Vehicle trajectory prediction method and device, vehicle, storage medium and program product

Abstract

The invention relates to a vehicle trajectory prediction method and device, a vehicle, a storage medium and a program product, and relates to the technical field of trajectory prediction. The method comprises the steps of determining a target lane line sampling area based on driving speed information of a vehicle; and determining a target lane line point on a lane line in the target lane line sampling area so as to perform trajectory prediction on the vehicle based on the target lane line point. Therefore, the trajectory prediction efficiency can be improved, and the driving safety risk caused by low trajectory prediction efficiency can be reduced.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, in particular to the field of trajectory prediction technology, and specifically to a vehicle trajectory prediction method, device, vehicle, storage medium and program product. Background Art

[0002] In intelligent driving systems, trajectory prediction of autonomous vehicles is crucial to ensure vehicle safety, achieve smarter driving, and avoid collisions with other vehicles.

[0003] At present, common trajectory prediction methods usually rely on dynamic models to calculate the vehicle's driving trajectory within a certain period of time, or introduce lane line information to constrain the vehicle's trajectory to predict its driving path. However, due to the complex and changeable vehicle driving scenarios, trajectory prediction presents high uncertainty and complexity, resulting in low trajectory prediction efficiency and difficulty in predicting the vehicle's trajectory stably and quickly. Inefficient trajectory prediction may not be able to adjust the driving strategy in time, thereby increasing the risk of collision or traffic accidents. Therefore, how to improve the efficiency of vehicle trajectory prediction has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present application provides a vehicle trajectory prediction method, device, vehicle, storage medium and program product to at least solve the technical problem of how to improve the efficiency of vehicle trajectory prediction in the related art. The technical solution of the present application is as follows:

[0005] According to a first aspect of the present application, a vehicle trajectory prediction method is provided, the method comprising: determining a target lane line sampling area based on the vehicle's driving speed information; determining a target lane line point on the lane line within the target lane line sampling area, so as to predict the vehicle trajectory based on the target lane line point.

[0006] According to the above technical means, since the vehicle travels with different driving speed information, there are differences in the target lane line points that it may reach. If the target lane line points are determined on the lane lines within the fixed lane line sampling area, the efficiency of determining the target lane line points may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for a vehicle traveling at high speed, the fixed lane line sampling area may not cover the distant lane line points that the vehicle can actually reach, resulting in the vehicle determining the target lane line points based on less relevant lane line points, which not only increases the processing burden, but also may cause the predicted trajectory to be inaccurate. For vehicles traveling at low speeds, the fixed sampling area may introduce too many unnecessary or infeasible lane line points, increasing the complexity of processing. Therefore, by dynamically adjusting the target lane line sampling area based on the vehicle's driving speed information, it can better adapt to the current driving state of the vehicle, reduce the number of infeasible candidate lane line points on the lane lines within the target lane line sampling area, and thus when determining the target lane line points on the lane lines within the target lane line sampling area, the efficiency of determining the target lane line points can be improved, thereby improving the accuracy and efficiency of trajectory prediction for the vehicle based on the target lane line points.

[0007] In a possible implementation, a target lane line sampling area is determined based on the vehicle's driving speed information, including: determining a range of a first lane line sampling area in front of the vehicle based on the vehicle's driving speed information; determining a target lane line sampling area from the first lane line sampling area and the second lane line sampling area based on the range of the first lane line sampling area and the range of the second lane line sampling area in front of the vehicle; the range of the second lane line sampling area is determined based on a preset distance in front of the vehicle.

[0008] According to the above technical means, the range of the second lane line sampling area is determined based on the preset distance in front of the vehicle, that is, the second lane line sampling area is pre-set and relatively fixed, and the vehicle does not need to recalculate the second lane line sampling area every time it moves, thereby reducing the calculation burden. The range of the first lane line sampling area is determined based on the vehicle's driving speed information, that is, the first lane line sampling area is relatively dynamic and can be flexibly adapted according to the vehicle's driving speed to ensure that a high lane line detection accuracy can be maintained at various speeds. Through the combination of these two lane line sampling areas, a suitable sampling area can be selected based on the vehicle's driving speed information, which not only ensures the efficiency of determining the target lane line sampling area, but also improves the accuracy of determining the target lane line sampling area.

[0009] In another possible implementation, based on the range of the first lane line sampling area and the range of the second lane line sampling area, a target lane line sampling area is determined from the first lane line sampling area and the second lane line sampling area, including: when the range of the first lane line sampling area exceeds the range of the second lane line sampling area, the first lane line sampling area is determined as the target lane line sampling area; when the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, the second lane line sampling area is determined as the target lane line sampling area.

[0010] According to the above technical means, if the range of the first lane line sampling area exceeds the range of the second lane line sampling area, it can be said that the vehicle's driving speed is fast, and the distance between the target lane line point that the vehicle may reach and the current position may be large. The pre-set second lane line sampling area can cover the target lane line point that the vehicle may reach. Therefore, the second lane line sampling area can be determined as the target lane line sampling area. If the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, it can be said that the vehicle's driving speed is slow, and the distance between the target lane line point that the vehicle may reach and the current position may be small. If a fixed second lane line sampling area is used, too many unnecessary or infeasible candidate lane line points may be introduced, increasing the complexity of processing. Therefore, the first lane line sampling area dynamically determined based on the driving speed information can be determined as the target lane line sampling area, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0011] In another possible implementation, the driving speed information includes the current driving speed of the vehicle at the current position, and the range of the target lane line sampling area is positively correlated with the current driving speed; determining the target lane line point on the lane line within the target lane line sampling area includes: sampling on the lane line within the target lane line sampling area to obtain a candidate lane line point set; the candidate lane line point set includes at least one candidate lane line point; constraining the candidate lane line point set based on constraint information to obtain a candidate lane line point set that meets the constraint information; the constraint information includes: road network information and vehicle performance information; determining the target lane line point from the candidate lane line point set that meets the constraint information.

[0012] According to the above technical means, due to the positive correlation between the current driving speed included in the vehicle's driving speed information and the range of the target lane line sampling area, if the vehicle's driving speed is faster, the vehicle can reach a farther target lane line point, and the range of the target lane line sampling area will also be larger. However, a larger target lane line sampling area may cover too many unnecessary or infeasible candidate lane line points. Therefore, the candidate lane line point set is constrained by constraint information, and candidate lane line points that do not meet the constraint information are eliminated, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0013] In another possible implementation, a target lane line point is determined on a lane line within a target lane line sampling area, including: acquiring vehicle characteristics of the vehicle and map characteristics of the area where the vehicle is located; the vehicle characteristics are used to characterize road information around the vehicle and other vehicle information around the vehicle, and the map characteristics are used to characterize road information in the area where the vehicle is located; sampling is performed on the lane lines within the target lane line sampling area to obtain a candidate lane line point set; the candidate lane line point set includes at least one candidate lane line point; based on the interaction characteristics and the map characteristics, the confidence of each candidate lane line point in the candidate lane line point set is determined; based on the confidence of the candidate lane line point, the target lane line point is determined.

[0014] According to the above technical means, since the vehicle characteristics are used to represent the information of the vehicle itself, the road information around the vehicle detected by the sensors configured on the vehicle, and the information of other vehicles around the vehicle, it can reflect the complex interactive relationship between the vehicle and the roads around the vehicle and other vehicles. When determining the target lane line point based on the vehicle characteristics and map characteristics, it can better adapt to the complex real traffic scenes. Thereby improving the accuracy of determining the target lane line point.

[0015] In another possible implementation, the method also includes: determining target position information of the vehicle traveling to the target position based on the target lane line point; determining the target driving scenario of the vehicle based on the target position and the current position of the vehicle; determining a set of candidate predicted trajectories of the vehicle in the target driving scenario based on the current driving state information and current position information of the vehicle at the current position, the target position information, and the target driving state information of the vehicle traveling to the target position in the target driving scenario; the candidate predicted trajectory set includes at least one candidate predicted trajectory; constraining the candidate predicted trajectory set based on constraint information to obtain a set of candidate predicted trajectories that meet the constraint information; the constraint information includes road network information; and determining the target predicted trajectory from the set of candidate predicted trajectories that meet the constraint information.

