CN111462321B - Point cloud map processing method, processing device, electronic device and vehicle - Google Patents
Point cloud map processing method, processing device, electronic device and vehicle Download PDFInfo
- Publication number
- CN111462321B CN111462321B CN202010227293.2A CN202010227293A CN111462321B CN 111462321 B CN111462321 B CN 111462321B CN 202010227293 A CN202010227293 A CN 202010227293A CN 111462321 B CN111462321 B CN 111462321B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- initial pose
- merging
- pose
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000012545 processing Methods 0.000 title claims abstract description 51
- 238000003672 processing method Methods 0.000 title claims abstract description 31
- 238000005457 optimization Methods 0.000 claims abstract description 37
- 238000000034 method Methods 0.000 claims description 22
- 230000000694 effects Effects 0.000 abstract description 8
- 238000005259 measurement Methods 0.000 description 13
- 238000004422 calculation algorithm Methods 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/74—Determining position or orientation of objects or cameras using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/08—Indexing scheme for image data processing or generation, in general involving all processing steps from image acquisition to 3D model generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/32—Indexing scheme for image data processing or generation, in general involving image mosaicing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Remote Sensing (AREA)
- Computer Graphics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Navigation (AREA)
- Traffic Control Systems (AREA)
Abstract
The application discloses a point cloud map processing method. The point cloud map processing method comprises the following steps: combining the point cloud data and the initial pose to obtain a combined point cloud, processing the combined point cloud to obtain a road marking point and a measuring point, and constructing a target equation according to the road marking point and the measuring point to optimize the initial pose. In the point cloud map processing method, the point cloud data and the initial pose are combined, and the road mark points are obtained after processing, so that finer optimization can be performed on a scene, the initial pose of each frame of point cloud is optimized based on the minimization of convergence errors, a target equation is constructed, the calculated amount and the memory consumption are reduced, and a better optimization effect can be obtained. The application also discloses a processing device, an electronic device and a vehicle.
Description
Technical Field
The application relates to the technical field of automatic driving, in particular to a point cloud map processing method, a processing device, an electronic device and a vehicle.
Background
With the development of automotive electronics, autopilot or unmanned technology is becoming a hotspot. The generation of the point cloud map of the vehicle running environment is a key in the unmanned technology, and the optimization of the point cloud map directly influences the splicing quality and precision of the map and finally influences the precision and effect of the point cloud positioning. In the related art, it is difficult to optimize the small error, or the calculation amount is large, which results in large memory consumption.
Disclosure of Invention
In view of this, embodiments of the present application provide a point cloud map processing method, a processing apparatus, an electronic apparatus, a vehicle, and a computer-readable storage medium.
The application provides a point cloud map processing method, which comprises the following steps:
combining point cloud data and an initial pose to obtain a combined point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
processing the merging point cloud to obtain a road marking point and a measuring point; and
and constructing a target equation according to the road mark points and the measuring points so as to optimize the initial pose.
In some embodiments, the processing the merged point cloud to obtain the landmark points and the measurement points includes:
downsampling the merging point cloud through a grid unit at a predetermined resolution;
when the number of points in the grid unit is larger than a preset threshold value, determining the central point of the grid unit as the road sign point; and
and searching the nearest point to the road mark point in the grid unit as the measuring point.
In some embodiments, the constructing a target equation from the landmark points and the measurement points to optimize the initial pose includes:
constructing the target equation according to the road marking point and the measuring point based on the beam method adjustment; and
and solving the target equation to obtain the optimized pose.
In some embodiments, the point cloud map processing method further includes:
judging whether the result after optimizing the initial pose meets the preset requirement or not; and
and carrying out iterative processing on the initial pose under the condition that the result does not meet the preset requirement.
In some embodiments, performing the iterative processing on the initial pose if the result does not meet the preset requirement includes:
under the condition that the result does not meet the preset requirement, updating the initial pose by adopting the result to obtain an updated initial pose;
the merging point cloud data and the initial pose to obtain a merging point cloud includes:
and merging the point cloud data and the updated initial pose to obtain the merged point cloud.
