CN110400490B

CN110400490B - Trajectory prediction method and apparatus

Info

Publication number: CN110400490B
Application number: CN201910729763.2A
Authority: CN
Inventors: 钱祥隽
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2019-08-08
Filing date: 2019-08-08
Publication date: 2022-02-25
Anticipated expiration: 2039-08-08
Also published as: CN110400490A

Abstract

The embodiment of the invention discloses a track prediction method and a track prediction device; the embodiment of the invention can obtain map information, running information of a target vehicle, a lane prediction model and a motion prediction model; determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle; predicting a target lane into which a target vehicle is to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model; predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model; and calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information. In the embodiment of the invention, the target lane of the vehicle and the motion information of the vehicle on the target lane are predicted through different prediction models, so that the track calculation is carried out. Therefore, the prediction accuracy of the track prediction can be improved by the scheme.

Description

Trajectory prediction method and apparatus

Technical Field

The invention relates to the field of computers, in particular to a track prediction method and a track prediction device.

Background

With the popularization of the internet of things, vehicles can plan routes and avoid collision accidents through the assistance of computer technology, and meanwhile, the driving efficiency is improved. For example, the vehicle-mounted computer may recognize a static obstacle around the vehicle, predict the traveling tracks of other vehicles around the vehicle, and determine whether the vehicle poses a potential security threat to the vehicle according to the traveling tracks of the vehicles around the vehicle.

However, the prediction accuracy of the current trajectory prediction method is low.

Disclosure of Invention

The embodiment of the invention provides a track prediction method and a track prediction device, which can improve the prediction precision of track prediction.

The embodiment of the invention provides a track prediction method, which comprises the following steps:

obtaining map information, running information of a target vehicle, a lane prediction model and a motion prediction model, wherein the lane prediction model and the motion prediction model are trained by training samples;

determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle;

predicting a target lane to which a target vehicle is about to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model;

predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model;

and calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information.

In some embodiments, the motion prediction model comprises a first motion prediction model, a second motion prediction model, the motion information on the target lane comprises target speed information and target distance information relative to the target lane at preset times, the target lane comprises a target lane centerline;

the predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting the motion prediction model comprises the following steps:

predicting target speed information of a target vehicle relative to the target lane at a preset moment by adopting a first motion prediction model according to the driving characteristics;

predicting target distance information of the target vehicle relative to the central line of the target lane at a preset moment by adopting a second motion prediction model according to the driving characteristics;

the calculating the motion trail of the target vehicle based on the driving information, the target lane and the motion information comprises:

and calculating the motion trail of the target vehicle based on the running information, the target lane, the target speed information and the target distance information.

In some embodiments, the travel information includes initial position information, initial speed information;

calculating a movement trajectory of a target vehicle based on the travel information, the target lane, the target speed information, and the target distance information, including:

determining target position information of the target vehicle at a preset moment according to the target lane and the target distance information;

and calculating the motion trail of the target vehicle based on the initial position information, the target position information, the initial speed information and the target speed information.

In some embodiments, the predicting, by using a lane prediction model, a target lane to which a target vehicle is about to enter in the associated lane according to the driving characteristics includes:

calculating the entrance probability of the target vehicle entering the associated lane at a preset moment by adopting a lane prediction model according to the driving characteristics;

and determining a target lane from the associated lanes according to the entrance probability.

In some embodiments, the lane prediction model includes a plurality of lane prediction submodels, and the calculating, using the lane prediction model, an entry probability that the target vehicle enters the associated lane at a preset time according to the driving characteristics includes:

adopting a lane prediction submodel to calculate the probability of an entrance sub of the target vehicle entering the associated lane at a preset time according to the driving characteristics;

and carrying out weighted summation on the driving sub-probabilities to obtain the driving probability that the target vehicle drives into the associated lane at the preset moment.

In some embodiments, obtaining the map information, the driving information of the target vehicle, the lane prediction model, and the motion prediction model further comprises:

acquiring training samples and an initial prediction model, wherein each training sample corresponds to a plurality of sample labels;

removing sample labels corresponding to the training samples to obtain lane training samples and motion training samples;

training an initial prediction model by adopting the lane training sample to obtain a lane prediction model;

and training an initial prediction model by adopting the motion training sample to obtain a motion prediction model.

In some embodiments, the sample labels include lane labels, distance labels, speed labels, and the motion training samples include distance training samples and speed training samples;

removing the sample labels corresponding to the training samples to obtain lane training samples and motion training samples, and the method comprises the following steps:

discarding the distance labels and the speed labels of the training samples to obtain lane training samples only retaining lane labels;

discarding the lane labels and the distance labels of the training samples to obtain speed training samples only retaining the speed labels;

and discarding the lane labels and the speed labels of the training samples to obtain the distance training samples only keeping the distance labels.

In some embodiments, the motion training samples comprise speed training samples, distance training samples, the speed training samples comprise speed training subsamples, the distance training samples comprise distance training subsamples, and the motion prediction model comprises a speed prediction model, a distance prediction model;

training an initial prediction model by adopting the motion training sample to obtain a motion prediction model, wherein the motion prediction model comprises the following steps:

training an initial prediction model by adopting the speed training subsample to obtain a speed prediction model;

and training the initial prediction model by adopting the distance training subsample to obtain a distance prediction model.

In some embodiments, determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle comprises:

determining the current lane of the target vehicle according to the map information and the driving information of the target vehicle;

performing topology analysis on the current lane of the target vehicle according to the map information to obtain an associated lane associated with the current lane;

and calculating the driving characteristics of the target vehicle relative to the associated lane based on the associated lane and the driving information of the target vehicle.

