标志物点云即可 实现激光雷达外参的高精度标定。针对具有较小的竖直视场角、无法扫描近处地面点云的激光雷达；由于场地宽度受限，无法采集到有效的地面点云的侧激光雷达；安装俯仰角过大，导致无法扫描地面点云或扫描到的地面点云较少的激光雷达；安装高度较高，扫描到的地面点云距离较远的激光雷达等等，均可实现高精度的外参标定。
举例来说，在激光雷达的视场角较小的场景中，图14示出了根据本申请一实施例中一种激光雷达扫描场景的示意图，如图14所示，主激光雷达竖直视场角较小、主激光雷达无法扫描到距离车辆较近的地面，即无法获取近处地面点云。若采用相关技术中依赖于地面信息标定主激光雷达外参的方案标定精度通常较低。因此，可以参照如图4(a)-图4(e)等场景在道路两边设置标志物，通过执行上述步骤1201-1204，可以仅利用主激光雷达采集的标志物点云，在主激光雷达的视场角较小的情况下，完成主激光雷达外参的高精度标定。
举例来说，在产线环境中，流水线宽度有限，图15(a)-图15(b)示出了根据本申请一实施例中一种产线环境的示意图，如图15(a)-图15(b)所示，由于场地宽度受限，车辆的侧激光雷达无法扫描到地面，即无法获取地面点云；或者扫描到的地面区域较小，即所获取的地面点云数量较少。若采用相关技术中依赖于地面信息标定侧激光雷达外参的方案，通常无法正常运行。因此，可以参照如图4(a)-图4(e)等场景在产线场地设置标志物，通过执行上述步骤1201-步骤1204，可以仅利用侧激光雷达采集的标志物点云，在空间狭窄的产线场地中，完成侧激光雷达外参的高精度标定。
举例来说，针对上述图4(e)中路面不平整的场景，道路上设置有标志物，通过执行上述步骤1201-步骤1204，从而仅利用激光雷达采集的标志物点云，在局部区域有上下起伏的道路中，完成侧激光雷达外参的高精度标定。图16示出了根据本申请一实施例中标定俯仰角的对比示意图，如图16所示，在不平整道路上，采用相关技术中依赖于地面信息标定主激光雷达外参的方案，所标定的俯仰角偏差较大；在该不平整道路上设置标志物(如图4(e)所示)，采用本申请实施例中方法，仅利用标志物点云标定俯仰角，能够减少或避免不平整的地面信息的影响，从而提升所标定的俯仰角的精度。
在一种可能的实现方式中，还可以提取所采集的点云中的地面点云，利用地面点云及标志物点云标定激光雷达外参，从而充分利用地面信息及标志物信息，提高外参标定精度；示例性地，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且俯仰角的数值大于第二预设阈值的情况下，根据标志物点云和地面点云，对激光雷达的外参进行标定。这样，在存在有效地面点云的情况下，可以充分利用地面点云，从而进一步提高标定精度和稳定性。
示例性地，可以利用标志物矢量
Figure PCTCN2021115418-appb-000028
的位置矢量
Figure PCTCN2021115418-appb-000029
及地面点云，基于公式(10)所示的目标函数，联合标定激光雷达的俯仰角、偏航角、翻滚角。
argmin 0.5α∑∑(f(P′ ij，l j(0，0，1，d j))-r) 2+0.5β∑P′ G[z].............(10)
在公式(10)中，α和β是残差项比例系数，r为标志物截面的半径，z为激光雷达的高度，j为所扫描到的标志物的标号，P ′ i j为旋转后的第j个标志物的第i个激光点，l j(0，0，1，d j)表示第j个标志物矢量，d j为第j个标志物与XY平面的交点；P′ G为旋转后的 地面点云。
在一帧点云中包含多个标志物对应的激光点，且标志物的间距未知的情况下：基于公式(10)优化激光雷达的俯仰角、偏航角、翻滚角及激光雷达的高度。此时，公式(10)中P′ ij＝R rollR pitchR yawP ij+[0，0，z] T，P′ G＝R rollR pitchR yawP G+[0，0，z] T，其中，z为激光雷达的高度，R yaw表示偏航角，R pitch表示俯仰角，R roll表示翻滚角；P ij为旋转前第j个标志物的第i个激光点，P G表示旋转前的地面点云。
在一帧点云中包含多个标志物对应的激光点，且标志物的间距已知的情况下：基于公式(10)，优化激光雷达的俯仰角、偏航角、翻滚角、激光雷达的高度及该帧点云中第一个标志物的位置d 0＝[x 0，y 0] T，其中，x 0，y 0为d 0的坐标值，设W j表示第j个标志物到第j+1个标志物的距离，则该帧点云中第二个标志物的位置为d 1＝[x 1+W 0，y 0] T。
进一步地，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达的俯仰角的数值不大于第二预设阈值的情况下，根据标志物点云及地面点云，对激光雷达的外参进行标定。
在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达的俯仰角的数值不大于第二预设阈值的情况下，激光雷达的朝向偏上，同时可以扫描到地面，即存在地面点云，因此，可以利用地面点云及标志物点云，对激光雷达的外参进行标定，从而充分利用地面信息，提高激光雷达的外参标定精度。
在一种可能的实现方式中，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达的俯仰角的数值不大于第二预设阈值的情况下，根据地面点云，标定激光雷达的俯仰角和翻滚角，还可以标定激光雷达的高度；根据拟合线的位置信息，标定激光雷达的偏航角。示例性地，可以利用标志物矢量
Figure PCTCN2021115418-appb-000030
的位置矢量
Figure PCTCN2021115418-appb-000031
标定激光雷达的偏航角。
这样，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达俯仰角的数值不大于预设阈值的情况下，激光雷达可以扫描到地面，存在地面点云，因此，可以利用地面点云，对激光雷达的俯仰角和翻滚角进行标定，提高了标定精度和稳定性。
考虑到车辆行驶过程中，较难实现直线行驶，本申请实施例中，不限制车辆的行驶偏角，可以在完成上述标定之后，结合车辆运动信息，优化标定结果。在一种可能的实现方式中：可以获取车辆的位置信息及激光雷达的位置信息；根据车辆的位置信息及激光雷达的位置信息，确定车辆的航向角；根据航向角，优化标定后的激光雷达的偏航角。