The application provides a processing device of a point cloud map, which comprises:
the merging module is used for merging the point cloud data and the initial pose to obtain merged point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
the processing module is used for processing the merging point cloud to obtain a road sign point and a measuring point;
and the optimization module is used for constructing a target equation according to the road mark points and the measurement points so as to optimize the initial pose.
The application provides an electronic device, comprising a processor for:
combining point cloud data and an initial pose to obtain a combined point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
processing the merging point cloud to obtain a road marking point and a measuring point; and
and constructing a target equation according to the road mark points and the measuring points so as to optimize the initial pose.
In certain embodiments, the processor is further configured to:
downsampling the merging point cloud through a grid unit at a predetermined resolution;
when the number of points in the grid unit is larger than a preset threshold value, determining the central point of the grid unit as the road sign point; and
and searching the nearest point to the road mark point in the grid unit as the measuring point.
In certain embodiments, the processor is further configured to:
constructing the target equation according to the road marking point and the measuring point based on the beam method adjustment; and
and solving the target equation to obtain the optimized pose.
In certain embodiments, the processor is further configured to:
judging whether the result after optimizing the initial pose meets the preset requirement or not; and
and carrying out iterative processing on the initial pose under the condition that the result does not meet the preset requirement.
In certain embodiments, the processor is further configured to:
under the condition that the result does not meet the preset requirement, updating the initial pose by adopting the result to obtain an updated initial pose; and
and merging the point cloud data and the updated initial pose to obtain the merged point cloud.
The application provides a vehicle, comprising one or more processors and a memory; and one or more programs, wherein the one or more programs are stored in the memory and executed by the one or more processors, the programs comprising instructions for performing the point cloud map processing method as described above.
The present application provides a non-transitory computer-readable storage medium containing computer-executable instructions that, when executed by one or more processors, cause the processors to perform the point cloud map processing method.
In the point cloud map processing method, the processing device, the electronic device, the vehicle and the computer readable storage medium, the road mark points are obtained after processing by combining the point cloud data and the initial pose, so that finer optimization can be performed on a scene, the initial pose of each frame of point cloud is optimized based on minimization of convergence errors, a target equation is constructed, the calculated amount and the memory consumption are reduced, and a better optimization effect can be obtained.
Drawings
The foregoing and/or additional aspects and advantages of the application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
fig. 1 is a flow chart of a point cloud map processing method according to some embodiments of the present application.
Fig. 2 is a block diagram of a point cloud map processing apparatus according to some embodiments of the present application.
Fig. 3 is a flow chart of a point cloud map processing method according to some embodiments of the present application.
Fig. 4 is a flow chart of a point cloud map processing method according to some embodiments of the present application.
Fig. 5 is a flow chart of a point cloud map processing method according to some embodiments of the present application.
Fig. 6 is a flow chart of a point cloud map processing method according to some embodiments of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present application and should not be construed as limiting the application.
Referring to fig. 1, the present application provides a point cloud map processing method. Comprising the following steps:
s10: merging the point cloud data and the initial pose to obtain a merging point cloud;
s20: processing the merging point cloud to obtain a road marking point and a measuring point; and
s30: and constructing a target equation according to the road mark points and the measuring points to optimize the initial pose.
Referring to fig. 2, an embodiment of the application provides an electronic device. The electronic device includes a processor. The processor is used for merging the point cloud data and the initial pose to obtain a merging point cloud, processing the merging point cloud to obtain a road marking point and a measuring point, and constructing a target equation according to the road marking point and the measuring point to optimize the initial pose. The electronic device may be a driving computer or a computing device for processing data in the cloud.
The embodiment of the application also provides a processing device 110, and the point cloud map processing method of the embodiment of the application can be realized by the processing device 110 of the embodiment of the application.
Specifically, the processing device 110 includes a merge module 112, a process module 114, and an optimization module 116. S10 may be implemented by the merge module 112, S20 may be implemented by the process module 114, and S30 may be implemented by the optimize module 116. Alternatively, the merging module 112 is configured to merge the point cloud data and the initial pose to obtain a merged point cloud. The processing module 114 is configured to process the merging point cloud to obtain the road sign point and the measurement point. The optimization module 116 is configured to construct a target equation from the landmark points and the measurement points to optimize the initial pose.