An embodiment of the present invention further provides a trajectory prediction apparatus, including:

the system comprises an acquisition unit, a processing unit and a control unit, wherein the acquisition unit is used for acquiring map information, running information of a target vehicle, a lane prediction model and a motion prediction model, and the lane prediction model and the motion prediction model are trained by training samples;

the association unit is used for determining an associated lane of the target vehicle and the driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle;

the lane unit is used for predicting a target lane into which a target vehicle is to drive in the associated lane according to the driving characteristics by adopting a lane prediction model;

the movement unit is used for predicting the movement information of the target vehicle on the target lane according to the driving characteristics by adopting a movement prediction model;

and the track unit is used for calculating the motion track of the target vehicle based on the running information, the target lane and the motion information.

The embodiment of the invention can obtain map information, running information of a target vehicle, a lane prediction model and a motion prediction model, wherein the lane prediction model and the motion prediction model are trained by training samples; determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle; predicting a target lane into which a target vehicle is to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model; predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model; and calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information.

In the embodiment of the invention, the target lane of the vehicle and the motion information of the vehicle on the target lane are predicted through different prediction models, so that the track calculation is carried out. Therefore, the prediction accuracy of the track prediction can be improved by the scheme.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1a is a schematic view of a scene of a trajectory prediction method according to an embodiment of the present invention;

FIG. 1b is a schematic flow chart of a trajectory prediction method according to an embodiment of the present invention;

fig. 1c is a schematic diagram of a map layer structure of a high-precision map provided in an embodiment of the present invention;

FIG. 1d is a schematic diagram of a target vehicle in relation to a current lane, according to an embodiment of the present invention;

FIG. 1e is a schematic structural diagram of a random forest model according to an embodiment of the present invention;

FIG. 1f is a schematic diagram illustrating a motion trajectory calculation according to an embodiment of the present invention;

FIG. 2 is a flow chart of a trajectory prediction method including a model prediction process according to an embodiment of the present invention;

FIG. 3a is a schematic diagram of a first structure of a trajectory prediction apparatus according to an embodiment of the present invention;

FIG. 3b is a schematic diagram of a second structure of the trajectory prediction device according to the embodiment of the present invention;

FIG. 3c is a schematic diagram of a third structure of a trajectory prediction apparatus according to an embodiment of the present invention;

FIG. 3d is a diagram illustrating a fourth structure of a trajectory prediction apparatus according to an embodiment of the present invention;

FIG. 3e is a schematic diagram of a fifth structure of the trajectory prediction apparatus according to the embodiment of the present invention;

fig. 4 is a schematic structural diagram of a network device according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The embodiment of the invention provides a track prediction method and a track prediction device.

The trajectory prediction device may be specifically integrated in an electronic device, and the electronic device may be a terminal, a server, or the like. The terminal can be an autopilot, a smart phone, a tablet Personal Computer, an intelligent Bluetooth device, a notebook Computer, a Personal Computer (PC) and other devices; the server may be a single server or a server cluster composed of a plurality of servers.

In some embodiments, the trajectory prediction apparatus may also be integrated into a plurality of electronic devices, for example, the trajectory prediction apparatus may be integrated into a plurality of servers, and the trajectory prediction method of the present invention is implemented by the plurality of servers.

For example, the electronic device may be an autopilot and mounted on a vehicle, and the autopilot may communicate with a network and acquire travel information of a surrounding vehicle through a sensor mounted on the vehicle. For example, referring to fig. 1a, the trajectory prediction device is integrated in an autopilot, and the autopilot mounted on the host vehicle can obtain the traveling information of other vehicles around the host vehicle by using the sensors of the host vehicle. For example, map information, a lane prediction model, a motion prediction model, and traveling information of other surrounding vehicles (for example, traveling speeds of other surrounding vehicles, types of vehicles, distances from the host vehicle, traveling directions, and the like) may be acquired through a network; then, the automatic driver can determine the current associated lanes of other surrounding vehicles and the driving characteristics of other surrounding vehicles relative to the associated lanes based on the map information and the driving information of other surrounding vehicles; predicting a target lane to which other surrounding vehicles are about to drive in the associated lane according to the driving characteristics by adopting a lane prediction model, and predicting the motion information of the other surrounding vehicles on the target lane according to the driving characteristics by adopting a motion prediction model; and finally, calculating the motion tracks of other surrounding vehicles based on the driving information, the target lane and the motion information.

The following are detailed below. The numbers in the following examples are not intended to limit the order of preference of the examples.

In this embodiment, a trajectory prediction method is provided, as shown in fig. 1b, a specific flow of the trajectory prediction method may be as follows:

101. and acquiring map information, running information of the target vehicle, a lane prediction model and a motion prediction model.

The map information may be image information in which spatial information such as roads, traffic conditions, and administrative areas in the real world is depicted, or may be customized map data of the virtual world. Map information may be used for ground traffic control, vehicle navigation, vehicle routing, and the like.

In some embodiments, the map information may include a high-precision map, which may contain a static high-precision map layer and a dynamic high-precision map layer.

The static high-precision map layer may include a road layer, a road component layer, a road attribute layer, and other map layers that include static information. In particular, the lane layer may include road detail information, such as lane lines, lane center lines, lane width, curvature, gradient, heading, lane rules, and the like. The road component layer may include road components such as traffic signs, pavement markers, etc., for example, recording the precise position and height of traffic lights, etc.

The dynamic high-precision map layer can comprise map layers containing dynamic traffic information, such as a road congestion layer, a construction condition layer, a traffic accident layer, a traffic control layer, an air layer and the like. For example, the construction situation layer may contain information such as refurbishment, wear and repainting of the road marking, changes in traffic signs, etc.

For example, as shown in fig. 1c, a map layer structure diagram of a high-precision map is provided, which includes a static high-precision map layer and a dynamic high-precision map layer, where the static high-precision map layer includes a road layer and a road component layer, and the dynamic high-precision map layer includes a weather layer.

In the embodiment of the present invention, the vehicle equipped with the trajectory prediction device may be referred to as a host vehicle, and the target vehicle may be another vehicle within a certain distance around the host vehicle, other than the host vehicle. The distance may be set by a user, may be set by a technician, or may be related to the vehicle sensor sensing distance.