[0016] According to the above technical means, the driving state of the vehicle from the current position to the target position may be different in different driving scenarios. For example, in the lane change scenario, the vehicle may not be located at the center line of the lane, and in the lane keeping scenario, the vehicle may be located at the center line of the lane, which results in different acceleration and speed of the vehicle at the target position. Therefore, when determining the candidate predicted trajectory, in addition to considering the current driving state information, current position information, and target position information, it is also necessary to combine the target driving state information of the vehicle driving to the target position in the target driving scenario. This method can adapt to different driving scenarios, improve the accuracy of the candidate predicted trajectory set, and thus improve the accuracy of the target trajectory prediction. In addition, since the constraint information includes traffic restriction information, under the constraint of the constraint information, unsafe trajectories that do not meet the constraint information can be excluded, so that the predicted trajectory is closer to the actual situation, thereby improving the accuracy and feasibility of the predicted trajectory.

[0017] In another possible implementation, the target driving scenario is determined based on a lateral driving scenario of the vehicle and a longitudinal driving scenario of the vehicle. The lateral driving scenario is used to indicate the relationship between the lane in which the vehicle is located at a current position and the lane in which the vehicle is located at a target position; the longitudinal driving scenario is used to indicate the change in the vehicle's driving speed from the current position to the target position.

[0018] According to the above technical means, since the movement of the vehicle can be decoupled into longitudinal movement along the center line of the road and lateral movement perpendicular to the center line of the road, when determining the target driving scene, the determination of the target driving scene can also be decoupled into the determination of the lateral driving scene and the determination of the longitudinal driving scene. This decoupling of the two-dimensional motion problem into two one-dimensional motion problems can reduce the complexity of trajectory prediction.

[0019] In another possible implementation, determining a target prediction trajectory from a set of candidate prediction trajectories that meet the constraint information includes: obtaining a target trajectory optimization function corresponding to a target driving scenario; and optimizing the candidate prediction trajectory set based on the target trajectory optimization function to obtain the target prediction trajectory.

[0020] According to the above technical means, since the vehicle trajectory may be different in different driving scenarios, if the optimization is only based on a fixed optimization target, the target prediction trajectory may be inaccurate. Therefore, by obtaining the target trajectory optimization function corresponding to the target driving scenario, the candidate prediction trajectory set can be optimized using the scenario-specific target trajectory optimization function to obtain a more accurate target prediction trajectory.

[0021] According to the second aspect provided by the present application, a vehicle trajectory prediction device is provided, which includes: a determination module, used to determine a target lane line sampling area based on the vehicle's driving speed information; determine a target lane line point on the lane line within the target lane line sampling area, so as to predict the vehicle trajectory based on the target lane line point.

[0022] According to the third aspect provided by the present application, a vehicle is provided, comprising: a processor; a memory for storing processor executable instructions; wherein the processor is configured to execute instructions to implement the method of the above-mentioned first aspect and any possible implementation manner thereof.

[0023] According to the fourth aspect provided by the present application, a computer-readable storage medium is provided. When the instructions in the computer-readable storage medium are executed by the processor of the vehicle, the vehicle is enabled to execute the method in the above-mentioned first aspect and any possible implementation method thereof.

[0024] According to the fifth aspect provided by the present application, a computer program product is provided, the computer program product comprising computer instructions, and when the computer instructions are executed on a vehicle, the vehicle executes the method of the above-mentioned first aspect and any possible implementation manner thereof.

[0025] Therefore, the above technical features of the present application have the following beneficial effects:

[0026] (1) Since vehicles travel at different speeds, the target lane points they may reach are different. If the target lane points are determined on the lanes within a fixed lane sampling area, the efficiency of determining the target lane points may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for vehicles traveling at high speeds, the fixed lane sampling area may not cover the distant lane points that the vehicle can actually reach, resulting in the vehicle determining the target lane points based on less relevant lane points, which not only increases the processing burden but also may lead to inaccurate predicted trajectories. For vehicles traveling at low speeds, the fixed sampling area may introduce too many unnecessary or infeasible lane points, increasing the complexity of processing. Therefore, by dynamically adjusting the target lane sampling area based on the vehicle's driving speed information, it can better adapt to the current driving state of the vehicle and reduce the number of infeasible candidate lane points on the lanes within the target lane sampling area. Therefore, when determining the target lane points on the lanes within the target lane sampling area, the efficiency of determining the target lane points can be improved, thereby improving the accuracy and efficiency of trajectory prediction for the vehicle based on the target lane points.

[0027] (2) The range of the second lane line sampling area is determined based on the preset distance in front of the vehicle, that is, the second lane line sampling area is pre-set and relatively fixed, and the vehicle does not need to recalculate the second lane line sampling area every time it moves, thereby reducing the calculation burden. The range of the first lane line sampling area is determined based on the vehicle's driving speed information, that is, the first lane line sampling area is relatively dynamic and can be flexibly adapted according to the vehicle's driving speed to ensure that a high lane line detection accuracy can be maintained at various speeds. Through the combination of these two lane line sampling areas, a suitable sampling area can be selected based on the vehicle's driving speed information, which not only ensures the efficiency of determining the target lane line sampling area, but also improves the accuracy of determining the target lane line sampling area.

[0028] (3) If the range of the first lane line sampling area exceeds the range of the second lane line sampling area, it can be said that the vehicle's driving speed is fast, and the distance between the target lane line point that the vehicle may reach and the current position may be large. The pre-set second lane line sampling area can cover the target lane line point that the vehicle may reach. Therefore, the second lane line sampling area can be determined as the target lane line sampling area. If the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, it can be said that the vehicle's driving speed is slow, and the distance between the target lane line point that the vehicle may reach and the current position may be small. If a fixed second lane line sampling area is used, too many unnecessary or infeasible candidate lane line points may be introduced, increasing the complexity of processing. Therefore, the first lane line sampling area dynamically determined based on the driving speed information can be determined as the target lane line sampling area, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0029] (4) Due to the positive correlation between the current driving speed included in the vehicle's driving speed information and the range of the target lane line sampling area, if the vehicle's driving speed is faster, the vehicle can reach a farther target lane line point, and the range of the target lane line sampling area will also be larger. However, a larger target lane line sampling area may cover too many unnecessary or infeasible candidate lane line points. Therefore, the candidate lane line point set is constrained by constraint information, and candidate lane line points that do not meet the constraint information are eliminated, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0030] (5) Since vehicle features are used to represent the information of the vehicle itself, the road information around the vehicle detected by the sensors configured on the vehicle, and the information of other vehicles around the vehicle, they can reflect the complex interactive relationship between the vehicle and the road around the vehicle and other vehicles. When determining the target lane line point based on vehicle features and map features, it can better adapt to complex real traffic scenes. Thereby improving the accuracy of determining the target lane line point.

[0031] (6) In different driving scenarios, the driving state of the vehicle from the current position to the target position may be different. For example, in the lane change scenario, the vehicle may not be located at the center line of the lane, and in the lane keeping scenario, the vehicle may be located at the center line of the lane, which results in different acceleration and speed of the vehicle at the target position. Therefore, when determining the candidate predicted trajectory, in addition to considering the current driving state information, current position information, and target position information, it is also necessary to combine the target driving state information of the vehicle traveling to the target position in the target driving scenario. This method can adapt to different driving scenarios, improve the accuracy of the candidate predicted trajectory set, and thus improve the accuracy of the target trajectory prediction. In addition, since the constraint information includes traffic restriction information, under the constraint of the constraint information, unsafe trajectories that do not meet the constraint information can be excluded, so that the predicted trajectory is closer to the actual situation, thereby improving the accuracy and feasibility of the predicted trajectory.

[0032] (7) Since the motion of the vehicle can be decoupled into longitudinal motion along the center line of the road and lateral motion perpendicular to the center line of the road, when determining the target driving scene, the determination of the target driving scene can also be decoupled into the determination of the lateral driving scene and the determination of the longitudinal driving scene. This decoupling of the two-dimensional motion problem into two one-dimensional motion problems can reduce the complexity of trajectory prediction.

[0033] (8) Since the vehicle trajectory may be different in different driving scenarios, if the optimization is only based on a fixed optimization target, the target prediction trajectory may be inaccurate. Therefore, by obtaining the target trajectory optimization function corresponding to the target driving scenario, the candidate prediction trajectory set can be optimized using the scenario-specific target trajectory optimization function, thereby obtaining a more accurate target prediction trajectory.

[0034] It should be noted that the technical effects brought about by any implementation method in the second to fifth aspects can refer to the technical effects brought about by the corresponding implementation method in the first aspect, and will not be repeated here.