According to the point cloud map processing method, the processing device and the electronic device, the point cloud data and the initial pose are combined, and the road mark points are obtained after processing, so that finer optimization can be performed on a scene, the initial pose of each frame of point cloud is optimized by constructing a target equation based on minimization of convergence errors, the calculated amount and the memory consumption are reduced, and a better optimization effect can be obtained.
Specifically, the point cloud data generally refers to information of points on the surface of an object measured by a 3D scanning device, such as a laser radar, a stereo camera, a TOF camera, or the like, where each point contains three-dimensional coordinates, and may contain color information, reflected intensity information, or the like. The point cloud data are collected by the scanning device and output in the form of relevant data, so that the point cloud data can be read by all post-processing devices.
In this embodiment, the point cloud data refers to points where object information acquired by the lidar presents a series of scattered points with accurate angle and distance information.
With the development of automotive electronics, related technologies such as sensing technology, image processing, artificial intelligence, etc. are gradually applied in the automotive field, and automatic driving or unmanned driving technologies are gradually becoming hot spots. By controlling the vehicle, the vehicle can automatically and safely run according to a preset running path, and automatic driving is realized. Generating a point cloud map of the vehicle's driving environment is a key in the automated driving technique. In the related art, data is generally collected by a vehicle-mounted laser radar (Light Detection And Ranging, liDAR), and a point cloud map is generated in combination with GPS data.
In LiDAR SLAM (Simultaneous localization and mapping, synchronous positioning and mapping), five modules are usually included, such as sensor data, odometer, back end, mapping and loop-back detection.
The sensor data are mainly used for collecting various types of original data in an actual environment, such as point cloud data and the like. The odometer is mainly used for estimating the relative position of a moving target at different moments. The rear end is mainly used for optimizing accumulated errors brought by the odometer. The map is used for three-dimensional map construction. Loop detection is mainly used for eliminating space accumulated errors.
It can be appreciated that the backend needs to continuously update and optimize the point cloud map, so as to obtain a better point cloud global map. The existence of a point cloud global map is a key for loop detection, and aims to overcome the accumulated drift of a vehicle in real-time positioning. Or, the optimization of the point cloud map is a key in automatic driving based on LiDAR, and the quality of the optimization effect directly influences the splicing quality and precision of the point cloud map, and finally influences the precision and effect of the point cloud positioning.
In the related art, poseGraph is a commonly used point cloud map optimization algorithm, the basic principle is to construct a map by taking the pose of each frame of point cloud as a vertex and the error of the pose as an edge, and the optimization aim is to minimize the sum of squares of the cumulative lengths of the edges in the map. However, the PoseGraph algorithm is generally used for optimization of a wide range of scenes, and is mainly oriented to large errors, and for small errors, the PoseGraph algorithm is difficult to perform relatively fine optimization.
For another point cloud map optimization algorithm S-GICP, the GICP algorithm for two-frame point cloud splicing is expanded, so that multi-frame point cloud can be optimized at one time, however, the calculated amount of the algorithm is large, and the memory consumption of processing equipment is large. For example, for a local map with N frames of point clouds, averaging N points per frame of point cloud, the number of edges that need to be built using this algorithm is (N-1) +|! X N.
In this embodiment, the pose of the laser point cloud data stored in a single frame form is optimized, wherein the vehicle-mounted sensor comprises a vehicle-mounted laser radar and an odometer, the point cloud data is obtained through the laser radar, and the initial pose is obtained through the odometer. Converting the initial pose into point cloud data through coordinate system conversion, and combining the point cloud data with the point cloud data obtained through a laser radar to obtain a combined point cloud, wherein the combined point cloud can be represented by the following formula:
wherein,,is the j-th point of the i-th frame point cloud, T i And the pose of the point cloud of the ith frame.
The combined point clouds are processed in a predetermined manner, such as downsampling, to obtain road marking points and measuring points. Furthermore, a target equation is constructed according to the road mark points and the measuring points, and it can be understood that the light rays from different frame images, which are directed to the same 3D point, are converged at one point, and based on the target equation, the optimization target is to minimize the convergence error, so that the optimization of the initial pose is realized.
Compared with the PoseGraph algorithm, the processing method of the point cloud map provided by the embodiment of the application has the advantages that the road mark points are arranged, and the pose of each frame of point cloud is optimized through the road mark points, so that a finer optimization effect can be obtained.