示例性地，基于标志物，激光雷达可以定位激光雷达的位置信息；车辆可以通过自身底盘等信息估计出车辆的位置信息；利用车辆的位置信息及激光雷达的位置信息，滤波估计出车辆的航向角；利用车辆航向角的平行约束，补偿偏航角的动态变化。
进一步地，对于步骤1201中提取的一帧点云进行上述处理后，可以对积累一段时间的多帧点云，进行上述处理，并将得到的标定结果进行统计分析，从而优化得到最终的激光雷达的外参。
这样，通过上述步骤1201-1204，基于竖直设置于目标区域的标志物(例如，一排平行的竖直标志物)，利用竖直标志物的拟合线需要符合竖直约束的原理，得到激光 雷达外参的数值；在一些示例中，可以基于计算的俯仰角的数值，在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值时，对俯仰角数值是否大于第二预设阈值进行判断，从而能够自动化判断地面点云的可用性，提高标定效率和自动化；并仅利用标志物点云，即可同时标定单激光雷达的俯仰角、偏航角及翻滚角，在一些示例中，可以结合车辆运动信息补偿偏航角的动态变化，从而实现了单激光雷达的高精度动态标定。
本申请实施例中，目标区域的至少一边竖直设置有标志物，标志物设置简单，降低了对场地的要求，施工成本低。根据标志物点云，得到标志物的拟合线信息，基于竖直标志物的拟合线需要符合竖直约束，得到激光雷达外参的数值，这样，根据标志物点云，即可计算激光雷达外参的数值；在一些示例中，可以在激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且激光雷达俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，即可完成对激光雷达外参的高精度标定；同时，由于标定过程中不依赖于地面点云，可以适用于地面信息不充分的场景(例如，小竖直视场角的激光雷达无法采集近处地面点云、由于场地尺寸受限侧激光雷达无法采集到有效的地面点云、安装俯仰角过大的激光雷达向上抬导致地面点云缺失或较少等等)，实现了地面信息不充分的场景中单激光雷达的高精度标定。此外，相对于建图标定等方式，可以自动执行整个标定过程，提高了单激光雷达标定的效率。
进一步地，在车辆安装有多个激光雷达时，还可以根据任意两个激光雷达到车体坐标系的标定结果，对任一激光雷达的标定结果进行进一步优化，实现多激光标定。例如，可以根据通过上述步骤1201-步骤1204标定的主激光雷达的外参和通过上述步骤1201-步骤1204标定的从激光雷达的外参，对该从激光雷达的外参进行优化。
示例性地，主激光雷达与从激光雷达可以没有共视区域，或者共视区域较小。在一帧点云中，主激光雷达可以扫描到至少两个标志物；通过预测及匹配标志物的位置，从而优化从激光雷达的外参。
图17示出了根据本申请一实施例中一种激光雷达标定方法的流程图；如图17所示，该方法可以包括以下步骤：
步骤1701、根据主激光雷达的标定后的外参及主激光雷达采集的多个标志物点云，确定多个标志物的位置信息。
其中，主激光雷达的外参可以为通过上述步骤1201-步骤1204所标定的外参。多个标志物的位置信息可以包括基于上述标定外参转换后的多个标志物的拟合线的位置信息。
示例性地，可以用主激光雷达采集的多个标志物点云，拟合得到多个标志物的位置矢量
Figure PCTCN2021115418-appb-000032
进而利用主激光雷达的标定后的外参，将位置矢量
Figure PCTCN2021115418-appb-000033
转换到竖直方向，此时得到的位置矢量即为多个标志物的位置信息。
在一段时间内，主激光雷达连续跟踪标志物，通过扫描先后顺序，判断并记录标志物的唯一编号j，同时，可以测量出任意两个标志物间的距离Wj；设Wj表示第j个标志物到第j+1个标志物的距离。
举例来说，图18(a)-18(b)示出了根据本申请一实施例中一种多激光雷达联合优化的示意图；如图18(a)所示，各标志物之间可以间距相等，主激光雷达对连续 跟踪到的标志物进行编号，即W0、W1…Wj；如图18(b)所示，激光雷达当前可以跟踪到标志物W0、W1，同时可以基于确定的标志物W0、W1的位置信息，得到W0、W1之间的距离。
步骤1702、根据多个标志物的位置信息，得到第一标志物的预测位置。
其中，第一标志物为从激光雷达可以跟踪到的、在上述多个标志物之后的标志物。
示例性地，可以利用第一标志物的编号和标志物间的距离Wj，推算出第一标志物的预测位置。第一标志物的预测位置可以包括第一标志物的拟合线的位置信息(如，第一标志物的中心线的位置信息)。
在一种可能的实现方式中，可以先进行方向预测，再进行距离预测，从而得到第一标志物的预测位置；示例性地，由于同一侧的多个竖直标志物平行，基于两个标志物的位置信息，即可确定一条直线，可知其余标志物均在这条直线上，进而结合两个标志物的先后顺序，即可得到该直线上标志物的排布方向；然后基于该直线上标志物的排布方向，通过主激光雷达跟踪标志物估计出的距离Wj，如W2,W3，估计出第一标志物的预测位置。例如，如图18(a)所示，从激光雷达可以跟踪到W2，即第一标志物可以为W2，如图18(b)所示，可以利用W0、W1的位置信息，确定W0、W1所在直线上标志物排布方向，进而基于该方向，利用上述W0、W1之间的距离，及W1的位置信息，推算得到W2的预测位置。
步骤1703、根据从激光雷达的标定后的外参及从激光雷达采集的第一标志物点云，得到第一标志物的测量位置。
其中，从激光雷达的外参可以为通过上述步骤1201-步骤1204所标定的外参。示例性地，第一标志物的位置信息可以包括基于上述标定外参转换后的第一标志物的中心线的位置信息。
示例性地，可以利用从激光雷达采集的第一标志物点云，按照前文所述方式，得到第一标志物的位置矢量
Figure PCTCN2021115418-appb-000034
进而利用从激光雷达的标定后的外参，将位置矢量
Figure PCTCN2021115418-appb-000035
转换到竖直方向，此时得到的位置矢量即为第一标志物的测量位置。
步骤1704、通过预测位置与测量位置，优化从激光雷达的外参。