[0035] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification are used to explain the principles of the present application, and do not constitute an improper limitation on the present application.

[0037] Figure 1 is a flow chart of a vehicle trajectory prediction method according to an exemplary embodiment;

[0038] Figure 2 is a schematic diagram showing a lane line sampling area according to an exemplary embodiment;

[0039] Figure 3 is a flowchart of a method for determining a candidate prediction trajectory set according to an exemplary embodiment;

[0040] Figure 4 is a flow chart of another vehicle trajectory prediction method according to an exemplary embodiment;

[0041] Figure 5 is a block diagram of a vehicle trajectory prediction device according to an exemplary embodiment;

[0042] Figure 6 is a block diagram of a vehicle according to an exemplary embodiment. DETAILED DESCRIPTION

[0043] In order to enable ordinary persons in the art to better understand the technical solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0044] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0045] In order to improve the efficiency of vehicle trajectory prediction, the present application proposes a vehicle trajectory prediction method. Since the vehicle travels with different driving speed information, there are differences in the target lane line points that it may reach. If the target lane line points are determined on the lane lines within a fixed lane line sampling area, the efficiency of determining the target lane line points may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for vehicles traveling at high speeds, the fixed lane line sampling area may not cover the distant lane line points that the vehicle can actually reach, causing the vehicle to determine the target lane line points based on less relevant lane line points, which not only increases the processing burden, but may also cause the predicted trajectory to be inaccurate. For vehicles traveling at low speeds, fixed sampling areas may introduce too many unnecessary or infeasible lane line points, increasing the complexity of processing.

[0046] Therefore, by dynamically adjusting the target lane line sampling area based on the vehicle's driving speed information, it can better adapt to the vehicle's current driving state and reduce the infeasible candidate lane line points on the lane lines within the target lane line sampling area. Thus, when determining the target lane line points on the lane lines within the target lane line sampling area, the determination efficiency of the target lane line points can be improved, thereby improving the accuracy and efficiency of vehicle trajectory prediction based on the target lane line points.

[0047] It should be understood that the vehicle trajectory prediction method provided in the embodiment of the present application can be applied to electronic devices such as personal computers (PCs), laptops, mobile devices, tablet computers, laptop computers, etc. The embodiment of the present application does not limit the specific form of the electronic device. Alternatively, the vehicle trajectory prediction method provided in the embodiment of the present application can also be applied to a processor on a vehicle. The embodiment of the present application does not impose any restrictions on this. For ease of understanding, the vehicle trajectory prediction method provided in the present application is specifically introduced below in conjunction with the accompanying drawings.

[0048] Figure 1 is a flow chart of a vehicle trajectory prediction method according to an exemplary embodiment. Figure 1 As shown, the vehicle trajectory prediction method includes the following steps:

[0049] S101. Determine a target lane line sampling area based on vehicle driving speed information.

[0050] It is understandable that since the vehicle travels at different driving speed information, there are differences in the target lane line points that it may reach. If the target lane line points are determined on the lane lines within the fixed lane line sampling area, the efficiency of determining the target lane line points may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for a vehicle traveling at high speed, the fixed lane line sampling area may not cover the distant lane line points that the vehicle can actually reach, causing the vehicle to determine the target lane line points based on less relevant candidate lane line points, which not only increases the processing burden, but also may cause the predicted trajectory to be inaccurate. For vehicles traveling at low speeds, the fixed sampling area may introduce too many unnecessary or infeasible candidate lane line points, increasing the complexity of processing. Therefore, the target lane line sampling area can be dynamically adjusted through the vehicle's driving speed information, which can better adapt to the current driving state of the vehicle, reduce infeasible candidate lane line points, thereby improving the efficiency of determining the target lane line points, and then improving the accuracy and response speed of trajectory prediction.

[0051] The driving speed information may include the current driving speed, current driving acceleration and driving speed limit of the vehicle at the current position. The driving speed limit refers to the maximum driving speed or minimum driving speed of the vehicle specified by the traffic restriction information.

[0052] As a possible implementation, the target lane line sampling area can be determined from a first lane line sampling area in front of the vehicle and a second lane line sampling area in front of the vehicle based on the vehicle's driving speed information, wherein the range of the first lane line sampling area is determined based on the vehicle's driving speed information, and the range of the second lane line sampling area is determined based on a preset distance in front of the vehicle.

[0053] It can be understood that the range of the second lane line sampling area is determined based on the preset distance in front of the vehicle, that is, the second lane line sampling area is pre-set and relatively fixed, and the vehicle does not need to recalculate the second lane line sampling area every time it moves, thereby reducing the calculation burden. The range of the first lane line sampling area is determined based on the vehicle's driving speed information, that is, the first lane line sampling area is relatively dynamic and can be flexibly adapted according to the vehicle's driving speed to ensure that a high lane line detection accuracy can be maintained at various speeds.

[0054] In addition, regarding determining the target lane line sampling area from the first lane line sampling area in front of the vehicle and the second lane line sampling area in front of the vehicle, reference may be made to the specific descriptions in S201 and S202 below, which will not be elaborated in this application.

[0055] As another possible implementation, the current driving speed included in the driving speed information of the vehicle is positively correlated with the range of the target lane line sampling area. The greater the current driving speed, the larger the range of the target lane line sampling area. Conversely, the smaller the current driving speed, the smaller the range of the target lane line sampling area. Therefore, the target lane line sampling area can be determined based on the positive correlation between the current driving speed included in the driving speed information of the vehicle and the range of the target lane line sampling area.

[0056] It can be understood that directly adjusting the range of the target lane line sampling area according to the driving speed information reduces the data processing process compared to determining the target lane line sampling area from the first lane line sampling area and the second lane line sampling area in the above implementation method, and can adapt to different driving speed information more quickly.

[0057] S102: Determine a target lane line point on a lane line within a target lane line sampling area, so as to predict a trajectory of the vehicle based on the target lane line point.

[0058] As a possible implementation, the number of lane lines in the target lane line sampling area can be one or more. For each lane line in the target lane line sampling area, at least one lane line point can be determined along the lane line according to a preset sampling step length, and each lane line point in the at least one lane line point can be used as a candidate lane line point and constitute a candidate lane line point set. Furthermore, the confidence of each candidate lane line point can be determined, and the candidate lane line point with the highest confidence can be determined as the target lane line point. The preset sampling step length refers to the distance between two adjacent candidate lane line points, for example, the preset sampling step length is 3 meters.

[0059] In addition, regarding the process of determining the target lane line point based on the confidence of each candidate lane line point, reference may be made to the description in S401-S404 below, which will not be elaborated in this application.

[0060] As another possible implementation method, the vehicle's kinematic model (such as a vehicle dynamics model, a motion trajectory model, etc.) can be used to predict the target position that the vehicle may reach in the future. Based on the target position that may be reached, a lane line point that best matches the target position is selected on the lane line within the target lane line sampling area as the target lane line point.

[0061] As another possible implementation, when the target lane line sampling area is based on the positive correlation between the current driving speed included in the driving speed information of the vehicle and the range of the target lane line sampling area, after sampling the lane lines in the target lane line sampling area to obtain the candidate lane line point set, it is also necessary to filter out the candidate lane line point set that meets the constraint information from the candidate lane line point set based on the constraint information to reduce unnecessary or infeasible candidate lane line points. Then, the target lane line point is determined from the candidate lane line point set that meets the constraint information.

[0062] In addition, regarding the selection of a candidate lane line point set that meets the constraint information from the candidate lane line point set based on the constraint information, reference may be made to the specific description in S301-S302 below, which will not be elaborated in this application.

[0063] Based on the above S101 and S102, the vehicle trajectory prediction method provided by the present application has differences in the target positions that may be reached due to the vehicles traveling at different driving speed information. If the target lane line point is determined on the lane line within the fixed lane line sampling area, the efficiency of determining the target lane line point may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for a vehicle traveling at high speed, the fixed lane line sampling area may not cover the distant lane line points that the vehicle can actually reach, resulting in the vehicle determining the target lane line point based on the less relevant lane line points, which not only increases the processing burden, but also may cause the predicted trajectory to be inaccurate. For vehicles traveling at low speeds, the fixed sampling area may introduce too many unnecessary or infeasible lane line points, increasing the complexity of processing. Therefore, by dynamically adjusting the target lane line sampling area based on the vehicle's driving speed information, it can better adapt to the current driving state of the vehicle, reduce the number of infeasible candidate lane line points on the lane line within the target lane line sampling area, and thus when determining the target lane line point on the lane line within the target lane line sampling area, the efficiency of determining the target lane line point can be improved, thereby improving the accuracy and efficiency of trajectory prediction for the vehicle based on the target lane line point.