Taking N frames of point clouds and taking point cloud map data with N points in each frame of point clouds as an example, the S-GICP algorithm needs to store N frames of point clouds simultaneously, the space consumption is O (N), the number of edges to be established is (N-1) +.! XN, time spent O ((N-1) |XN). In this embodiment, only one frame of point cloud needs to be stored at the same time, so that the space consumption is O (1), the time consumption is O (n×n), the maximum number of edges in the optimization equation is n×n, the calculated amount and the memory consumption are obviously reduced, and a better optimization effect can be obtained.
Referring to fig. 3, in some embodiments, S20 includes:
s21: downsampling the merging point cloud through a grid unit at a predetermined resolution;
s22: when the number of points in the grid cells is larger than a preset threshold value, determining the central point of the grid cells as a road sign point; and
s23: and searching the point closest to the road mark point in the grid unit as a measuring point.
In some embodiments, S21-S23 may be implemented by the processing module 114. That is, the processing module 114 is configured to perform downsampling processing on the merged point cloud through the grid unit with a predetermined resolution, and determine a center point of the grid unit as a road sign point and find a point closest to the road sign point in the grid unit as a measurement point when the number of points in the grid unit is greater than a predetermined threshold.
In some embodiments, the processor is configured to downsample the merged point cloud through the grid cell at a predetermined resolution, and to determine a center point of the grid cell as a landmark point and to find a point of the grid cell closest to the landmark point as a measurement point when the number of points in the grid cell is greater than a predetermined threshold.
Specifically, the merging point cloud obtained after merging is subjected to downsampling processing by utilizing a grid according to a preset resolution. To achieve better processing results, points in the grid cells that are less than a predetermined threshold need to be filtered out during the downsampling of the combined point cloud. When the number of points in the grid cell is greater than a predetermined threshold, then the center point of the grid cell is determined to be the landmark point.
It should be noted that the predetermined threshold may be set according to the calculation requirement, and in some examples, the predetermined threshold may be set directly, for example, may be 2, 3, or the like, which is not limited herein. In other examples, to achieve better denoising, the number of points in all grid cells may be counted using a histogram, taking the number of points in the previous predetermined proportion of grid cells as a predetermined threshold, e.g., the number of points in the previous 10% grid cells.
After the landmark points are determined, counting the center points of the middle distance grids in the grid units, namely the landmark points, and determining the nearest point as a measuring point.
The process of downsampling and finding the measurement points can be expressed by the following relation:
wherein n is the number of points in the jth grid cell,to be a point falling in the grid cell, p j For the coordinates of the central point within the unit mesh, +.>The measurement point is the observation of the ith frame point cloud to the jth 3D point, or the coordinate of the jth 3D point in the ith frame point cloud. The search (·) is a search algorithm that searches for a distance from the center point p for each frame of point cloud that falls within the jth cell grid j The nearest measurement point.
Referring to fig. 4, in some embodiments, S30 includes:
s31: constructing a target equation according to the road mark points and the measuring points based on the beam method adjustment; and
s32: and solving a target equation to obtain the optimized pose.
In some embodiments, S31, S32 may be implemented by the optimization module 116, that is, the optimization module 116 is configured to construct a target equation from the road marking points and the measurement points based on the beam method adjustment, and to solve the target equation to obtain the optimized pose.
In some embodiments, the processor is configured to construct a target equation from the landmark points and the measurement points based on the beam method adjustment, and to solve the target equation to obtain the optimized pose.
Specifically, in computer vision, the beam method adjustment BA (Bundle Adjustment, BA) is often used for optimizing the pose of the camera and the reconstructed three-dimensional scene, and ideally, an object should correspond to the same point under different viewing angles, however, when the feature point matching is performed in the actual three-dimensional reconstruction process, an incorrect feature point matching pair is generated, and the function of the beam method is to eliminate the incorrect matching. The principle is that the projection error relation between the reconstruction point and the pixel point is constructed, so that the projection error is minimized. The target equation is as follows:
wherein T is i Is the pose of the i-th frame point cloud, p j Is the j-th 3D point, i.e. reconstruction point, z ij The method is that the ith frame image is observed on the jth 3D point, namely the pixel coordinate of the jth 3D point on the ith frame image; h (T, p) is a projection function, and pixel points are calculated according to the designated pose and reconstruction points. It will be appreciated that the BA algorithm in computer vision is based on the fact that light from different frames of images directed to the same 3D point should converge at one point, or that the light convergence error should be minimized.