The travel information of the target vehicle refers to information that can be detected by the host vehicle during travel of the target vehicle, such as a positioning position, a travel speed, a travel direction, a vehicle license plate, a vehicle type, and the like.

The lane prediction model may be a mathematical model for predicting a lane where the target vehicle is located at a future time (for example, after 3 seconds), and similarly, the motion prediction model is a mathematical model for predicting the motion information of the target vehicle at the future time.

The map information, the traveling information of the target vehicle, the lane prediction model, and the motion prediction model may be acquired in the same manner or in different manners.

For example, the map information, the lane prediction model, and the movement prediction model may be read from the local memory, and the driving information of the target vehicle may be acquired by the sensor system.

For another example, the lane prediction model and the motion prediction model may be acquired through a network, the map information may be read from the local memory, and the travel information of the target vehicle may be acquired through a sensor system, and so on.

102. And determining the associated lane of the target vehicle and the driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle.

Since the vehicle may keep the lane where the vehicle is located at the present time, may change lanes to the left, to the right, or even continuously change two lanes at a future time, it is necessary to screen out the lanes where the vehicle is likely to travel at the future time based on the map information and the travel information of the target vehicle and to note these lanes as the associated lanes before the trajectory prediction is performed.

The associated lane refers to a lane related to the lane in which the target vehicle is currently located. Specifically, the associated lane refers to a lane into which the target vehicle can enter from the current lane.

It is noted that the associated lane may be the lane in which the target vehicle is currently located.

In some embodiments, step 102 may specifically include the following steps:

(1) and determining the current lane of the target vehicle according to the map information and the running information of the target vehicle.

The driving information of the target vehicle is relative position relation information between the vehicle and the target vehicle, which is acquired by the vehicle through a sensor system, the accurate position of the target vehicle on the high-precision map can be deduced by combining the position of the vehicle on the high-precision map, and the current lane of the target vehicle on the high-precision map is determined according to the accurate position of the target vehicle on the high-precision map.

For example, in the present embodiment, referring to fig. 1d, a schematic diagram of the current lane where the target vehicle is located on the high-precision map is provided, and it is known that the position coordinate of the host vehicle on the high-precision map is (x is 0, y is 0), the distance between the target vehicle and the host vehicle is acquired as d is 3 meters, and the position angle between the target vehicle and the host vehicle is θ is 60 °, then the relative distance between the target vehicle and the host vehicle on the coordinate axis of the precision map can be calculated, that is, if the distance between the target vehicle and the host vehicle on the X axis is d × sin θ and the distance between the Y axis is d × cos θ, the position coordinates of the target vehicle on the high-precision map are (X is 1.5 and Y is-2.6), and it is understood that, the coordinates fall within the range of the area of lane 2 (1< x <2, -20< y <20), and it is determined that the current lane where the target vehicle is located is lane 2.

(2) And carrying out topology analysis on the current lane of the target vehicle according to the map information to obtain an associated lane associated with the current lane.

After the current lane where the target vehicle is located is determined, topology analysis can be performed on the current lane of the target vehicle to obtain an associated lane associated with the current lane, and specifically, by obtaining a topology relationship of the current lane where the target vehicle is located, according to the topology relationship, the associated lane associated with the current lane where the target vehicle is located can be obtained.

The topological relation may include relations such as adjacency, association, inclusion, and connection, and in some embodiments, the map information may include topological relations among all roads.

(3) And calculating the driving characteristics of the target vehicle relative to the associated lane based on the associated lane and the driving information of the target vehicle.

In the above steps, one or more associated lanes corresponding to each target vehicle may be obtained, and the driving characteristics of the target vehicle with respect to the associated lanes may be calculated through the driving information of the target vehicle.

The driving characteristics refer to physical characteristics of the vehicle when the vehicle is running, for example, the driving characteristics of the target vehicle relative to the associated lane may be a distance between the target vehicle and the associated lane, a relative speed between the target vehicle and the associated lane, a relative distance between the target vehicle and an obstacle on the associated lane, and the like.

For example, the driving characteristics of the target vehicle relative to the associated lane may include a relative speed between the target vehicle and an obstacle on the associated lane, where the obstacle may be a static object or a dynamic object on the associated lane, and the obstacle may be a traffic light, a motor vehicle, a non-motor vehicle, a green belt, or the like.

In particular, the driving characteristics of the target vehicle relative to the associated lane may include a footprint, size, type, motion state of an obstacle in the associated lane, and a relative speed, relative distance, etc. between the obstacle and the target vehicle in the associated lane.

103. And predicting a target lane into which the target vehicle is to enter in the associated lanes according to the driving characteristics by adopting a lane prediction model.

The lane prediction model may be a Neural network model, such as a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), a Recurrent Neural network (CNN), or the like.

For example, in some embodiments, the lane prediction model may be a deep neural network model based on a random forest algorithm.

The random forest algorithm is an algorithm for training and predicting a sample by using a plurality of decision trees. The random forest model may include a plurality of decision trees, for example, the random forest model shown in fig. 1e, which includes two decision trees, and the category output by the random forest model may be determined by the category output by the partial decision tree model, and the values are:

wherein, P₁(c | f) is the output of the decision tree on the left in FIG. 1e, P_n(c | f) is the output of the right decision tree in FIG. 1e, and P (c | f) is the output of the random forest model.

Specifically, in some embodiments, step 103 may include the steps of:

(1) and calculating the entrance probability of the target vehicle entering the associated lane at a preset moment by adopting a lane prediction model according to the driving characteristics.

For example, in some embodiments, the lane prediction model may be a deep neural network model based on a random forest algorithm in order to reduce overfitting, improve efficiency of processing high-dimensional features, and accommodate a large variety of prediction data. At this time, the lane prediction model is adopted to process the driving characteristics, so that a plurality of entering sub-probabilities that the target vehicle enters one associated lane at a preset time can be obtained, and the entering sub-probabilities are weighted and summed to obtain the entering probability that the target vehicle enters the associated lane at the preset time.