利用上述第一标志物的预测位置和从激光雷达实际的测量位置，优化侧激光雷达的外参，使得第一标志物的位置矢量重合度最佳。示例性地，可以利用从激光雷达相对于主激光雷达的外参，将测量位置进行变换，进而利用变换后的测量位置与预测位置进行匹配处理，完成外参的联合优化。可以理解的是，如果从激光雷达相对于主激光雷达的外参存在偏差，则从激光雷达得到的测量位置经过存在偏差的相对外参转换后，与预测位置不一致。
示例性地，可以采用最小二乘形式，基于优化目标函数：0.5[(Δa) 2+(Δb) 2+(Δc) 2]优化求解，其中，Δa，Δb，Δc分别是主激光雷达-从激光雷达的方向矢量三个分量的差。
这样，通过上述步骤1701-步骤1704，基于单激光标定的结果，利用解算的标志物两两之间距离和/或朝向，预测其余标志物的位置，实现任意朝向多激光雷达间俯仰角、偏航角、翻滚角的联合优化。针对多激光雷达由于安装位置和角度偏差较大，点云投射到不同空间位置，不适合直接进行点云配准的场景，该方法可以有效提高标定 的精度，从而实现了没有共视区域、或者共视区域较小的多激光雷达联合标定。此外，该方法不需要提前建图，显著提高了多激光雷达标定的效率。
上述图12或图17所示的激光雷达标定方法可以运用在产线末端标定、服务标定、存在高清地图的在线自校准等等场景中；在一些示例中，产线末端标定需要满足多车型、多传感器适配；快速标定；成本低，场地简单，能够在不同工厂中推广和建造；需要有较大角度标定容忍范围，安装存在较大偏差时，需要有异常报警；全天候工作，导航系统不可用时也要能正常工作等等需求；此时，车辆从本申请实施例提供的图4(a)-图4(e)中任一道路一端驶入从另一端驶出，车辆通过短距离的行驶，进而基于上述本申请实施例的单激光雷达外参标定和/或多激光雷达联合优化的方法，即可完成车载激光雷达标定，满足产线末端标定的需求；并且可以更新系统中外参，保障智能驾驶功能的安全使用，提升智能驾驶性能。
基于上述方法实施例的同一发明构思，本申请的实施例还提供了一种激光雷达的标定装置，该激光雷达的标定装置可以用于执行上述方法实施例所描述的技术方案。
图19示出了根据本申请一实施例中一种激光雷达标定装置的结构图，如图19所示，所述装置包括：获取模块1901，用于获取车辆通过目标道路时激光雷达采集的点云，所述目标道路的至少一边设置有标志物；筛选模块1902，用于根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定；第一提取模块1903，用于对所述初步筛选出的点云进行多次拟合处理，得到地面点云；第二提取模块1904，用于提取所采集的点云中的标志物点云；标定模块1905，用于根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
在一种可能的实现方式中，所述装置还包括：转换模块，用于根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息；第三提取模块，用于根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，其中，所述交叉域表示平行于所述车辆行进的方向，以所述交叉特征点为中心的区域；优化模块，用于根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化。
在一种可能的实现方式中，所述多个激光雷达包括主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方的环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述转换模块，还用于：根据主激光雷达的标定后的外参，将主激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述主激光雷达对应的标志物点云的位置信息及地面点云的位置信息；根据从激光雷达的标定后的外参，将所述从激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述从激光雷达对应的标志物点云的位置信息及地面点云的位置信息；所述第三提取模块，还用于：根据所述主激光雷达对应的地面点云的位置信息及所述从激光雷达对应的地面点云的位置信息，得到第一交叉特征点或第一交叉域；根据所述主激光雷达对应的标志物点云的位置信息及所述从激光雷达对应的标志物点云的位置信息，得到第二交叉特征点或第二交叉域；所述优化模块，还用于：根据所述第一交叉特征点或第一交叉域，对所述从激光雷达标定后的俯仰角和翻滚角进行优化；根据所述第二交叉特征点或第二交叉域，对所述从激光雷达标定后的偏航角进行优化。
在一种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项；所述标定模块，还用于：根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据所述标志物点云，标定所述激光雷达的偏航角。
在一种可能的实现方式中，所述第二提取模块，还用于：在所采集的点云中过滤掉所述地面点云；沿垂直于车辆行进的方向，将过滤后的点云划分为多个切片；提取所述标志物点云，所述标志物点云包括满足预设条件的切片集合中的特征点，其中，所述切片集合包括相邻的一个或多个目标切片，所述目标切片中特征点数量超过阈值。
在一种可能的实现方式中，所述装置还包括降采样模块，用于：获取地面点云的第一线束信息，根据所述第一线束信息，对所述地面点云进行降采样处理；和/或，获取标志物点云的第二线束信息，根据所述第二线束信息，对所述标志物点云进行降采样处理；所述标定模块，还用于：根据降采样处理后的所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
在一种可能的实现方式中，所获取的点云为车辆在沿直线行驶状态下，激光雷达采集的点云。
在一种可能的实现方式中，所述标志物包括路沿、护栏、建筑物中的至少一项。
上述实施例中，激光雷达标定装置及其各种可能的实现方式的技术效果及具体描述可参见上述激光雷达标定方法，此处不再赘述。