In this embodiment, a target equation is constructed to optimize the point cloud map based on the optimization objective of minimizing the light convergence error. The objective equation was constructed as follows:
wherein T is i Is the pose of the ith frame of image, p j Is the j-th 3D point and,is the measurement of the ith frame point cloud data on the jth 3D, i.e. the coordinates of the jth 3D point in the ith frame point cloud.
When solving, the disturbance vector is delta zeta= [ delta p delta phi ] which can be calculated by multiplying the disturbance by the lie algebra] T For the same error term, namely the error term corresponding to the ith frame point cloud, the jacobian matrix is as follows:
for the jacobian matrix, the g2o or CeresSolver nonlinear optimization library can be conveniently adopted to carry out nonlinearityOptimizing to obtain the optimized pose T * 。
It will be appreciated that in the above formula, p j Is an unknown quantity. For small error level, such as point map optimization with a spliced average error smaller than 1 meter, combining all point clouds to be optimized, then performing downsampling processing on the combined point clouds through grid cells according to a preset resolution, wherein the obtained center downsampled point clouds of the grid cells are regarded as road mark points, namely { p } in the above formula j And p is j Each frame point in a certain radius range around is a measuring point, namely the above
In this way, the objective equation of the present embodiment is constructed based on light convergence, so as to add the road mark points in the optimization process, and the optimization is aimed at minimizing the convergence error, and optimizing the initial pose of each frame point cloud with this as the constraint.
Referring to fig. 5, in some embodiments, the processing method further includes:
s40: judging whether the result after optimizing the initial pose meets the preset requirement or not; and
s50: and carrying out iterative processing on the initial pose under the condition that the result does not meet the preset requirement.
In some embodiments, the processing device 110 further includes a determination module. S40 may be implemented by the determination module, and S50 may be implemented by the optimization module 116. Or, the judging module is configured to judge whether the result after optimizing the initial pose meets the preset requirement, and the optimizing module 116 is configured to perform iterative processing on the initial pose if the result does not meet the preset requirement.
In some embodiments, the processor is configured to determine whether the result after optimizing the initial pose meets a preset requirement, and to perform iterative processing on the initial pose if the result does not meet the preset requirement.
It can be appreciated that after the initial pose is optimized, it is necessary to compare the pre-optimization and post-optimization to determine whether to converge so as to confirm whether the optimized pose meets the requirement. For the solution of the constructed objective equation, the solution is actually a high-dimensional nonlinear optimization problem. The minimum value of the objective function may be gradually converged on by an iterative method.
It should be noted that, the determination of convergence is related to a specific implementation of the initial pose, for example, the implementation of the initial pose may use a rotation matrix, a quaternion, a lie algebra or an euler angle, and in addition, the determination of the road marking point is related to a predetermined threshold set in the determination of the road marking point, which is not limited herein.
Referring to fig. 6, in some embodiments, S50 further includes:
s51: and under the condition that the result does not meet the preset requirement, updating the initial pose by adopting the result to obtain the updated initial pose.
S10 comprises the following steps:
s11: and merging the point cloud data and the updated initial pose to obtain a merging point cloud.
In some embodiments, S51 may be implemented by the optimization module 116 and S11 may be implemented by the merge module 112. Alternatively, the optimizing module 116 is configured to update the initial pose with the result of the current calculation to obtain an updated initial pose, and the merging module 112 is configured to merge the point cloud data and the updated initial pose to obtain a merged point cloud if the result does not meet the preset requirement.
In some embodiments, the processor is configured to update the initial pose with the result to obtain an updated initial pose, and to merge the point cloud data and the updated initial pose to obtain a merged point cloud if the result does not meet the preset requirement.