(2) And determining a target lane from the associated lanes according to the driving-in probability.

For example, in some embodiments, the lane prediction model may include a plurality of lane prediction submodels (e.g., a decision tree model), and using the lane prediction model to calculate the entry probability of the target vehicle entering the associated lane at the preset time according to the driving characteristics may specifically include the following steps:

(a) calculating the probability of the target vehicle entering the associated lane at a preset moment by adopting a lane prediction submodel according to the driving characteristics;

(b) and carrying out weighted summation on the driving sub-probabilities to obtain the driving probability of the target vehicle driving into the associated lane at the preset moment.

One or more target lanes may be determined from the associated lanes based on the target vehicle entry probability into the associated lanes.

The method for determining one or more target lanes from the associated lanes has various methods, for example, in some embodiments, the entrance probabilities are sorted from high to low, and the associated lanes corresponding to the entrance probabilities of the previous preset number are marked as target lanes; for example, the preset number is 3, the entrance probabilities are sorted from large to small, and the associated lanes corresponding to the first 3 entrance probabilities are marked as target lanes.

In some embodiments, the entrance probability may be compared with a preset probability range, and the associated lane corresponding to the entrance probability belonging to the preset probability range is marked as the target lane. For example, if the preset probability range is [0.8, 1], all the associated lanes corresponding to the entrance probability with the probability value greater than or equal to 0.8 are marked as target lanes.

The preset number and the preset probability range may be obtained by reading a local memory, may be obtained from a server through a network, may be set by a user, and the like.

104. The method is used for predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model.

The motion information may include information such as a driving speed, a driving direction, a relative distance, a positioning position, and the like on the target lane when the target vehicle is driving.

The motion prediction model may be of various types, for example, the motion prediction model may also be a neural network model, such as a convolutional neural network model, a deep neural network model, a cyclic neural network model, and so on.

Similarly, the motion prediction model may also be a deep neural network model based on a random forest algorithm.

In some embodiments, the motion prediction model includes a first motion prediction model, a second motion prediction model, the motion information includes target speed information and target distance information, the target lane may include a target lane center line, the motion information of the target vehicle on the target lane may refer to the target speed information and the target distance information of the target vehicle relative to the target lane at a preset time, and the step 104 may include the steps of:

(1) predicting target speed information of a target vehicle relative to a target lane at a preset moment by adopting a first motion prediction model according to the driving characteristics;

(2) and predicting target distance information of the target vehicle relative to the central line of the target lane at a preset moment by adopting a second motion prediction model according to the driving characteristics.

The target lane center line refers to a lane center line from the target lane. Specifically, the lane center line is a line formed by connecting center points between the edges of the lane subgrade in sequence from the start point to the end point of the lane. Lane center line information of each lane may be stored in the high-precision map.

105. And calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information.

In some embodiments, after obtaining the target speed information of the target vehicle relative to the target lane at the preset time and the target distance information of the target vehicle relative to the target lane at the preset time, step 105 may specifically be calculating the motion trajectory of the target vehicle based on the driving information, the target lane, the target speed information and the target distance information.

In some embodiments, the driving information of the target vehicle includes initial position information and initial speed information, where the initial position information may refer to a distance between the target vehicle and a lane center line of a lane where the target vehicle is currently located, and the initial speed information may refer to a current speed of the target vehicle, and step 105 may specifically include the following steps:

(1) determining target position information of the target vehicle at a preset moment according to the target lane and the target distance information;

(2) and calculating the motion trail of the target vehicle based on the initial position information, the target position information, the initial speed information and the target speed information.

There are various methods of calculating the movement trajectory of the target vehicle from the present to the future time. For example, trajectory calculation may be performed using a cubic polynomial curve, a quintic polynomial curve, an S-curve, a trapezoidal curve, or the like.

For example, with reference to fig. 1f, V1 is the current speed of the target vehicle, and V2 is the predicted target speed of the target vehicle at a future time (e.g., after 3 seconds); d1 is the distance between the current lane center line of the target vehicle and the current lane center line of the target vehicle, d2 is the distance between the center line of the target lane of the target vehicle and the future time of the target vehicle, and d is the distance between the current lane of the target vehicle and the target lane.

The lane center line can be obtained from a high-precision map, and the calculation formula of the distance D between the target vehicle and the future target vehicle is as follows:

D＝d1+d2+d

at this time, the movement locus of the target vehicle from the present to the future time can be calculated by the distance D and the current speed V1 of the target vehicle, and the target speed V2 of the target vehicle at the future time.

Specifically, the method for calculating the motion trajectory Q by using the fifth-order polynomial curve is as follows:

Q(t_start,t_end)＝q(t_start)+q(t_start+1)+q(t_start+2)+...q(t_end)＝∑q(t_i)

wherein q is a position point where the target vehicle is located at a certain moment; q is a line formed by connecting a plurality of position points, namely a motion track; t is t_startIs the initial time; t is t_endIs the future time; t is t_iThe sampling time between the initial time and the future time may be configured by a technician or may be set by a user.

TargetThe position point q (t) of the vehicle at the sampling time_i) This can be found by the following equation:

q(t_i)＝q(t_satrt)+w₁(t_i-t_strat)+w₂(t_i-t_strat)²+w₃(t_i-t_strat)³+w₄(t_i-t_strat)⁴+w₅(t_i-t_strat)⁵

wherein, w₁、w₂、w₃、w₄、w₅Is the coefficient of the fifth order polynomial.