[0064] In some embodiments, since the range of the second lane line sampling area is determined based on a preset distance in front of the vehicle, that is, the second lane line sampling area is pre-set and relatively fixed, the vehicle does not need to recalculate the area every time during driving, thereby reducing the calculation burden and improving the stability of the target lane line sampling area, which helps to reduce the potential safety risks caused by the instability of the target lane line sampling area. The range of the first lane line sampling area is dynamically adjusted based on the vehicle's driving speed information, that is, the first lane line sampling area can be adapted in real time according to the vehicle's driving speed to ensure that the sampling of lane lines at different speeds can maintain a high accuracy. Therefore, by determining the target lane line sampling area from these two areas through the vehicle's driving speed information, a good balance can be found between safety and flexibility. In order to determine the target lane line sampling area from the first lane line sampling area in front of the vehicle and the second lane line sampling area in front of the vehicle based on the vehicle's driving speed information.

[0065] Therefore, as a possible implementation method, the above S101 can be specifically implemented as the following steps:

[0066] S201. Determine a range of a first lane line sampling area in front of the vehicle based on the vehicle's driving speed information.

[0067] The first lane line sampling area may be an irregular area or a regular circular area.

[0068] It can be understood that a regular circular area has a fixed shape and boundary, and can more easily adapt to various sampling algorithms, making it easier and more efficient to sample lane line points. Therefore, the first lane line sampling area can be a regular circular area.

[0069] The irregular area can flexibly adapt to complex road environments. For example, in complex scenes such as curved lanes or intersections, the irregular area can more effectively cover the lane line and improve the accuracy and completeness of lane line point sampling. Therefore, the first lane line sampling area can be an irregular area.

[0070] As a possible implementation method, when the first lane line sampling area is a regular circular area, the radius of the first lane line sampling area can be determined based on the vehicle's driving speed information, and then based on the radius of the first lane line sampling area, the scope of the first lane line sampling area can be determined.

[0071] For example, when the current driving speed of the vehicle exceeds the maximum driving speed in the driving speed limit, the distance traveled by the vehicle within a preset time period with the maximum driving speed in the driving speed limit and the current driving acceleration as conditions can be determined, and then the distance can be determined as the radius of the first lane line sampling area.

[0072] When the current driving speed of the vehicle is lower than the minimum driving speed in the restricted driving speed, the distance traveled by the vehicle within a preset time period with the minimum driving speed in the restricted driving speed and the current driving acceleration as conditions can be determined, and then the distance can be determined as the radius of the first lane line sampling area.

[0073] As another possible implementation, when the first lane line sampling area is an irregular area, the distance traveled by the vehicle within a preset time period can be determined by the vehicle's driving speed information, thereby determining the range of a circular area with a radius of the distance. Then, based on the distance and a predefined irregular area boundary division method, the boundary of the first lane line sampling area is determined so that the range of the first lane line sampling area matches the range of the circular area.

[0074] In some embodiments, the first lane sampling area can be determined to be an irregular area or a regular circular area according to the road scene the vehicle is in. It is understandable that a suitable shape of the first lane sampling area can be selected according to the road scene the vehicle is in, thereby improving the sampling effect of the lane line points.

[0075] S202: Determine a target lane line sampling area from the first lane line sampling area and the second lane line sampling area based on the range of the first lane line sampling area and the range of the second lane line sampling area in front of the vehicle.

[0076] As a possible implementation manner, the target lane line sampling area may be determined based on a size relationship between a range of the first lane line sampling area and a range of the second lane line sampling area.

[0077] In one example, when the range of the first lane line sampling area exceeds the range of the second lane line sampling area, the second lane line sampling area is determined as the target lane line sampling area.

[0078] It can be understood that if the range of the first lane line sampling area exceeds the range of the second lane line sampling area, it can be explained that the vehicle is traveling at a faster speed, the distance between the target lane line point that the vehicle may reach and the current position may be larger, and the range of the first lane line sampling area is larger. If the first lane line sampling area is determined as the target lane line sampling area, although it can cover the farther position that the vehicle may reach, it will also cover too many unnecessary or infeasible candidate lane line points, increasing the amount of calculation for determining the target lane line point. In contrast, although the second lane line sampling area may not be able to fully cover the farther position that the vehicle may reach, since the second lane line sampling area is usually pre-set based on industry experience, it can usually provide a stable and reliable candidate lane line point. Therefore, in order to reduce unnecessary or infeasible candidate lane line points while ensuring the safety of trajectory prediction, the second lane line sampling area can be determined as the target lane line sampling area.

[0079] In another example, when the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, the first lane line sampling area is determined as the target lane line sampling area.

[0080] It is understandable that if the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, it can be said that the vehicle is traveling at a slow speed, and the distance between the target lane line point that the vehicle may reach and the current position may be small. If a fixed second lane line sampling area is used, too many unnecessary or infeasible candidate lane line points may be introduced, increasing the complexity of processing. Therefore, the first lane line sampling area dynamically determined based on the driving speed information can be determined as the target lane line sampling area, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0081] For example, when the first lane line sampling area is a regular circular area, Figure 2 As shown, if the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, the first lane line sampling area is determined as the target lane line sampling area to determine the target lane line point among the candidate lane line points on the lane line within the target lane line sampling area.

[0082] Based on the above S201 and S202, the range of the second lane line sampling area is determined based on the preset distance in front of the vehicle, that is, the second lane line sampling area is pre-set and relatively fixed, and the vehicle does not need to recalculate the second lane line sampling area every time it moves, thereby reducing the calculation burden. The range of the first lane line sampling area is determined based on the vehicle's driving speed information, that is, the first lane line sampling area is relatively dynamic and can be flexibly adapted according to the vehicle's driving speed to ensure that a high lane line detection accuracy can be maintained at various speeds. Through the combination of these two lane line sampling areas, a suitable sampling area can be selected based on the vehicle's driving speed information, which not only ensures the efficiency of determining the target lane line sampling area, but also improves the accuracy of determining the target lane line sampling area.

[0083] In some embodiments, based on the positive correlation between the current speed of the vehicle and the range of the target lane line sampling area, if the vehicle's speed is faster, the vehicle can reach a farther target lane line point, and the range of the target lane line sampling area will also be larger. However, a larger target lane line sampling area may cover too many unnecessary or infeasible candidate lane line points. Therefore, in order to reduce unnecessary or infeasible candidate lane line points, some candidate lane line points can be eliminated according to the constraint information.

[0084] As a possible implementation manner, the above S102 may also be implemented as the following steps:

[0085] S301. Sampling is performed on lane lines within a target lane line sampling area to obtain a candidate lane line point set.

[0086] The candidate lane line point set includes at least one candidate lane line point.

[0087] In addition, reference may be made to the specific description of determining the candidate lane line point set in the above S102, which will not be elaborated in this application.

[0088] S302: constrain the candidate lane line point set based on the constraint information to obtain a candidate lane line point set that meets the constraint information.

[0089] The constraint information includes: traffic restriction information and vehicle performance information. Traffic restriction information can be used to specify the maximum or minimum driving speed of the vehicle. The vehicle performance information is determined based on the vehicle hardware information, and the vehicle performance information can include the vehicle's braking information, the vehicle's acceleration information, etc.

[0090] As a possible implementation, the maximum deceleration and maximum acceleration of the vehicle can be determined based on the constraint information. Then, the distance that the vehicle travels within a preset time period at the current speed and the maximum deceleration is determined, and the near-end deceleration range is determined based on the distance. In addition, the distance that the vehicle travels within a preset time period at the current speed and the maximum acceleration is determined, and the far-end acceleration range is determined based on the distance.

[0091] Next, the candidate lane line points within the proximal deceleration range and the candidate lane line points outside the proximal deceleration range are eliminated to obtain a set of candidate lane line points that meet the constraint information.

[0092] S303: Determine a target lane line point from a set of candidate lane line points that meet the constraint information.

[0093] In addition, regarding determining the target lane line point from the candidate lane line point set, reference may be made to the specific description in S102 above, and this application will not elaborate on it here.