图20示出了根据本申请一实施例中一种激光雷达标定装置的结构图，如图20所示，所述装置包括：获取模块2001，用于获取车辆通过目标区域时激光雷达采集的点云；所述目标区域的至少一边竖直设置有标志物；提取模块2002，用于提取所采集的点云中的标志物点云；拟合模块2003，用于根据标志物点云，得到标志物的拟合线信息；所述拟合线信息包括拟合线的位置及方向信息；计算模块2004，用于根据所述拟合线信息，得到所述激光雷达外参的数值；
在一种可能的实现方式中，所述外参包括俯仰角；所述装置还包括：标定模块，用于在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，对所述激光雷达的外参进行标定，其中，所述第二预设阈值由所述激光雷达的竖直视场角确定。
在一种可能的实现方式中，所述外参包括偏航角；所述装置还包括：优化模块，用于获取所述车辆的位置信息及所述激光雷达的位置信息；根据所述车辆的位置信息及所述激光雷达的位置信息，确定所述车辆的航向角；根据所述航向角，优化标定后的所述激光雷达的偏航角。
在一种可能的实现方式中，所述车辆安装有主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述装置还包括：确定模块，用于根据主激光雷达的标定后的外参及所述主激光雷达采集的多个标志物点云，确定所述多个标志物的位置信息；预测模块，用于根据所述多个标志物的位置信息，得到第一标志物的预测位置；测量模块，用于根据从激光雷达的标定后的外参及所述从激光雷达采集的所述第一标志物点云，得到所述第一标志物的测量位置；匹配模块，用于通过对所述预测位置与所述测量位置，优化所述从激光雷达的外参。
在一种可能的实现方式中，所述提取模块还用于：提取所采集的点云中的地面点云；所述标定模块，还用于：根据标志物点云和所述地面点云，对所述激光雷达的外参进行标定。
在一种可能的实现方式中，所述拟合模块，还用于：根据标志物点云，确定旋转角初始值，所述旋转角初始值使标志物点云旋转后，在所述激光雷达坐标系中水平平面的投影区域最小；根据所述旋转角初始值，对标志物点云进行旋转处理；利用所述旋转后的标志物点云，得到标志物的拟合线信息。
在一种可能的实现方式中，所述标定模块，还用于：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据标志物点云及地面点云，对所述激光雷达的外参进行标定。
在一种可能的实现方式中，所述外参包括俯仰角、翻滚角、偏航角中的至少一项；所述标定模块，还用于：在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据拟合线的位置信息，标定所述激光雷达的偏航角。
在一种可能的实现方式中，所述目标区域的至少一边竖直设置有多个标志物，且所述多个标志物与地面的交点在同一直线上。
上述实施例中，激光雷达标定装置及其各种可能的实现方式的技术效果及具体描述可参见上述激光雷达标定方法，此处不再赘述。
本申请的实施例提供了一种激光雷达标定装置，包括：处理器以及用于存储处理器可执行指令的存储器；其中，所述处理器被配置为执行所述指令时实现上述激光雷达标定方法。
图21示出根据本申请一实施例的一种激光雷达标定装置的结构示意图，如图21所示，该激光雷达的标定装置可以包括：至少一个处理器2101，通信线路2102，存储器2103以及至少一个通信接口2104。
处理器2101可以是一个通用中央处理器(central processing unit，CPU)，微处理器，特定应用集成电路(application-specific integrated circuit，ASIC)，或一个或多个用于控制本申请方案程序执行的集成电路。
通信线路2102可包括一通路，在上述组件之间传送信息。
通信接口2104，使用任何收发器一类的装置，用于与其他设备或通信网络通信，如以太网，RAN，无线局域网(wireless local area networks，WLAN)等。
存储器2103可以是只读存储器(read-only memory，ROM)或可存储静态信息和指令的其他类型的静态存储设备，随机存取存储器(random access memory，RAM)或者可存储信息和指令的其他类型的动态存储设备，也可以是电可擦可编程只读存储器(electrically erasable programmable read-only memory，EEPROM)、只读光盘(compact disc read-only memory，CD-ROM)或其他光盘存储、光碟存储(包括压缩光碟、激光碟、光碟、数字通用光碟、蓝光光碟等)、磁盘存储介质或者其他磁存储设备、或者能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质，但不限于此。存储器可以是独立存在，通过通信线路2102与处理器 相连接。存储器也可以和处理器集成在一起。本申请实施例提供的存储器通常可以具有非易失性。其中，存储器2103用于存储执行本申请方案的计算机执行指令，并由处理器2101来控制执行。处理器2101用于执行存储器2103中存储的计算机执行指令，从而实现本申请上述实施例中提供的方法。
可选的，本申请实施例中的计算机执行指令也可以称之为应用程序代码，本申请实施例对此不作具体限定。
示例性地，处理器2101可以包括一个或多个CPU，例如图21中的CPU0和CPU1。
示例性地，激光雷达的标定装置可以包括多个处理器，例如图21中的处理器2101和处理器2107。这些处理器中的每一个可以是一个单核(single-CPU)处理器，也可以是一个多核(multi-CPU)处理器。这里的处理器可以指一个或多个设备、电路、和/或用于处理数据(例如计算机程序指令)的处理核。
在具体实现中，作为一种实施例，激光雷达标定装置还可以包括输出设备2105和输入设备2106。输出设备2105和处理器2101通信，可以以多种方式来显示信息。例如，输出设备2105可以是液晶显示器(liquid crystal display，LCD)，发光二级管(light emitting diode，LED)显示设备，阴极射线管(cathode ray tube，CRT)显示设备，或投影仪(projector)等。输入设备2106和处理器2101通信，可以以多种方式接收用户的输入。例如，输入设备2106可以是鼠标、键盘、触摸屏设备或传感设备等。