The known physical formula is as follows, at t_iAt the moment, the speed v can be obtained by taking the first derivative from the position point q, and the motion acceleration a can be obtained by taking the second derivative from the position point q:

q′(t_i)＝v(t_i)，q″(t_i)＝a(t_i)

the coefficient w can be derived₁、w₂、w₃、w₄、w₅The calculation formula of (a) is as follows:

w₁＝v_start

wherein h is q (t)_i)-q(t_start)。

As can be seen from the above, the embodiment of the present invention can obtain map information, driving information of a target vehicle, a lane prediction model, and a motion prediction model, wherein the lane prediction model and the motion prediction model are trained from training samples; determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle; predicting a target lane into which a target vehicle is to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model; predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model; and calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information.

Therefore, the target lane of the vehicle and the motion information of the vehicle on the target lane can be predicted through different prediction models, and therefore the track is calculated. Therefore, the prediction accuracy of the track prediction can be improved by the scheme.

The method described in the above embodiments is further described in detail below.

The trajectory prediction scheme provided by the embodiment of the invention can be applied to various traffic scenes and used for automatic driving vehicle systems of various levels so as to realize the prediction of the motion trajectories of the automatic driving vehicles on surrounding vehicles.

In this embodiment, a lane prediction model and a motion prediction model are trained by a server, and a model obtained by an autonomous vehicle through the training of the server predicts a motion trajectory of a surrounding vehicle within 3 seconds, and a method according to an embodiment of the present invention is described in detail with reference to fig. 2, and the specific flow thereof is as follows:

201. the server obtains training samples and an initial prediction model, wherein each training sample corresponds to a plurality of sample labels.

For example, the server obtains the training sample through a network, the server directly reads the initial prediction model in a local memory of the server, the server obtains training data through the network, and labels the training data to obtain an available training sample, and the like.

The training data can be acquired by an acquisition vehicle equipped with sensors such as a laser radar, a camera, a high-precision inertial navigation system and the like and a corresponding perception algorithm. The collection vehicle may be a vehicle dedicated to training data collection, or a vehicle equipped with an automatic driving system.

For example, the collection vehicle may collect traffic conditions around the collection vehicle and record the location where the traffic conditions occur according to a high-precision map. For example, the capturing vehicle may capture a picture of obstacles around the vehicle, a lane where the vehicle is located, a location where the vehicle is located, a speed, a moving direction, an acceleration, an angular velocity, and a number of the obstacles, and the like. The obstacles may include vehicles, traffic lights, pedestrians, and other people and objects that may obstruct the vehicle from traveling.

For example, the collecting vehicle may collect the driving speed and the driving direction of other vehicles within 15 meters around within 3 seconds, and the starting position and the final position within the 3 seconds, so as to obtain the lane where other vehicles within 15 meters around are located at the starting time and the target lane where other vehicles are located after 3 seconds.

In some embodiments, the server may label the training data to obtain an available training sample, and the specific steps are as follows:

(1) the associated lane of the obstacle is determined, as well as the target lane.

The target lane is a lane where the obstacle is located at a future time (e.g., after 3 seconds), and the target lane information may be included in the training data.

The lane associated with the obstacle is a lane associated with the lane where the obstacle is currently located, and there are various specific ways of determining the lane associated with the obstacle, for example, as shown in the following steps:

a. determining the current lane of the obstacle according to the high-precision map and the positioning position of the obstacle;

b. and carrying out topology analysis on the current lane of the obstacle according to the high-precision map to obtain the associated lane of the obstacle.

The positioning position of the obstacle can be deduced by collecting the relative position relation between the vehicle and the obstacle and collecting the position of the vehicle on a high-precision map.

The topological relation can be obtained from a high-precision map, and the topological relation can comprise relations such as adjacency, association, inclusion and connection.

(2) And determining the driving characteristics of the obstacle for the associated lane according to the training data.

The driving characteristics of the obstacle for the associated lane may include a floor area, a size, a type, a motion state of the obstacle in the associated lane, and a relative speed, a relative distance, and the like between the obstacle and the target vehicle in the associated lane.

For example, referring to the driving feature list shown in table 1, the obstacle has a plurality of driving features for the associated lane:

TABLE 1

(3) And marking the training data according to the target lane.

For example, in some embodiments, each training sample may correspond to three sample labels, namely, a lane label, a distance label, a speed label, and the like.

In this embodiment, the associated lanes in the training data may be labeled according to the target lane. For example, when the target lane is the same as the associated lane, the lane label generating the training data corresponding to the associated lane is a positive sample label, and when the target lane is different from the associated lane, the lane label generating the training data corresponding to the associated lane is a negative sample label.

For example, if the lanes associated with the obstacle are lane a, lane B, and lane C, the feature sets [ a ], [ B ], [ C ] may be generated according to the lanes.

For example, if the lane where the obstacle is located after 3 seconds is lane B, the target lane of the obstacle is lane B, and it is known that the target lane of the obstacle is different from lane a, the same as lane B, and different from lane C, the feature set [ a ] is used as a negative sample and is labeled as [ a, 0 ]; taking the feature set [ B ] as a positive sample, and marking as [ B, 1 ]; the feature set [ C ] is taken as a negative example, labeled [ C, 0 ].

In some embodiments, the feature set may be further labeled according to the driving speed of the obstacle after 3 seconds, wherein the data labeled as the negative sample does not need to label the driving speed, for example, if the driving speed of the obstacle after 3 seconds is v, the negative sample [ a, 0] is labeled as [ a, 0, 0 ]; labeling the positive sample [ B, 1] as [ B, 1, v ]; the negative sample [ C, 0] is labeled [ C, 0, 0 ].

In some embodiments, the feature set may also be labeled according to the distance of the obstacle after 3 seconds from the target lane, wherein the data labeled as negative samples need not label the distance, e.g., the distance of the obstacle after 3 seconds from the target lane is D, then the negative sample [ a, 0, 0] is labeled as [ a, 0, 0, 0], the positive sample [ B, 1, v ] is labeled as [ B, 1, v, D ], and the negative sample [ C, 0, 0] is labeled as [ C, 0, 0, 0 ].

202. And the server eliminates the sample labels corresponding to the training samples to obtain lane training samples and motion training samples.

After the sample labels of different types corresponding to the training samples are removed, the training samples can retain part of the sample labels. For example, in some embodiments, the sample labels include lane labels, distance labels, and speed labels, the motion training samples include distance training samples and speed training samples, and the removing process of the sample labels corresponding to the training samples to obtain the lane training samples and the motion training samples may specifically include the following steps:

(a) discarding the distance labels and the speed labels of the training samples to obtain lane training samples only retaining lane labels;

(b) discarding the lane labels and the distance labels of the training samples to obtain speed training samples only retaining the speed labels;

(c) and discarding the lane labels and the speed labels of the training samples to obtain the distance training samples only keeping the distance labels.

For example, when the training samples are a negative sample [ a, 0, 0, 0], a positive sample [ B, 1, v, D ], and a negative sample [ C, 0, 0, 0], the first term of the training samples refers to the candidate lane, the second term refers to the sample type (e.g., positive sample type, negative sample type) of the training samples, the third term refers to the driving speed of the obstacle on the target lane, and the fourth term refers to the distance between the obstacle and the target lane.

For example, the step a is performed on the training samples, i.e. the speed labels of the three items and the distance label of the fourth item of the training samples can be discarded, and then the lane training samples [ a, 0], [ B, 1], [ C, 0] are obtained.

For example, step b is performed on the training samples, that is, the lane label of the first item and the distance label of the fourth item of the training samples can be discarded, so as to obtain the speed training samples [0, 0], [1, v ], [0, 0 ].

For example, step c is performed on the training samples, that is, the lane label of the first item and the speed label of the third item of the training samples can be discarded, so as to obtain distance training samples [0, 0], [1, D ], [0, 0 ].

203. And the server trains the initial prediction model by adopting the lane training sample to obtain a lane prediction model.

For example, the server trains the initial prediction model by using lane training samples [ A, 0], [ B, 1], [ C, 0] until convergence, so as to obtain the lane prediction model.

204. And the server trains the initial prediction model by adopting the motion training sample to obtain a motion prediction model.

In some embodiments, the motion training samples include a speed training sample and a distance training sample, the speed training sample includes a speed training positive sample, the distance training sample includes a distance training sub-sample, the motion prediction model includes a speed prediction model and a distance prediction model, and the training of the initial prediction model by using the motion training samples to obtain the motion prediction model may specifically include the following steps:

(a) training the initial prediction model by adopting a speed training subsample to obtain a speed prediction model;

(b) and training the initial prediction model by adopting a distance training subsample to obtain a distance prediction model.

For example, the first item in the velocity training samples [0, 0], [1, v ], [0, 0] is the sample type, where 0 represents a negative sample and 1 represents a positive sample, and the velocity training samples [0, 0], [1, v ], [0, 0] include two negative samples [0, 0] and one positive sample [1, v ].

205. The autonomous vehicle acquires map information, a lane prediction model and a motion prediction model from a server, and acquires driving information of a target vehicle through a sensor.

The autonomous vehicle may obtain the map information, the lane prediction model, and the motion prediction model from the server via a network, or may import the map information, the lane prediction model, and the motion prediction model from the server via a storage device.

206. The autonomous vehicle predicts a movement locus of the target vehicle using a lane prediction model and a movement prediction model based on the map information and the travel information of the target vehicle.

Step (six) may refer to

steps

102, 103, 104, and 105, which are not described herein again.

As can be seen from the above, in the embodiment of the present invention, the server obtains the lane training samples and the motion training samples by obtaining the training samples and the initial prediction model and removing the sample labels corresponding to the training samples, and the server may train the initial prediction model by using the lane training samples to obtain the lane prediction model and train the initial prediction model by using the motion training samples to obtain the motion prediction model. The automatic driving vehicle acquires map information, a lane prediction model and a motion prediction model from a server, and acquires driving information of a target vehicle through a sensor; the autonomous vehicle acquires map information, a lane prediction model and a motion prediction model from a server, and acquires driving information of a target vehicle through a sensor.

Therefore, in the embodiment of the invention, the target lane of the vehicle and the motion information of the vehicle on the target lane are predicted through different prediction models, so that the track calculation is carried out. Therefore, the method and the device can improve the efficiency of the track prediction and improve the prediction accuracy of the track prediction.

In order to better implement the above method, an embodiment of the present invention further provides a trajectory prediction apparatus, where the trajectory prediction apparatus may be specifically integrated in an electronic device, and the electronic device may be a terminal, a server, or other devices. The electronic device may be specifically an autopilot, a server, and the like.

For example, in the present embodiment, the method according to the embodiment of the present invention will be described in detail by taking the example that the trajectory prediction device is integrated in the server.

For example, as shown in fig. 3a, the trajectory prediction apparatus may include an acquisition unit 301, an association unit 302, a lane unit 303, a motion unit 304, and a trajectory unit 305 as follows:

the acquisition unit 301:

the obtaining unit 301 is configured to obtain map information, driving information of a target vehicle, a lane prediction model, and a motion prediction model, where the lane prediction model and the motion prediction model are trained by training samples.

Referring to fig. 3b, in some embodiments, the acquisition unit 301 may further include an acquisition subunit 3011, a culling subunit 3012, a lane model subunit 3013, and a motion model subunit 3014, as follows:

(1) acquisition subunit 3011:

the acquisition subunit is used for acquiring training samples and an initial prediction model, wherein each training sample corresponds to a plurality of sample labels;

(2) the rejecting subunit 3012:

the removing subunit is used for removing the sample labels corresponding to the training samples to obtain lane training samples and motion training samples;

(3) lane model subunit 3013:

the lane model subunit is used for training the initial prediction model by adopting a lane training sample to obtain a lane prediction model;

(4) motion model subunit 3014:

and the motion model subunit is used for training the initial prediction model by adopting the motion training sample to obtain a motion prediction model.