[0094] Based on the above S301-S303, due to the positive correlation between the current driving speed included in the vehicle's driving speed information and the range of the target lane line sampling area, if the vehicle's driving speed is faster, the vehicle can reach a farther target lane line point, and the range of the target lane line sampling area will also be larger. However, a larger target lane line sampling area may cover too many unnecessary or infeasible candidate lane line points. Therefore, the candidate lane line point set is constrained by constraint information, and candidate lane line points that do not meet the constraint information are eliminated, thereby reducing unnecessary or infeasible candidate lane line points and improving the efficiency of determining the target lane line points.

[0095] In some other embodiments, in order to determine the target lane line point on the lane line within the target lane line sampling area, the above S102 may also be implemented as the following steps:

[0096] S401. Obtain vehicle characteristics of the vehicle and map characteristics of the area where the vehicle is located.

[0097] Among them, vehicle features are used to represent the information of the vehicle itself, the road information around the vehicle detected by the sensors configured on the vehicle, and the information of other vehicles around the vehicle, which can reflect the complex interactive relationship between the vehicle and the roads around the vehicle and other vehicles. Map features are used to represent the road information in the map of the area where the vehicle is located.

[0098] As a possible implementation method, after the road information around the vehicle and other vehicle information around the vehicle are detected by the sensors configured on the vehicle, the road information around the vehicle, other vehicle information around the vehicle and the vehicle's own information can be feature-encoded through a neural network model to obtain vehicle features. Similarly, after the road information in the map of the area where the vehicle is located is obtained through a high-precision map, the road information features in the map of the area where the vehicle is located can be feature-encoded through a neural network model to obtain map features.

[0099] For example, the road information around the vehicle, the information of other vehicles around the vehicle, and the information of the vehicle itself can be embedded into a low-dimensional vector space through a multilayer perceptron (MLP) to obtain the vehicle characteristics of the vehicle. And through the nonlinear transformation of the multilayer network, the complex interaction relationship between the vehicle itself and the roads around the vehicle and other vehicles can be learned, that is, the vehicle characteristics can reflect the complex interaction relationship between the vehicle itself and the roads around the vehicle and other vehicles. Similarly, the road information in the map of the area where the vehicle is located can be embedded into a low-dimensional vector space through MLP to obtain the map characteristics of the vehicle.

[0100] S402: Sampling is performed on lane lines within the target lane line sampling area to obtain a candidate lane line point set.

[0101] In addition, reference may be made to the specific description of determining the candidate lane line point set in the above S102, which will not be elaborated in this application.

[0102] S403: Determine the confidence level of each candidate lane line point in the candidate lane line point set based on the vehicle characteristics and the map characteristics.

[0103] As a possible implementation, a target historical trajectory matching the vehicle characteristics and the map characteristics can be determined from the vehicle's historical trajectory library through the vehicle characteristics and the map characteristics, and the target historical trajectory refers to a historical trajectory that is similar or consistent with the vehicle characteristics and the map characteristics. Further, based on the target historical trajectory, the probability of each candidate lane line point being determined as the target lane line point in the target historical trajectory can be determined, and the probability is determined as the confidence of each candidate lane line point.

[0104] In addition, the historical trajectory library of a vehicle may include the historical trajectory of the vehicle itself, and may also include the historical trajectory of other vehicles.

[0105] As another possible implementation, the confidence of each candidate lane line point can be determined through a machine learning model.

[0106] Exemplarily, based on the vehicle features and map features obtained in the above S401 example, a query vector, a key vector, and a value vector can be constructed through MLP, wherein the query vector is determined based on the vehicle's own information, the key vector is determined based on the road information around the vehicle, and the value vector is determined based on the information of other vehicles around the vehicle.

[0107] Next, the vehicle's attention coefficient for each candidate lane line point can be determined based on the query vector and the key vector. The attention coefficient of the candidate lane line point can be expressed as:

[0108]

[0109] Among them, α j is the attention coefficient of the candidate lane line point; Q i is the query vector; K j is the key vector; D is the vector dimension.

[0110] Finally, a soft maximum operation is performed on the attention coefficient of each candidate lane line point to obtain the confidence of each candidate lane line point. For example, the SoftMax function can be used to perform a soft maximum operation on the attention coefficient of each candidate lane line point to obtain a normalized attention coefficient, and the normalized attention coefficient is used as the confidence of the candidate lane line.

[0111] In addition, after determining the confidence of each candidate lane line point, the value vector can be weighted and summed according to the confidence of each candidate lane line point to obtain a weighted feature. The weighted feature can be expressed as:

[0112]

[0113] Among them, e attj is the weighted feature; J is the number of candidate lane line points; V j is a value vector.

[0114] It is understandable that due to the different confidences of different candidate lane points, the candidate lane points with higher confidence contribute more to the weighted features, while the candidate lane points with lower confidence contribute less to the weighted features. By combining the information of each candidate lane point through weighted summation, the candidate lane point that is most likely to be the target lane point can be highlighted, thereby improving the accuracy of trajectory prediction.

[0115] S404: Determine the target lane line point based on the confidence level of the candidate lane line point.

[0116] As a possible implementation method, the candidate lane line point with the highest confidence level may be determined as the target lane line point.

[0117] As another possible implementation method, each candidate lane line point has a corresponding confidence weight. Based on the confidence weight and confidence of each candidate lane line point, the weighted confidence of each candidate lane line can be determined, and then the candidate lane line point with the largest weighted confidence can be determined as the target lane line point to reduce the impact of noise on the confidence of the candidate lane line point, thereby improving the accuracy of the target lane line point selection.

[0118] In some embodiments, after determining the target lane line point, the model for determining the confidence of the target lane line point can be back-propagated and updated by determining the cross entropy loss between the confidence of the target lane line point and the confidence of the actual lane line point, wherein the confidence of the actual lane line point defaults to 1.

[0119] Based on the above S401-S404, since the vehicle characteristics are used to represent the information of the vehicle itself, the road information around the vehicle detected by the sensors configured on the vehicle, and the information of other vehicles around the vehicle, it can reflect the complex interactive relationship between the vehicle and the roads around the vehicle and other vehicles. When determining the target lane line point based on the vehicle characteristics and map characteristics, it can better adapt to the complex real traffic scene. Thereby improving the accuracy of determining the target lane line point.

[0120] In some embodiments, after determining the target lane line point, the vehicle trajectory can be predicted based on the target lane line point. Therefore, the vehicle trajectory prediction method provided in the embodiment of the present application can also include the following steps:

[0121] S501. Determine target position information of a vehicle traveling to a target position based on a target lane line point.

[0122] The position information (such as target position information, current position information) may include the offset information of the vehicle.

[0123] As a possible implementation manner, the target position may be a target lane line point.

[0124] S502: Determine a target driving scenario for the vehicle based on the target position and the current position of the vehicle.

[0125] It can be understood that by analyzing the target position and the current position, the road curvature of the vehicle's driving path from the current position to the target position can be predicted. Different road curvatures have an impact on the vehicle trajectory prediction. Therefore, the vehicle's driving scenes are divided by the target position and the vehicle's current position to adapt to the influence of different road curvatures, thereby improving the accuracy of trajectory prediction.

[0126] As a possible implementation manner, the target driving scene of the vehicle may be determined based on the relationship among the target position, the current position and the target driving scene, the target position and the current position.

[0127] Exemplarily, in the Frenet coordinate system, the target driving scene can be determined based on the lateral driving scene of the vehicle and the longitudinal driving scene of the vehicle. Therefore, the relationship between the target position, the current position and the target driving scene includes: the lateral relationship between the target position, the current position and the lateral driving scene, and the longitudinal relationship between the target position, the current position and the longitudinal driving scene. In this way, the lateral driving scene and the longitudinal driving scene can be determined based on the target position, the current position, the lateral relationship and the longitudinal relationship, and then the target driving scene can be determined based on the lateral driving scene and the longitudinal driving scene.

[0128] Among them, the horizontal driving scene is used to indicate the relationship between the lane of the vehicle at the current position and the lane of the vehicle at the target position. For example, the horizontal driving scene includes the lane keeping scene and the lane changing scene. The longitudinal driving scene is used to indicate the change of the driving speed of the vehicle from the current position to the target position. For example, the longitudinal driving scene includes the speed keeping scene, the parking scene, and the acceleration scene.

[0129] In some embodiments, the target driving state information of the vehicle driving to the target location is different in different target driving scenarios. Therefore, the target driving state information of the vehicle driving to the target location can be determined based on the target driving scenario.

[0130] For example, in the Frenet coordinate system, since the target driving scene can be determined based on the lateral driving scene of the vehicle and the longitudinal driving scene of the vehicle, the target driving state information of the vehicle when it travels to the target position can be determined based on the lateral driving state information corresponding to the lateral driving scene and the longitudinal driving state information corresponding to the longitudinal driving scene.

[0131] For the lateral driving scene, the lateral driving state information corresponding to the lateral driving scene is as follows:

[0132] Lane keeping scenario: In the lane keeping scenario, the target position is in the same lane as the current position, so the lateral driving status information corresponding to the lane keeping scenario is:

[0133]

[0134] Where d(t0) is the lateral offset of the candidate predicted trajectory at the initial time t0; is the lateral velocity of the candidate predicted trajectory at the initial time t0; is the lateral acceleration of the candidate predicted trajectory at the initial time t0.

[0135] Lane change scenario: In the lane change scenario, there is a difference between the lane where the target position is located and the lane where the current position is located. Therefore, there is a lateral offset between the target position and the current position. The lateral driving status information corresponding to the lane change scenario is:

[0136]

[0137] d i ∈(-d lane / 2,d lane / 2)

[0138] Among them, d i is the lateral offset of the target position, d lane is the width of the lane the vehicle is in.

[0139] For the longitudinal driving scene, the longitudinal driving state information corresponding to the longitudinal driving scene is as follows:

[0140] Uniform acceleration scenario: In order to simplify the trajectory prediction process, longitudinal driving scenarios, such as speed maintenance scenario, parking scenario, acceleration scenario, etc., can be unified into uniform acceleration scenarios. The longitudinal driving state information corresponding to the uniform acceleration scenario is:

[0141]

[0142] Among them, s(t0) is the longitudinal offset of the candidate prediction trajectory at the initial time t0; is the longitudinal velocity of the candidate predicted trajectory at the initial time t0; The longitudinal acceleration of the candidate predicted trajectory at the initial time t0.

[0143] It should be noted that the initial time is the time when the vehicle travels from the current position.

[0144] S503, determining a set of candidate predicted trajectories of the vehicle in the target driving scenario based on the current driving state information and current position information of the vehicle at the current position, the target position information, and the target driving state information of the vehicle driving to the target position in the target driving scenario.

[0145] The driving state information (such as target driving state information and current driving state information) may include driving speed, driving acceleration and the like.

[0146] Since trajectory prediction is to predict the path that a vehicle may travel in the future, and the description of the path depends on the coordinate system, trajectory prediction is usually performed in the coordinate system, and the driving state information and position information are usually represented in the form of the coordinate system.

[0147] Exemplarily, the position information may be the lateral position information and longitudinal position information of the vehicle in the Frenet coordinate system, and the driving state information may be the lateral driving state information and longitudinal driving state information of the vehicle in the Frenet coordinate system. Specifically, based on the actual trajectory of the vehicle, the driving state information of the vehicle in the x reference direction and the driving state information in the y reference direction in the Cartesian coordinate system, as well as the position information in the x reference direction and the position information in the y reference direction may be obtained. Further, the center line of the lane where the target lane line point is located is determined as the reference line of the Frenet coordinate system, and the driving state information of the vehicle in the x reference direction and the driving state information in the y reference direction in the Cartesian coordinate system, as well as the position information in the x reference direction and the position information in the y reference direction, are converted into the lateral position information and longitudinal position information of the vehicle in the Frenet coordinate system using the conversion formula for converting the Cartesian coordinate system to the Frenet coordinate system.

[0148] It can be understood that in the Frenet coordinate system, the motion of the vehicle can be decoupled into longitudinal motion along the centerline of the road and lateral motion perpendicular to the centerline of the road. This decoupling of the two-dimensional motion problem into two one-dimensional motion problems can reduce the complexity of trajectory prediction.

[0149] As a possible implementation method, a vehicle trajectory prediction model can be determined based on a vehicle kinematic model, and then the vehicle's current driving state information, current position information, target position information, and target driving state information are input into the vehicle trajectory prediction model to obtain a set of candidate predicted trajectories output by the vehicle trajectory prediction model.

[0150] It can be understood that since the vehicle kinematic model takes into account the basic physical characteristics and motion laws of the vehicle, such as displacement, speed, acceleration, steering and other factors, and the vehicle trajectory prediction model is determined based on the vehicle kinematic model, therefore, when trajectory prediction is performed based on the vehicle trajectory prediction model, a more accurate trajectory can be generated.

[0151] In one example, a quintic polynomial lateral trajectory prediction model and a quintic polynomial longitudinal trajectory prediction model of the vehicle are determined according to the vehicle kinematic model. Then, in the Frenet coordinate system, the vehicle's current lateral position information, current lateral driving state information, lateral target driving state information, and lateral target position information can be input into the quintic polynomial lateral trajectory prediction model to obtain a candidate lateral trajectory set output by the quintic polynomial lateral trajectory prediction model. The candidate lateral trajectory set can be expressed as:

[0152]

[0153] Among them, dset is the candidate lateral trajectory set, is the i-th candidate lateral trajectory, and t is the N sampling time points.

[0154] The vehicle's current longitudinal position information, current longitudinal driving state information, longitudinal target driving state information, and longitudinal target position information are input into the quintic polynomial longitudinal trajectory prediction model to obtain a candidate longitudinal trajectory set output by the quintic polynomial longitudinal trajectory prediction model. The candidate longitudinal trajectory set can be expressed as:

[0155]

[0156] Among them, s set is the candidate lateral trajectory set, is the jth candidate lateral trajectory.

[0157] Furthermore, each candidate lateral trajectory in the candidate lateral trajectory set is merged with the corresponding candidate longitudinal trajectory in the candidate longitudinal trajectory set to obtain a candidate prediction trajectory set. The candidate prediction trajectory set can be expressed as:

[0158]

[0159] Among them, f set is the candidate prediction trajectory set.

[0160] Exemplarily, the quintic polynomial lateral trajectory planning model of the vehicle can be expressed as:

[0161] d(t)=α0+α1t+α2t 2 +α3t 3 +α4t 4 +α5t 5

[0162] Among them, α0=d(t0),

[0163] In the above formula, if the candidate prediction trajectory ends at t f , T = t f -t0, then α3, α4, α5 can be determined by the following formula.

[0164]

[0165] Exemplarily, the quintic polynomial longitudinal trajectory planning model of the vehicle can be expressed as:

[0166] s(t)=β0+β1t+β2t 2 +β3t 3 +β4t 4 +β5t 5

[0167] Where, β0 = s(t0),

[0168] In the above formula, if the candidate prediction trajectory ends at t f , T = t f -t0, then β3, β4, β5 can be determined by the following formula.

[0169]

[0170] It should be noted that the end time refers to the time when the vehicle travels from the current position to the target position.

[0171] S504 , constraining the candidate prediction trajectory set based on the constraint information to obtain a candidate prediction trajectory set that meets the constraint information.

[0172] The constraint information includes traffic restriction information. In addition, the constraint information in S504 may be the same as or different from the constraint information in S302, and this application does not impose any specific limitation on this.

[0173] As a possible implementation method, the maximum acceleration limit and maximum curvature limit of the vehicle can be determined based on the constraint information, and then based on the maximum acceleration limit and the maximum curvature limit, the candidate predicted trajectories that do not meet the maximum acceleration limit and the maximum curvature limit are eliminated to obtain at least one candidate predicted trajectory that meets the maximum acceleration limit and the maximum curvature limit.

[0174] Exemplarily, the candidate predicted trajectory that meets the maximum acceleration limit and the maximum curvature limit can be expressed as:

[0175] Acc(t) <Max_Acc

[0176] Curv(t) <Max_Curv

[0177] Wherein, Acc(t) is the acceleration at each moment in the candidate prediction trajectory; Max_Acc is the maximum acceleration limit;

[0178] Curv(t) is the curvature at each moment in the candidate prediction trajectory; Max_Curv is the maximum curvature limit.

[0179] S505 : Determine a target prediction trajectory from a set of candidate prediction trajectories that meet the constraint information.

[0180] As a possible implementation method, a target trajectory optimization function corresponding to the target driving scenario can be used to determine a target prediction trajectory that meets the optimization goal of the target trajectory optimization function from a set of candidate prediction trajectories that meet the constraint information.

[0181] As another possible implementation manner, in a set of candidate prediction trajectories that meet the constraint information, each candidate prediction trajectory has a degree of compliance with the constraint information, and the candidate prediction trajectory with the highest degree of compliance is determined as the target prediction trajectory.