本申请实施例还提供一种激光雷达的标定系统，所述激光雷达的标定系统，包括至少一个本申请上述实施例提到的激光雷达的标定装置。
本申请实施例还提供一种车辆，所述车辆包括至少一个本申请上述实施例提到的激光雷达的标定装置，或激光雷达的标定系统。
本申请的实施例提供了一种非易失性计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令被处理器执行时实现上述方法。
本申请的实施例提供了一种计算机程序产品，包括计算机可读代码，或者承载有计算机可读代码的非易失性计算机可读存储介质，当所述计算机可读代码在电子设备的处理器中运行时，所述电子设备中的处理器执行上述方法。
计算机可读存储介质可以是可以保持和存储由指令执行设备使用的指令的有形设备。计算机可读存储介质例如可以是――但不限于――电存储设备、磁存储设备、光存储设备、电磁存储设备、半导体存储设备或者上述的任意合适的组合。计算机可读存储介质的更具体的例子(非穷举的列表)包括：便携式计算机盘、硬盘、随机存取存储器(Random Access Memory，RAM)、只读存储器(Read Only Memory，ROM)、可擦式可编程只读存储器(Electrically Programmable Read-Only-Memory，EPROM或闪存)、静态随机存取存储器(Static Random-Access Memory，SRAM)、便携式压缩盘只读存储器(Compact Disc Read-Only Memory，CD-ROM)、数字多功能盘(Digital Video Disc，DVD)、记忆棒、软盘、机械编码设备、例如其上存储有指令的打孔卡或凹槽内凸起结构、以及上述的任意合适的组合。
这里所描述的计算机可读程序指令或代码可以从计算机可读存储介质下载到各个计算/处理设备，或者通过网络、例如因特网、局域网、广域网和/或无线网下载到外部计算机或外部存储设备。网络可以包括铜传输电缆、光纤传输、无线传输、路由器、 防火墙、交换机、网关计算机和/或边缘服务器。每个计算/处理设备中的网络适配卡或者网络接口从网络接收计算机可读程序指令，并转发该计算机可读程序指令，以供存储在各个计算/处理设备中的计算机可读存储介质中。
用于执行本申请操作的计算机程序指令可以是汇编指令、指令集架构(Instruction Set Architecture，ISA)指令、机器指令、机器相关指令、微代码、固件指令、状态设置数据、或者以一种或多种编程语言的任意组合编写的源代码或目标代码，所述编程语言包括面向对象的编程语言—诸如Smalltalk、C++等，以及常规的过程式编程语言—诸如“C”语言或类似的编程语言。计算机可读程序指令可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络—包括局域网(Local Area Network，LAN)或广域网(Wide Area Network，WAN)—连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。在一些实施例中，通过利用计算机可读程序指令的状态信息来个性化定制电子电路，例如可编程逻辑电路、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或可编程逻辑阵列(Programmable Logic Array，PLA)，该电子电路可以执行计算机可读程序指令，从而实现本申请的各个方面。
这里参照根据本申请实施例的方法、装置(系统)和计算机程序产品的流程图和/或框图描述了本申请的各个方面。应当理解，流程图和/或框图的每个方框以及流程图和/或框图中各方框的组合，都可以由计算机可读程序指令实现。
这些计算机可读程序指令可以提供给通用计算机、专用计算机或其它可编程数据处理装置的处理器，从而生产出一种机器，使得这些指令在通过计算机或其它可编程数据处理装置的处理器执行时，产生了实现流程图和/或框图中的一个或多个方框中规定的功能/动作的装置。也可以把这些计算机可读程序指令存储在计算机可读存储介质中，这些指令使得计算机、可编程数据处理装置和/或其他设备以特定方式工作，从而，存储有指令的计算机可读介质则包括一个制造品，其包括实现流程图和/或框图中的一个或多个方框中规定的功能/动作的各个方面的指令。
也可以把计算机可读程序指令加载到计算机、其它可编程数据处理装置、或其它设备上，使得在计算机、其它可编程数据处理装置或其它设备上执行一系列操作步骤，以产生计算机实现的过程，从而使得在计算机、其它可编程数据处理装置、或其它设备上执行的指令实现流程图和/或框图中的一个或多个方框中规定的功能/动作。
附图中的流程图和框图显示了根据本申请的多个实施例的装置、系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段或指令的一部分，所述模块、程序段或指令的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个连续的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。
也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框 的组合，可以用执行相应的功能或动作的硬件(例如电路或ASIC(Application Specific Integrated Circuit，专用集成电路))来实现，或者可以用硬件和软件的组合，如固件等来实现。
尽管在此结合各实施例对本发明进行了描述，然而，在实施所要求保护的本发明过程中，本领域技术人员通过查看所述附图、公开内容、以及所附权利要求书，可理解并实现所述公开实施例的其它变化。在权利要求中，“包括”(comprising)一词不排除其他组成部分或步骤，“一”或“一个”不排除多个的情况。单个处理器或其它单元可以实现权利要求中列举的若干项功能。相互不同的从属权利要求中记载了某些措施，但这并不表示这些措施不能组合起来产生良好的效果。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应以所述权利要求的保护范围为准。

Claims (26)

  1. 一种激光雷达的标定方法，其特征在于，所述方法包括：
    获取车辆通过目标道路时激光雷达采集的点云，所述目标道路的至少一边设置有标志物；
    根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定；
    对所述初步筛选出的点云进行多次拟合处理，得到地面点云；
    提取所采集的点云中的标志物点云；
    根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息；
    根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，其中，所述交叉域表示平行于所述车辆行进的方向，以所述交叉特征点为中心的区域；
    根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化。
  3. 根据权利要求2所述的方法，其特征在于，所述多个激光雷达包括主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方的环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；
    所述根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，包括：
    根据主激光雷达的标定后的外参，将主激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述主激光雷达对应的标志物点云的位置信息及地面点云的位置信息；
    根据从激光雷达的标定后的外参，将所述从激光雷达对应的标志物点云及地面点云转换到车体坐标系中，得到所述从激光雷达对应的标志物点云的位置信息及地面点云的位置信息；
    所述根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，包括：
    根据所述主激光雷达对应的地面点云的位置信息及所述从激光雷达对应的地面点云的位置信息，得到第一交叉特征点或第一交叉域；根据所述主激光雷达对应的标志物点云的位置信息及所述从激光雷达对应的标志物点云的位置信息，得到第二交叉特征点或第二交叉域；
    所述根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化，包括：
    根据所述第一交叉特征点或第一交叉域，对所述从激光雷达标定后的俯仰角和翻滚 角进行优化；根据所述第二交叉特征点或第二交叉域，对所述从激光雷达标定后的偏航角进行优化。
  4. 根据权利要求1-3中任意一项所述的方法，其特征在于，所述外参包括俯仰角、翻滚角、偏航角中的至少一项，
    所述根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定，包括：
    根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；
    根据所述标志物点云，标定所述激光雷达的偏航角。
  5. 根据权利要求1-4中任意一项所述的方法，其特征在于，所述提取所采集的点云中的标志物点云，包括：
    在所采集的点云中过滤掉所述地面点云；
    沿垂直于车辆行进的方向，将过滤后的点云划分为多个切片；
    提取所述标志物点云，所述标志物点云包括满足预设条件的切片集合中的特征点，其中，所述切片集合包括相邻的一个或多个目标切片，所述目标切片中特征点数量超过阈值。
  6. 根据权利要求1-5中任意一项所述的方法，其特征在于，所述方法还包括：
    获取地面点云的第一线束信息，
    根据所述第一线束信息，对所述地面点云进行降采样处理；
    和/或，
    获取标志物点云的第二线束信息，
    根据所述第二线束信息，对所述标志物点云进行降采样处理；
    根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定，包括：
    根据降采样处理后的所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
  7. 根据权利要求1-6中任意一项所述的方法，其特征在于，所获取的点云为车辆在沿直线行驶状态下，激光雷达采集的点云。
  8. 根据权利要求1-7中任意一项所述的方法，其特征在于，所述标志物包括路沿、护栏、建筑物中的至少一项。
  9. 一种激光雷达的标定方法，其特征在于，所述方法包括：
    获取车辆通过目标区域时激光雷达采集的点云；所述目标区域的至少一边竖直设置有标志物；
    提取所采集的点云中的标志物点云；
    根据标志物点云，得到标志物的拟合线信息；所述拟合线信息包括拟合线的位置及方向信息；
    根据所述拟合线信息，得到所述激光雷达外参的数值。
  10. 根据权利要求9所述的方法，其特征在于，所述外参包括俯仰角；
    所述方法还包括：
    在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，对所述激光雷达的外参进行标定，其中，所述第二预设阈值由所述激光雷达的竖直视场角确定。
  11. 根据权利要求10所述的方法，其特征在于，所述外参包括偏航角；
    所述方法还包括：
    获取所述车辆的位置信息及所述激光雷达的位置信息；
    根据所述车辆的位置信息及所述激光雷达的位置信息，确定所述车辆的航向角；
    根据所述航向角，优化标定后的所述激光雷达的偏航角。
  12. 根据权利要求10或11所述的方法，其特征在于，所述车辆安装有主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；
    所述方法还包括：
    根据主激光雷达的标定后的外参及所述主激光雷达采集的多个标志物点云，确定所述多个标志物的位置信息；
    根据所述多个标志物的位置信息，得到第一标志物的预测位置；
    根据从激光雷达的标定后的外参及所述从激光雷达采集的所述第一标志物点云，得到所述第一标志物的测量位置；
    通过所述预测位置与所述测量位置，优化所述从激光雷达的外参。
  13. 根据权利要求10-12中任意一项所述的方法，其特征在于，所述方法还包括：提取所采集的点云中的地面点云；
    所述根据标志物点云，对所述激光雷达的外参进行标定，还包括：
    根据标志物点云和所述地面点云，对所述激光雷达的外参进行标定。
  14. 根据权利要求9-13中任意一项所述的方法，其特征在于，所述根据标志物点云，得到标志物的拟合线信息，包括：
    根据标志物点云，确定旋转角初始值，所述旋转角初始值使标志物点云旋转后，在所述激光雷达坐标系中水平平面的投影区域最小；
    根据所述旋转角初始值，对标志物点云进行旋转处理；
    利用所述旋转后的标志物点云，得到标志物的拟合线信息。
  15. 根据权利要求10-14中任意一项所述的方法，其特征在于，所述方法还包括：
    在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的 数值不大于所述第二预设阈值的情况下，根据标志物点云及地面点云，对所述激光雷达的外参进行标定。
  16. 根据权利要求9-15中任意一项所述的方法，其特征在于，所述外参包括俯仰角、翻滚角、偏航角中的至少一项，所述方法还包括：
    在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值不大于所述第二预设阈值的情况下，根据所述地面点云，标定所述激光雷达的俯仰角和翻滚角；根据拟合线的位置信息，标定所述激光雷达的偏航角。
  17. 根据权利要求9-16中任意一项所述的方法，其特征在于，所述目标区域的至少一边竖直设置有多个标志物，且所述多个标志物与地面的交点在同一直线上。
  18. 一种激光雷达的标定装置，其特征在于，所述装置包括：
    获取模块，用于获取车辆通过目标道路时激光雷达采集的点云，所述目标道路的至少一边设置有标志物；
    筛选模块，用于根据预设阈值，对所采集的点云进行初步筛选；所述预设阈值由所述激光雷达的安装高度确定；
    第一提取模块，用于对所述初步筛选出的点云进行多次拟合处理，得到地面点云；
    第二提取模块，用于提取所采集的点云中的标志物点云；
    标定模块，用于根据所述标志物点云及所述地面点云，对所述激光雷达的外参进行标定。
  19. 根据权利要求18所述的装置，其特征在于，所述装置还包括：
    转换模块，用于根据多个激光雷达的标定后的外参，得到各激光雷达对应的标志物点云的位置信息及地面点云的位置信息；
    第三提取模块，用于根据所述各激光雷达对应的标志物点云的位置信息及地面点云的位置信息，得到交叉特征点或交叉域，其中，所述交叉域表示平行于所述车辆行进的方向，以所述交叉特征点为中心的区域；
    优化模块，用于根据所述交叉特征点或交叉域，对所述多个激光雷达中任一激光雷达的标定后外参进行优化。
  20. 一种激光雷达的标定装置，其特征在于，所述装置包括：
    获取模块，用于获取车辆通过目标区域时激光雷达采集的点云；所述目标区域的至少一边竖直设置有标志物；
    提取模块，用于提取所采集的点云中的标志物点云；
    拟合模块，用于根据标志物点云，得到标志物的拟合线信息；所述拟合线信息包括拟合线的位置及方向信息；
    计算模块，用于根据所述拟合线信息，得到所述激光雷达外参的数值。
  21. 根据权利要求20所述的装置，其特征在于，所述外参包括俯仰角；
    所述装置还包括：标定模块，用于在所述激光雷达朝向与竖直向上方向的夹角小于第一预设阈值，且所述俯仰角的数值大于第二预设阈值的情况下，根据标志物点云，对所述激光雷达的外参进行标定，其中，所述第二预设阈值由所述激光雷达的竖直视场角确定。
  22. 根据权利要求21所述的装置，其特征在于，所述外参包括偏航角；所述装置还包括：优化模块，用于获取所述车辆的位置信息及所述激光雷达的位置信息；根据所述车辆的位置信息及所述激光雷达的位置信息，确定所述车辆的航向角；根据所述航向角，优化标定后的所述激光雷达的偏航角。
  23. 根据权利要求21或22所述的装置，其特征在于，所述车辆安装有主激光雷达和从激光雷达，其中，所述主激光雷达用于扫描所述车辆的前方环境，所述从激光雷达用于扫描所述车辆的侧方和/或后方环境；所述装置还包括：
    确定模块，用于根据主激光雷达的标定后的外参及所述主激光雷达采集的多个标志物点云，确定所述多个标志物的位置信息；
    预测模块，用于根据所述多个标志物的位置信息，得到第一标志物的预测位置；
    测量模块，用于根据从激光雷达的标定后的外参及所述从激光雷达采集的所述第一标志物点云，得到所述第一标志物的测量位置；
    匹配模块，用于通过所述预测位置与所述测量位置，优化所述从激光雷达的外参。
  24. 一种激光雷达的标定装置，其特征在于，包括：
    处理器；
    用于存储处理器可执行指令的存储器；
    其中，所述处理器被配置为执行所述指令时实现权利要求1-8中任意一项所述的方法，或者实现权利要求9-17中任意一项所述的方法。
  25. 一种非易失性计算机可读存储介质，其上存储有计算机程序指令，其特征在于，所述计算机程序指令被处理器执行时实现权利要求1-8中任意一项所述的方法，或者实现权利要求9-17中任意一项所述的方法。
  26. 一种包含指令的计算机程序产品，其特征在于，当其在计算机上运行时，使得所述计算机执行如权利要求1-8中任意一项所述的方法，或者实现权利要求9-17中任意一项所述的方法。
PCT/CN2021/115418 2021-08-30 2021-08-30 一种激光雷达的标定方法、装置及存储介质 WO2023028774A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202180006105.6A CN114829971A (zh) 2021-08-30 2021-08-30 一种激光雷达的标定方法、装置及存储介质
PCT/CN2021/115418 WO2023028774A1 (zh) 2021-08-30 2021-08-30 一种激光雷达的标定方法、装置及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/115418 WO2023028774A1 (zh) 2021-08-30 2021-08-30 一种激光雷达的标定方法、装置及存储介质