In some embodiments, the sample labels include lane labels, distance labels, and speed labels, the motion training samples include distance training samples and speed training samples, and the culling subunit 3012 may be specifically configured to:

In some embodiments, the motion training samples include speed training samples and distance training samples, the speed training samples include speed training positive samples, the distance training samples include distance training subsamples, the motion prediction model includes a speed prediction model and a distance prediction model, and the motion model subunit 3014 may be specifically configured to:

training the initial prediction model by adopting a speed training subsample to obtain a speed prediction model;

and training the initial prediction model by adopting a distance training subsample to obtain a distance prediction model.

(II) association unit 302:

an associating unit 302, configured to determine an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle.

In some embodiments, the association unit 302 may specifically be configured to:

determining a current lane of the target vehicle according to the map information and the driving information of the target vehicle;

Lane (iii) unit 303:

the lane unit 303 is configured to predict a target lane into which the target vehicle is to enter in the associated lane according to the driving characteristics by using a lane prediction model.

In some embodiments, referring to fig. 3c, the lane unit 303 may include a probability subunit 3031 and a lane subunit 3032, as follows:

(1) probability subunit 3031:

the probability subunit is used for calculating the entrance probability of the target vehicle entering the associated lane at the preset moment according to the driving characteristics by adopting a lane prediction model;

(2) lane subunit 3032:

and the lane subunit is used for determining a target lane from the associated lanes according to the entrance probability.

In some embodiments, the lane prediction model includes a plurality of lane prediction submodels, and the probability subunit 3031 may be specifically configured to:

calculating the probability of the target vehicle entering the associated lane at a preset moment by adopting a lane prediction submodel according to the driving characteristics;

and carrying out weighted summation on the driving sub-probabilities to obtain the driving probability of the target vehicle driving into the associated lane at the preset moment.

(iv) motion unit 304:

and the motion unit 304 is configured to predict motion information of the target vehicle on the target lane according to the driving characteristics by using a motion prediction model.

In some embodiments, the motion prediction model includes a first motion prediction model, a second motion prediction model, and the motion information includes object velocity information and object distance information, and referring to fig. 3d, the motion unit 304 may include a first motion subunit 3041 and a second motion subunit 3042 as follows:

(1) the first movement unit 3041:

the first motion unit is used for predicting target speed information of the target vehicle relative to the target lane at a preset moment according to the driving characteristics by adopting a first motion prediction model;

(2) second movement unit 3042:

and the second motion unit is used for predicting the target distance information of the target vehicle relative to the target lane at the preset moment according to the driving characteristics by adopting a second motion prediction model.

(V) track Unit 305:

a trajectory unit 305 for calculating a motion trajectory of the target vehicle based on the travel information, the target lane and the motion information.

In some embodiments, the trajectory unit 305 may be specifically configured to calculate a motion trajectory of the target vehicle based on the driving information, the target lane, the target speed information, and the target distance information.

In some embodiments, the travel information includes initial position information, initial speed information; referring to fig. 3e, the trajectory unit 305 may include a position subunit 3051 and a trajectory subunit 3052, as follows:

(1) position subunit 3051:

the position subunit is used for determining target position information of the target vehicle at a preset moment according to the target lane and the target distance information;

(2) trajectory subunit 3052:

and the track subunit is used for calculating the motion track of the target vehicle based on the initial position information, the target position information, the initial speed information and the target speed information.

In a specific implementation, the above units may be implemented as independent entities, or may be combined arbitrarily to be implemented as the same or several entities, and the specific implementation of the above units may refer to the foregoing method embodiments, which are not described herein again.

As can be seen from the above, the trajectory prediction device of the present embodiment acquires the map information, the travel information of the target vehicle, the lane prediction model, and the motion prediction model by the acquisition unit; determining, by the association unit, an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle; predicting a target lane to which a target vehicle is about to enter in the associated lanes according to the driving characteristics by using a lane prediction model by a lane unit; predicting the motion information of the target vehicle on the target lane by the motion unit according to the driving characteristics by adopting a motion prediction model; calculating, by a trajectory unit, a motion trajectory of the target vehicle based on the travel information, the target lane, and the motion information.

According to the scheme, the target lane of the vehicle and the motion information of the vehicle on the target lane can be predicted through different prediction models, so that the track calculation is carried out. Therefore, the prediction accuracy of the track prediction can be improved by the scheme.

The embodiment of the invention also provides electronic equipment which can be a smart phone, a smart watch, a tablet computer, a microcomputer, an automatic pilot, a server and the like. As shown in fig. 4, it shows a schematic structural diagram of an electronic device according to an embodiment of the present invention, specifically:

the electronic device may include a processor 401 of one or more processing cores, memory 402 of one or more computer-readable storage media, a power supply 403, an input unit 404, among other components, a sensor system 405, a positioning system 406, and so on. Those skilled in the art will appreciate that the electronic device configuration shown in fig. 4 does not constitute a limitation of the electronic device and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components. Wherein:

the processor 401 is a control center of the electronic device, connects various parts of the entire electronic device by various interfaces and lines, and performs various functions of the electronic device and processes data by driving or executing software programs and/or modules stored in the memory 402 and calling data stored in the memory 402, thereby performing overall monitoring of the electronic device. In some embodiments, processor 401 may include one or more processing cores; in some embodiments, processor 401 may integrate an application processor, which primarily handles operating systems, user interfaces, applications, etc., and a modem processor, which primarily handles wireless communications. It will be appreciated that the modem processor described above may not be integrated in the processor 401.

The memory 402 may be used to store software programs and modules, and the processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created according to use of the electronic device, and the like. Further, the memory 402 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 access to the memory 402.

The electronic device also includes a power supply 403 for supplying power to the various components, and in some embodiments, the power supply 403 may be logically coupled to the processor 401 via a power management system, such that the power management system may manage charging, discharging, and power consumption. The power supply 403 may also include any component of one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and the like.