[0182] Based on the above S501-S505, the driving state of the vehicle from the current position to the target position may be different in different driving scenarios. For example, in the lane change scenario, the vehicle may not be located at the center line of the lane, and in the lane keeping scenario, the vehicle may be located at the center line of the lane, which results in different acceleration and speed of the vehicle at the target position. Therefore, when determining the candidate predicted trajectory, in addition to considering the current driving state information, current position information, and target position information, it is also necessary to combine the target driving state information of the vehicle driving to the target position in the target driving scenario. This method can adapt to different driving scenarios, improve the accuracy of the candidate predicted trajectory set, and thus improve the accuracy of the target trajectory prediction. In addition, since the constraint information includes traffic restriction information, under the constraint of the constraint information, unsafe trajectories that do not meet the constraint information can be excluded, so that the predicted trajectory is closer to the actual situation, thereby improving the accuracy and feasibility of the predicted trajectory.

[0183] The following is an exemplary description of the process of determining a candidate prediction trajectory set based on different driving scenarios.

[0184] like Figure 3 As shown, when determining the target driving scene, the determination of the target driving scene can be decomposed into the determination of the lateral driving scene and the determination of the longitudinal driving scene. For the lateral driving scene, the lateral driving scene can be determined to be a lane keeping scene or a lane changing scene. Based on the lateral driving scene, the lateral target driving state information can be determined to be the target driving state information corresponding to the lane keeping scene or the target driving state information corresponding to the lane changing scene. Then, based on the lateral target driving state information and other lateral vehicle information, a lateral trajectory set can be determined.

[0185] For longitudinal driving scenarios, vehicle speed maintenance scenarios, acceleration scenarios, and parking scenarios can all be simplified to uniform acceleration scenarios. After determining that the longitudinal driving scenario is a uniform acceleration scenario, the longitudinal target driving state information can be determined as the target driving state information corresponding to the uniform acceleration scenario, and then based on the longitudinal target driving state information and other longitudinal vehicle information, the longitudinal trajectory set can be determined.

[0186] Finally, the horizontal trajectory set and the vertical trajectory set are combined to obtain the candidate prediction trajectory set.

[0187] In some embodiments, in order to determine the target predicted trajectory from the candidate predicted trajectory set that meets the constraint information, the above S505 can be implemented as the following steps:

[0188] S601. Obtain a target trajectory optimization function corresponding to a target driving scenario.

[0189] It is understandable that since the vehicle trajectory may be different in different driving scenarios, if the optimization is only based on a fixed optimization target, the target prediction trajectory may not be accurate enough. Therefore, by obtaining the target trajectory optimization function corresponding to the target driving scenario, the candidate prediction trajectory set is optimized using the scenario-specific target trajectory optimization function, thereby obtaining a more accurate target prediction trajectory.

[0190] Exemplarily, in the Frenet coordinate system, since the target driving scene can be determined based on the lateral driving scene and the longitudinal driving scene of the vehicle, the target trajectory optimization function corresponding to the target driving scene can be determined based on the lateral trajectory optimization function corresponding to the lateral driving scene and the longitudinal trajectory optimization function corresponding to the longitudinal driving scene. In this way, the lateral trajectory optimization function corresponding to the lateral driving scene and the longitudinal trajectory optimization function corresponding to the longitudinal driving scene can be obtained, and then the lateral trajectory optimization function and the longitudinal trajectory optimization function are merged to obtain the target trajectory optimization function corresponding to the target driving scene.

[0191] For the lateral driving scenario, the lateral trajectory optimization function is as follows:

[0192] Lane keeping scenario: In the lane keeping scenario, the lane at the target position is the same as the lane at the current position, and the vehicle does not change lanes. The lateral motion of the vehicle in the lane keeping scenario can be simplified to uniform acceleration motion. The lateral velocity of the vehicle at the current position is the same as the lateral velocity at the target position. Therefore, the lateral trajectory optimization function corresponding to the lane keeping scenario can be expressed as:

[0193]

[0194] Among them, C d is the lateral trajectory optimization function corresponding to the lane keeping scenario; J(d(t)) is the rate of change of lateral acceleration in time domain; K j is the weight coefficient of comfort, K d is the weight coefficient of the lateral trajectory behavior, t f For the end time, To predict the candidate prediction trajectory at the termination time t f The lateral velocity, is the candidate predicted trajectory derived according to uniform acceleration at the termination time t f The theoretical transverse velocity, Can be taken The lateral velocity of the vehicle at the target location

[0195] Lane changing scenario: In the lane changing scenario, the lane where the target position is located is different from the lane where the current position is located. The vehicle needs to change lanes. When reaching the target position, the vehicle is not necessarily located on the center line of the lane where the target position is located. Therefore, during the optimization process, the lateral target position can be sampled, and it is assumed that the speed and acceleration of the vehicle at the target position are zero. For this case, in order to improve the stability of the vehicle during the lane changing process, the lateral trajectory optimization function corresponding to the lane changing scenario can be expressed as:

[0196] C d =K j J(d(t))+K c C o 2

[0197] Among them, C d is the lateral trajectory optimization function corresponding to the lane change scenario; C o K is the distance between the lateral target position and the center line of the vehicle at the end time; c A weight factor representing the lateral trajectory behavior.

[0198] For the longitudinal driving scenario, the longitudinal trajectory optimization function is as follows:

[0199] Uniform acceleration scenario: In the uniform acceleration scenario, you can refer to the specific description of the lane keeping scenario in the above lateral driving scenario, and this application will not go into details here. The longitudinal trajectory optimization function corresponding to the uniform acceleration scenario can be expressed as:

[0200]

[0201] Among them, C s is the longitudinal trajectory optimization function corresponding to the uniform acceleration scenario; S(s(t)) is the rate of change of longitudinal acceleration in the time domain; K j is the weight coefficient of comfort, K s is the weight coefficient of the longitudinal trajectory behavior, t f For the end time, To predict the candidate prediction trajectory at the termination time t f The longitudinal velocity, is the candidate predicted trajectory derived according to uniform acceleration at the termination time t f The longitudinal theoretical velocity, Can be taken The longitudinal velocity of the vehicle at the target position

[0202] Based on the lateral trajectory optimization function and the longitudinal trajectory optimization function, the target trajectory optimization function corresponding to the target driving scenario can be expressed as:

[0203] C total =k d C d +k s C s

[0204] C total is the target trajectory optimization function, k d is the weight coefficient of the lateral trajectory optimization function, k s Weight coefficients for the longitudinal trajectory optimization function.

[0205] S602: Based on the target trajectory optimization function, the candidate prediction trajectory set is optimized to obtain the target prediction trajectory.

[0206] As a possible implementation method, the target trajectory optimization function has a corresponding optimization goal, such as minimizing the trajectory length, avoiding obstacles, etc., and then the candidate predictions that meet the optimization goals corresponding to the target trajectory optimization function can be concentrated as the target prediction trajectory.

[0207] Based on the above S601 and S602, since the vehicle trajectories may be different in different driving scenarios, if the optimization is performed only based on a fixed optimization target, the target prediction trajectory may be inaccurate. Therefore, by obtaining the target trajectory optimization function corresponding to the target driving scenario, the candidate prediction trajectory set can be optimized using the scenario-specific target trajectory optimization function, thereby obtaining a more accurate target prediction trajectory.

[0208] The following uses the graph attention module to determine the confidence of candidate lane line points as an example to illustrate the complete process of trajectory prediction.

[0209] like Figure 4 As shown, vehicle features and map features are input into the graph attention module to obtain the confidence of the candidate lane line points, and then the confidence of the candidate lane line points is input into the trajectory prediction module. The trajectory prediction module determines the target lane line points based on the confidence of the candidate lane line points, and then determines the target predicted trajectory of the vehicle based on the target lane line points.

[0210] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of the method. In order to achieve the above functions, the vehicle trajectory prediction device or the vehicle includes a hardware structure and / or software module corresponding to the execution of each function. It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0211] The embodiment of the present application can divide the functional modules of the vehicle trajectory prediction device or the vehicle according to the above method. For example, the vehicle trajectory prediction device or the vehicle may include various functional modules corresponding to the various functional divisions, or two or more functions may be integrated into one processing module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0212] Figure 5 is a block diagram of a vehicle trajectory prediction device according to an exemplary embodiment. Figure 5 The vehicle trajectory prediction device 500 includes: a determination module 501.

[0213] In one possible implementation, the determination module 501 is used to determine a target lane line sampling area based on the vehicle's driving speed information; determine a target lane line point on the lane line within the target lane line sampling area to predict the vehicle's trajectory based on the target lane line point.

[0214] According to the above technical means, the present application provides a vehicle trajectory prediction device. Since the vehicle travels with different driving speed information, the target position that it may reach is different. If the target lane line point is determined on the lane line within the fixed lane line sampling area, the efficiency of determining the target lane line point may be low, thereby affecting the accuracy and efficiency of trajectory prediction. For example, for a vehicle traveling at high speed, the fixed lane line sampling area may not cover the distant lane line points that the vehicle can actually reach, resulting in the vehicle determining the target lane line point based on less relevant lane line points, which not only increases the processing burden, but also may cause the predicted trajectory to be inaccurate. For a vehicle traveling at a low speed, the fixed sampling area may introduce too many unnecessary or infeasible lane line points, increasing the complexity of processing. Therefore, by dynamically adjusting the target lane line sampling area based on the vehicle's driving speed information, it can better adapt to the current driving state of the vehicle, reduce the number of infeasible candidate lane line points on the lane line within the target lane line sampling area, and thus when determining the target lane line point on the lane line within the target lane line sampling area, the efficiency of determining the target lane line point can be improved, thereby improving the accuracy and efficiency of trajectory prediction for the vehicle based on the target lane line point.

[0215] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0216] Figure 6 FIG. 1 is a block diagram of a vehicle according to an exemplary embodiment. Figure 6 As shown, vehicle 600 includes, but is not limited to, a processor 601 and a memory 602 .

[0217] The memory 602 is used to store executable instructions of the processor 601. It can be understood that the processor 601 is configured to execute instructions to implement the vehicle trajectory prediction method in the above embodiment.

[0218] It should be noted that those skilled in the art can understand that Figure 6 The vehicle structure shown in the figure does not constitute a limitation on the vehicle, and the vehicle may include Figure 6 More or fewer components may be shown, or certain components may be combined, or the components may be arranged differently.

[0219] The processor 601 is the control center of the vehicle, which uses various interfaces and lines to connect various parts of the entire vehicle, and executes various functions of the vehicle and processes data by running or executing software programs and / or modules stored in the memory 602, and calling data stored in the memory 602, so as to monitor the vehicle as a whole. The processor 601 may include one or more processing units. Optionally, the processor 601 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, etc., and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 601.

[0220] The memory 602 may be used to store software programs and various data. The memory 602 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 by at least one functional module (such as a determination unit, a processing unit, etc.), etc. In addition, the memory 602 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0221] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 602 including instructions, and the above instructions can be executed by a processor 601 of a vehicle 600 to implement the vehicle trajectory prediction method in the above embodiment.

[0222] In actual implementation, Figure 5 The functions of the determination module 501 in Figure 6 The processor 601 in the embodiment calls the computer program stored in the memory 602. The specific execution process can refer to the description of the method part in the above embodiment, which will not be repeated here.

[0223] Optionally, the computer-readable storage medium may be a non-temporary computer-readable storage medium, for example, the non-temporary computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0224] In an exemplary embodiment, the present application also provides a computer program product including one or more instructions, which can be executed by the processor 601 of the vehicle to complete the vehicle trajectory prediction method in the above embodiment.

[0225] It should be noted that when the instructions in the above-mentioned computer-readable storage medium or one or more instructions in the computer program product are executed by the vehicle's processor, the various processes of the above-mentioned method embodiment are implemented and the same technical effect as the above-mentioned method can be achieved. To avoid repetition, they will not be repeated here.

[0226] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0227] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0228] The units described as separate components may or may not be physically separated, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0229] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0230] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or the full classification part or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to execute the full classification part or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, disk or CD and other media that can store program code.

[0231] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto, and any changes or substitutions within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A vehicle trajectory prediction method, characterized in that: The method comprises: Based on the vehicle's driving speed information, determine the target lane line sampling area; A target lane line point is determined on the lane line within the target lane line sampling area to predict the trajectory of the vehicle based on the target lane line point.

2. The method according to claim 1, characterized in that The step of determining the target lane line sampling area based on the vehicle's driving speed information includes: Determine the range of a first lane line sampling area in front of the vehicle based on the driving speed information of the vehicle; Based on the range of the first lane line sampling area and the range of the second lane line sampling area in front of the vehicle, the target lane line sampling area is determined from the first lane line sampling area and the second lane line sampling area; the range of the second lane line sampling area is determined based on a preset distance in front of the vehicle.

3. The method according to claim 2, characterized in that The determining the target lane line sampling area from the first lane line sampling area and the second lane line sampling area based on the range of the first lane line sampling area and the range of the second lane line sampling area in front of the vehicle comprises: When the range of the first lane line sampling area exceeds the range of the second lane line sampling area, determining the second lane line sampling area as the target lane line sampling area; When the range of the first lane line sampling area does not exceed the range of the second lane line sampling area, the first lane line sampling area is determined as the target lane line sampling area.

4. The method according to claim 1, characterized in that: The driving speed information includes the current driving speed of the vehicle at the current position, and the range of the target lane line sampling area is positively correlated with the current driving speed; Determining a target lane line point on a lane line within the target lane line sampling area includes: Sampling is performed on the lane line within the target lane line sampling area to obtain a candidate lane line point set; the candidate lane line point set includes at least one candidate lane line point; Constraining the candidate lane line point set based on constraint information to obtain a candidate lane line point set that meets the constraint information; the constraint information includes: traffic restriction information and performance information of the vehicle; The target lane line point is determined from the set of candidate lane line points that meet the constraint information.

5. The method according to claim 1, characterized in that Determining a target lane line point on a lane line within the target lane line sampling area includes: Acquire vehicle features of the vehicle and map features of the area where the vehicle is located; the vehicle features are used to represent information about the vehicle itself, road information around the vehicle detected by sensors configured on the vehicle, and information about other vehicles around the vehicle; and the map features are used to represent road information in a map of the area where the vehicle is located; Sampling is performed on the lane line within the target lane line sampling area to obtain a candidate lane line point set; the candidate lane line point set includes at least one candidate lane line point; Determining, based on the interaction feature and the map feature, a confidence level of each candidate lane line point in the candidate lane line point set; The target lane line point is determined based on the confidence of the candidate lane line point.

6. The method according to claim 1, characterized in that The method further comprises: Determine target position information of the vehicle traveling to a target position based on the target lane line point; Determining a target driving scenario for the vehicle based on the target location and the current location of the vehicle; Determine a candidate prediction trajectory set of the vehicle in the target driving scenario based on the current driving state information and current position information of the vehicle at the current position, the target position information, and the target driving state information of the vehicle driving to the target position in the target driving scenario; the candidate prediction trajectory set includes at least one candidate prediction trajectory; Constraining the candidate prediction trajectory set based on constraint information to obtain a candidate prediction trajectory set that meets the constraint information; the constraint information includes traffic restriction information; A target predicted trajectory is determined from the set of candidate predicted trajectories that meet the constraint information.

7. The method according to claim 6, characterized in that The target driving scenario is determined based on the lateral driving scenario of the vehicle and the longitudinal driving scenario of the vehicle. The lateral driving scenario is used to indicate the relationship between the lane of the vehicle at the current position and the lane of the vehicle at the target position; the longitudinal driving scenario is used to indicate the change in the driving speed of the vehicle from the current position to the target position.

8. The method according to claim 6, characterized in that The determining a target predicted trajectory from the set of candidate predicted trajectories that meet the constraint information includes: Obtaining a target trajectory optimization function corresponding to the target driving scenario; Based on the target trajectory optimization function, the candidate prediction trajectory set is optimized to obtain the target prediction trajectory.

9. A vehicle trajectory prediction device, characterized in that: include: A determination module, used to determine a target lane line sampling area based on the vehicle's driving speed information; A target lane line point is determined on the lane line within the target lane line sampling area to predict the trajectory of the vehicle based on the target lane line point.

10. A vehicle, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: When the computer-executable instructions stored in the computer-readable storage medium are executed by a processor of a vehicle, the vehicle is capable of performing the method of any one of claims 1 to 8.

12. A computer program product, characterized in that The computer program product comprises computer instructions which, when executed on a vehicle, cause the vehicle to perform the method according to any one of claims 1 to 8.

Images