Specifically, in the case where it is determined that the optimized initial pose is not converged, the optimized result is used to update the initial pose, that is, the initial pose at this time is no longer obtained by the odometer. And after updating the initial pose, repeating the steps, and optimizing the current initial pose again until the optimized result is converged. Of course, an upper limit of the number of iterations may be defined in consideration of the processing performance and the calculation time of the computing device. In the actual processing process, when the iteration number reaches the maximum iteration number, the iteration processing is not performed any more, and the pose result of the last optimization is adopted.
The embodiment of the application also provides a computer readable storage medium. One or more non-transitory computer-readable storage media containing computer-executable instructions that, when executed by one or more processors, cause the processors to perform the point cloud map processing method of any of the above embodiments.
The embodiment of the application also provides a vehicle. The vehicle includes a lidar, a memory, and one or more processors, and one or more programs are stored in the memory and configured to be executed by the one or more processors. The program includes a program for executing the point cloud map processing method according to any one of the above embodiments.
The processor may be used to provide computing and control capabilities to support the operation of the entire vehicle. The memory in the vehicle provides an environment for computer readable instructions in the memory to run.
The point cloud map processing method of the embodiment of the application is applied to the vehicle and can be used for providing a quicker and more accurate point cloud map optimization processing scheme for realizing automatic driving of the vehicle.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program to instruct the associated hardware, and the program may be stored in a non-transitory computer readable storage medium, which when executed may include the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), or the like.
The foregoing examples illustrate only a few embodiments of the application, which are described in detail and are not to be construed as limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (9)
1. The point cloud map processing method is characterized by comprising the following steps of:
combining point cloud data and an initial pose to obtain a combined point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
downsampling the merging point cloud through a grid unit at a predetermined resolution;
when the number of points in the grid unit is larger than a preset threshold value, determining the central point of the grid unit as a road mark point;
searching a point closest to the road mark point in the grid unit as a measuring point;
constructing a target equation according to the road marking point and the measuring point based on the beam method adjustment; and
and solving the target equation to obtain the optimized pose.
2. The point cloud map processing method according to claim 1, characterized in that the point cloud map processing method further comprises:
judging whether the result after optimizing the initial pose meets the preset requirement or not; and
and carrying out iterative processing on the initial pose under the condition that the result does not meet the preset requirement.
3. The point cloud map processing method according to claim 2, wherein performing iterative processing on the initial pose if the result does not satisfy the preset requirement includes:
under the condition that the result does not meet the preset requirement, updating the initial pose by adopting the result to obtain an updated initial pose;
the merging point cloud data and the initial pose to obtain a merging point cloud includes:
and merging the point cloud data and the updated initial pose to obtain the merged point cloud.
4. A processing device for a point cloud map, the processing device comprising:
the merging module is used for merging the point cloud data and the initial pose to obtain merged point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
the processing module is used for carrying out downsampling processing on the merging point clouds through grid cells at a preset resolution;
when the number of points in the grid unit is larger than a preset threshold value, determining the central point of the grid unit as a road mark point; and
searching a point closest to the road mark point in the grid unit as a measuring point;
the optimization module is used for constructing a target equation according to the road mark points and the measuring points based on the beam method adjustment; and
and solving the target equation to obtain the optimized pose.
5. An electronic device comprising a processor configured to:
combining point cloud data and an initial pose to obtain a combined point cloud, wherein the point cloud data and the initial pose are obtained through a vehicle-mounted sensor;
downsampling the merging point cloud through a grid unit at a predetermined resolution;
when the number of points in the grid unit is larger than a preset threshold value, determining the central point of the grid unit as a road mark point; and
searching a point closest to the road mark point in the grid unit as a measuring point;
constructing a target equation according to the road marking point and the measuring point based on the beam method adjustment; and
and solving the target equation to obtain the optimized pose.
6. The electronic device of claim 5, wherein the processor is further configured to:
judging whether the result after optimizing the initial pose meets the preset requirement or not; and
and carrying out iterative processing on the initial pose under the condition that the result does not meet the preset requirement.
7. The electronic device of claim 6, wherein the processor is further configured to:
under the condition that the result does not meet the preset requirement, updating the initial pose by adopting the result to obtain an updated initial pose; and
and merging the point cloud data and the updated initial pose to obtain the merged point cloud.
8. A vehicle, characterized by comprising:
one or more processors, memory; and
one or more programs, wherein the one or more programs are stored in the memory and executed by the one or more processors, the programs comprising instructions for performing the point cloud map processing method of any of claims 1-3.