The electronic device may further include an input unit 404, and the input unit 404 may be configured to receive input numbers, characters, images, location information, and the like, and generate signal inputs of keys, virtual keyboards, steering wheels, levers, sensors, and the like related to user settings and function control, for example, the input unit may receive images, geographical locations, driving information, and the like input by the sensor system and the location system.

The electronic device may also include a sensor system 405, and the sensor system 405 may include a variety of sensors, such as radar, cameras, infrared sensors, and the like. The structure of the sensor system 405 may be centralized, distributed, hierarchical, hybrid, multi-level, etc., where the various sensors may include components such as a sensing element, a conversion element, an auxiliary power supply, and a conversion circuit, and the sensing element may directly sense and measure and output a physical quantity signal having a certain relationship with the measured physical quantity signal; the conversion element converts the physical quantity signal output by the sensitive element into an electric signal; the conversion circuit is responsible for amplifying and modulating the electric signal output by the conversion element; the conversion element and the conversion circuit typically also require an auxiliary power supply.

The electronic device may further comprise a positioning system 406, which positioning system 406 may receive, track, transform and measure position signals, giving the position and velocity of the carrier in real time. The positioning system 406 may be composed of an antenna unit, a receiver host unit and a power supply, where the antenna unit may convert the acquired positioning navigation signal into a current, and amplify and frequency convert the signal current; the receiver unit can track, process and measure the amplified and frequency converted signal power.

Although not shown, the electronic device may further include a display unit, a communication unit, and the like, which are not described in detail herein. Specifically, in this embodiment, the processor 401 in the electronic device loads the executable file corresponding to the process of one or more application programs into the memory 402 according to the following instructions, and the processor 401 runs the application programs stored in the memory 402, thereby implementing various functions as follows:

predicting a target lane into which a target vehicle is to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model;

The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.

As can be seen from the above, in the embodiment of the present invention, the electronic device may obtain map information, driving information of the target vehicle, a lane prediction model, and a motion prediction model; determining an associated lane of the target vehicle and driving characteristics of the target vehicle relative to the associated lane based on the map information and the driving information of the target vehicle; predicting a target lane into which a target vehicle is to drive in the associated lanes according to the driving characteristics by adopting a lane prediction model; predicting the motion information of the target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model; and calculating the motion trail of the target vehicle based on the running information, the target lane and the motion information. Therefore, the scheme can predict the target lane of the vehicle and the motion information of the vehicle on the target lane through different prediction models, so as to calculate the track. Therefore, the prediction accuracy of the track prediction can be improved by the scheme.

It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.

To this end, the present invention provides a storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any one of the trajectory prediction methods provided by the embodiments of the present invention. For example, the instructions may perform the steps of:

Wherein the storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.

Since the instructions stored in the storage medium may execute the steps in any of the trajectory prediction methods provided in the embodiments of the present invention, the beneficial effects that can be achieved by any of the trajectory prediction methods provided in the embodiments of the present invention can be achieved, which are detailed in the foregoing embodiments and will not be described again here.

The above detailed description is provided for a trajectory prediction method and apparatus provided by the embodiments of the present invention, and the specific examples are applied herein to illustrate the principles and embodiments of the present invention, and the above description of the embodiments is only used to help understanding the method and its core ideas of the present invention; meanwhile, for those skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

Claims

1. A trajectory prediction method, comprising:

determining a current lane of the target vehicle based on the map information and the driving information of the target vehicle;

performing topology analysis on the current lane of the target vehicle based on the map information to obtain an associated lane associated with the current lane;

calculating the driving characteristics of the target vehicle relative to the associated lane based on the associated lane and the driving information of the target vehicle, wherein the driving characteristics comprise distance and relative speed;

predicting motion information of a target vehicle on the target lane according to the driving characteristics by adopting a motion prediction model, wherein the motion information on the target lane comprises target speed information and target distance information relative to the target lane at a preset moment;

and calculating the motion trail of the target vehicle based on the target lane, the motion information and the running information.

2. The trajectory prediction method of claim 1, wherein the motion prediction model comprises a first motion prediction model, a second motion prediction model, the target lane comprising a target lane centerline;

3. The trajectory prediction method according to claim 2, wherein the travel information includes initial position information, initial speed information;

4. The trajectory prediction method according to claim 1, wherein the predicting, using a lane prediction model, a target lane to which a target vehicle is about to enter in the associated lane according to the driving characteristics includes:

5. The trajectory prediction method according to claim 4, wherein the lane prediction model includes a plurality of lane prediction submodels, and the calculating, using the lane prediction model, the entry probability of the target vehicle entering the associated lane at a preset time according to the driving characteristics includes:

6. The trajectory prediction method according to claim 1, wherein the obtaining of the map information, the traveling information of the target vehicle, the lane prediction model, and the motion prediction model further comprises:

7. The trajectory prediction method of claim 6, wherein the sample labels comprise lane labels, distance labels, and speed labels, and the motion training samples comprise distance training samples and speed training samples;

8. The trajectory prediction method of claim 6, wherein the motion training samples comprise speed training samples, distance training samples, the speed training samples comprise speed training subsamples, the distance training samples comprise distance training subsamples, and the motion prediction model comprises a speed prediction model, a distance prediction model;

9. A trajectory prediction device, comprising:

the association unit is used for determining the current lane of the target vehicle based on the map information and the running information of the target vehicle; performing topology analysis on the current lane of the target vehicle based on the map information to obtain an associated lane associated with the current lane; calculating the driving characteristics of the target vehicle relative to the associated lane based on the associated lane and the driving information of the target vehicle, wherein the driving characteristics comprise distance and relative speed;

the movement unit is used for predicting movement information of a target vehicle on the target lane according to the driving characteristics by adopting a movement prediction model, wherein the movement information on the target lane comprises target speed information and target distance information relative to the target lane at a preset moment;