Publications (1)

Publication Number Publication Date
WO2023028774A1 true WO2023028774A1 (zh) 2023-03-09

Family

ID=82527470

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/115418 WO2023028774A1 (zh) 2021-08-30 2021-08-30 一种激光雷达的标定方法、装置及存储介质

Country Status (2)

Country Link
CN (1) CN114829971A (zh)
WO (1) WO2023028774A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116168176A (zh) * 2023-04-23 2023-05-26 深圳大学 联合InSAR与激光点云的建筑物几何与变形提取方法
CN116381632A (zh) * 2023-06-05 2023-07-04 南京隼眼电子科技有限公司 雷达横滚角的自标定方法、装置及存储介质
CN116400334A (zh) * 2023-06-01 2023-07-07 未来机器人（深圳）有限公司 激光外参的标定验证方法、装置、电子设备及可存储介质
CN117590362A (zh) * 2024-01-19 2024-02-23 深圳海星智驾科技有限公司 一种多激光雷达外参标定方法、装置和设备
CN118604837A (zh) * 2024-08-08 2024-09-06 浙江吉利控股集团有限公司 激光雷达点云数据增强方法、装置、设备、介质及产品

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115372946B (zh) * 2022-08-15 2024-09-17 北京城市网邻信息技术有限公司 激光雷达采集设备与旋转轴标定方法、装置及存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106896353A (zh) * 2017-03-21 2017-06-27 同济大学 一种基于三维激光雷达的无人车路口检测方法
CN109100741A (zh) * 2018-06-11 2018-12-28 长安大学 一种基于3d激光雷达及图像数据的目标检测方法
US20190056484A1 (en) * 2017-08-17 2019-02-21 Uber Technologies, Inc. Calibration for an autonomous vehicle lidar module
CN109696663A (zh) * 2019-02-21 2019-04-30 北京大学 一种车载三维激光雷达标定方法和系统
US20200401823A1 (en) * 2019-06-19 2020-12-24 DeepMap Inc. Lidar-based detection of traffic signs for navigation of autonomous vehicles
CN112241007A (zh) * 2020-07-01 2021-01-19 北京新能源汽车技术创新中心有限公司 自动驾驶环境感知传感器的标定方法、布置结构及车辆
CN112258590A (zh) * 2020-12-08 2021-01-22 杭州迦智科技有限公司 一种基于激光的深度相机外参标定方法、设备及其存储介质
CN112946591A (zh) * 2021-02-26 2021-06-11 商汤集团有限公司 外参标定方法、装置、电子设备及存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102195164B1 (ko) * 2019-04-29 2020-12-24 충북대학교 산학협력단 다중 라이다를 이용한 다중 물체 인식 시스템 및 방법
US11598876B2 (en) * 2019-06-14 2023-03-07 Nvidia Corporation Segmenting ground points from non-ground points to assist with localization of autonomous vehicles
CN113093155B (zh) * 2021-03-02 2022-12-23 上海新纪元机器人有限公司 一种激光雷达联合标定方法及系统
CN112946612B (zh) * 2021-03-29 2024-05-17 上海商汤临港智能科技有限公司 外参标定方法、装置、电子设备及存储介质

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106896353A (zh) * 2017-03-21 2017-06-27 同济大学 一种基于三维激光雷达的无人车路口检测方法
US20190056484A1 (en) * 2017-08-17 2019-02-21 Uber Technologies, Inc. Calibration for an autonomous vehicle lidar module
CN109100741A (zh) * 2018-06-11 2018-12-28 长安大学 一种基于3d激光雷达及图像数据的目标检测方法
CN109696663A (zh) * 2019-02-21 2019-04-30 北京大学 一种车载三维激光雷达标定方法和系统
US20200401823A1 (en) * 2019-06-19 2020-12-24 DeepMap Inc. Lidar-based detection of traffic signs for navigation of autonomous vehicles
CN112241007A (zh) * 2020-07-01 2021-01-19 北京新能源汽车技术创新中心有限公司 自动驾驶环境感知传感器的标定方法、布置结构及车辆
CN112258590A (zh) * 2020-12-08 2021-01-22 杭州迦智科技有限公司 一种基于激光的深度相机外参标定方法、设备及其存储介质
CN112946591A (zh) * 2021-02-26 2021-06-11 商汤集团有限公司 外参标定方法、装置、电子设备及存储介质

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116168176A (zh) * 2023-04-23 2023-05-26 深圳大学 联合InSAR与激光点云的建筑物几何与变形提取方法
CN116400334A (zh) * 2023-06-01 2023-07-07 未来机器人（深圳）有限公司 激光外参的标定验证方法、装置、电子设备及可存储介质
CN116400334B (zh) * 2023-06-01 2023-09-12 未来机器人（深圳）有限公司 激光外参的标定验证方法、装置、电子设备及可存储介质
CN116381632A (zh) * 2023-06-05 2023-07-04 南京隼眼电子科技有限公司 雷达横滚角的自标定方法、装置及存储介质
CN116381632B (zh) * 2023-06-05 2023-08-18 南京隼眼电子科技有限公司 雷达横滚角的自标定方法、装置及存储介质
CN117590362A (zh) * 2024-01-19 2024-02-23 深圳海星智驾科技有限公司 一种多激光雷达外参标定方法、装置和设备
CN117590362B (zh) * 2024-01-19 2024-04-16 深圳海星智驾科技有限公司 一种多激光雷达外参标定方法、装置和设备
CN118604837A (zh) * 2024-08-08 2024-09-06 浙江吉利控股集团有限公司 激光雷达点云数据增强方法、装置、设备、介质及产品

Also Published As

Publication number Publication date
CN114829971A (zh) 2022-07-29

Similar Documents

Publication Publication Date Title
WO2023028774A1 (zh) 一种激光雷达的标定方法、装置及存储介质
US11874119B2 (en) Traffic boundary mapping
US11353589B2 (en) Iterative closest point process based on lidar with integrated motion estimation for high definition maps
EP3759562B1 (en) Camera based localization for autonomous vehicles
US11340355B2 (en) Validation of global navigation satellite system location data with other sensor data
EP3620823B1 (en) Method and device for detecting precision of internal parameter of laser radar
CN110832474B (zh) 更新高清地图的方法
JP6714688B2 (ja) 正確な道路データベースを生成および更新するために道路データ物体を突き合わせるためのシステムおよび方法
CN107851125B (zh) 通过车辆和服务器数据库进行两步对象数据处理以生成、更新和传送精确道路特性数据库的系统和方法
Khatab et al. Vulnerable objects detection for autonomous driving: A review
US11151394B2 (en) Identifying dynamic objects in a point cloud
CN107850672B (zh) 用于精确车辆定位的系统和方法
WO2021003452A1 (en) Determination of lane connectivity at traffic intersections for high definition maps
CN111656135A (zh) 基于高清地图的定位优化
CN111391823A (zh) 一种用于自动泊车场景的多层地图制作方法
Marinelli et al. Mobile mapping systems and spatial data collection strategies assessment in the identification of horizontal alignment of highways
US20210221398A1 (en) Methods and systems for processing lidar sensor data
US20210223373A1 (en) Methods and systems for processing lidar sensor data
CN113093128A (zh) 用于标定毫米波雷达的方法、装置、电子设备及路侧设备
US20230041031A1 (en) Systems and methods for efficient vehicle extent estimation
CN113743171A (zh) 目标检测方法及装置
CN111353453B (zh) 用于车辆的障碍物检测方法和装置
US20230121226A1 (en) Determining weights of points of a point cloud based on geometric features
CN113587937A (zh) 车辆的定位方法、装置、电子设备和存储介质
CN117079458A (zh) 数据补偿模型的训练方法、车辆控制方法、装置及设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21955354

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21955354

Country of ref document: EP

Kind code of ref document: A1