9. A non-transitory computer-readable storage medium of computer-executable instructions, which when executed by one or more processors, cause the processors to perform the point cloud map processing method of any of claims 1-3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010227293.2A CN111462321B (en) | 2020-03-27 | 2020-03-27 | Point cloud map processing method, processing device, electronic device and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010227293.2A CN111462321B (en) | 2020-03-27 | 2020-03-27 | Point cloud map processing method, processing device, electronic device and vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111462321A CN111462321A (en) | 2020-07-28 |
| CN111462321B true CN111462321B (en) | 2023-08-29 |
Family
ID=71685717
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010227293.2A Active CN111462321B (en) | 2020-03-27 | 2020-03-27 | Point cloud map processing method, processing device, electronic device and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111462321B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113947636B (en) * | 2021-10-19 | 2024-04-26 | 中南大学 | Laser SLAM positioning system and method based on deep learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109932713A (en) * | 2019-03-04 | 2019-06-25 | 北京旷视科技有限公司 | Localization method, device, computer equipment, readable storage medium storing program for executing and robot |
| CN109974693A (en) * | 2019-01-31 | 2019-07-05 | 中国科学院深圳先进技术研究院 | Unmanned plane localization method, device, computer equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SG10201700299QA (en) * | 2017-01-13 | 2018-08-30 | Otsaw Digital Pte Ltd | Three-dimensional mapping of an environment |
-
2020
- 2020-03-27 CN CN202010227293.2A patent/CN111462321B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109974693A (en) * | 2019-01-31 | 2019-07-05 | 中国科学院深圳先进技术研究院 | Unmanned plane localization method, device, computer equipment and storage medium |
| CN109932713A (en) * | 2019-03-04 | 2019-06-25 | 北京旷视科技有限公司 | Localization method, device, computer equipment, readable storage medium storing program for executing and robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111462321A (en) | 2020-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070615B (en) | Multi-camera cooperation-based panoramic vision SLAM method | |
| CN107796397B (en) | Robot binocular vision positioning method and device and storage medium | |
| CN108871353B (en) | Road network map generation method, system, equipment and storage medium | |
| CN108319655B (en) | Method and device for generating grid map | |
| US11830216B2 (en) | Information processing apparatus, information processing method, and storage medium | |
| US8199977B2 (en) | System and method for extraction of features from a 3-D point cloud | |
| CN107167826B (en) | Vehicle longitudinal positioning system and method based on variable grid image feature detection in automatic driving | |
| CN111612728B (en) | 3D point cloud densification method and device based on binocular RGB image | |
| Muñoz-Bañón et al. | Targetless camera-LiDAR calibration in unstructured environments | |
| Chen et al. | Transforming a 3-d lidar point cloud into a 2-d dense depth map through a parameter self-adaptive framework | |
| CN114217665B (en) | Method and device for synchronizing time of camera and laser radar and storage medium | |
| CN109900274B (en) | Image matching method and system | |
| CN113989758B (en) | Anchor guide 3D target detection method and device for automatic driving | |
| CN114994635A (en) | Intelligent driving travelable area detection method and device | |
| El Bouazzaoui et al. | Enhancing RGB-D SLAM performances considering sensor specifications for indoor localization | |
| CN115097419A (en) | External parameter calibration method and device for laser radar IMU | |
| CN111462321B (en) | Point cloud map processing method, processing device, electronic device and vehicle | |
| CN110851978A (en) | Camera position optimization method based on visibility | |
| CN113074634B (en) | Rapid phase matching method, storage medium and three-dimensional measurement system | |
| KR20170037804A (en) | Robust visual odometry system and method to irregular illumination changes | |
| CN112712062A (en) | Monocular three-dimensional object detection method and device based on decoupling truncated object | |
| CN115239899B (en) | Pose map generation method, high-precision map generation method and device | |
| CN113658260B (en) | Robot pose calculation method, system, robot and storage medium | |
| CN115471819A (en) | Target detection positioning method and device for roadside scene, electronic equipment and storage medium | |
| An et al. | Tracking an RGB-D camera on mobile devices using an improved frame-to-frame pose estimation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |