Path generation method and device, vehicle and storage medium

Abstract

Embodiments of the present application provide a path generation method and device, a vehicle and a storage medium. The method comprises: obtaining a space-time semantic corridor; discretizing and scattering a plurality of basic trajectory points in the space-time semantic corridor to obtain an initial solution space; determining an enhanced graph corresponding to the initial solution space based on the plurality of basic trajectory points, the enhanced graph comprising a plurality of enhanced points and a plurality of enhanced edges, the enhanced edges being edges associated with two enhanced points; and determining an automatic driving trajectory at a current time from the initial solution space based on the enhanced graph and a trajectory cost function. The enhanced graph is determined based on the space-time semantic corridor, and the automatic driving trajectory is generated according to the enhanced points and the enhanced edges in the enhanced graph in combination with the trajectory cost function. Compared with generating the automatic driving trajectory only by the space-time semantic corridor, the automatic driving trajectory is more efficiently obtained in the driving scene by generating the automatic driving trajectory based on the enhanced graph, and the obtained automatic driving trajectory is a globally optimal solution.

Description

Technical Field

[0001] The present application relates to the automotive field, and specifically relates to a path generation method, device, vehicle, and readable storage medium. Background Art

[0002] Related autonomous driving vehicle trajectory prediction methods generally use a time-space separation planning method. However, in some complex scenarios, it is difficult to efficiently obtain the autonomous driving trajectory, and it is impossible to obtain the optimal solution, and often only a suboptimal solution can be obtained. Summary of the Invention

[0003] In view of the above problems, the present application proposes a path generation method, device, vehicle and storage medium to improve the above problems.

[0004] In a first aspect, an embodiment of the present application provides a path generation method, the method comprising: obtaining a spatiotemporal semantic corridor, the spatiotemporal semantic corridor being obtained by processing an initial trajectory point based on vehicle information and traffic environment information of a target vehicle obtained at a current moment, wherein the initial trajectory point is determined based on a behavior prediction result of the target vehicle at the current moment; discretizing and scattering a plurality of basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, the plurality of basic trajectory points being used for dynamic path planning; determining an enhanced graph corresponding to the initial solution space based on the plurality of basic trajectory points, the enhanced graph comprising a plurality of enhanced points and a plurality of enhanced edges, the enhanced edge being an edge that associates two enhanced points, the plurality of enhanced points being determined based on the plurality of basic trajectory points; determining the automatic driving trajectory at the current moment from the initial solution space based on the enhanced graph and the trajectory cost function, so as to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

[0005] In a second aspect, an embodiment of the present application provides a path generation device, comprising: a spatiotemporal semantic corridor acquisition unit, configured to acquire a spatiotemporal semantic corridor, wherein the spatiotemporal semantic corridor is obtained by processing an initial trajectory point based on the vehicle information and traffic environment information of the target vehicle obtained at the current moment, wherein the initial trajectory point is determined based on the behavior prediction result of the target vehicle at the current moment; an initial solution space determination unit, configured to discretize and scatter multiple basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the multiple basic trajectory points are used for dynamic path planning; an enhanced graph determination unit, configured to determine an enhanced graph corresponding to the initial solution space based on the multiple basic trajectory points, wherein the enhanced graph includes multiple enhanced points and multiple enhanced edges, wherein the enhanced edge is an edge that associates two enhanced points, and the multiple enhanced points are determined based on the multiple basic trajectory points; a trajectory determination unit, configured to determine the automatic driving trajectory at the current moment from the initial solution space based on the enhanced graph and the trajectory cost function, so as to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

[0006] In a third aspect, an embodiment of the present application provides a vehicle comprising one or more processors and a memory; one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to execute the above-mentioned method.

[0007] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which program code is stored, wherein the above method is executed when the program code is run.

[0008] Embodiments of the present application provide a path generation method, apparatus, vehicle, and storage medium. This method determines an enhanced graph based on spatiotemporal semantic corridors and generates autonomous driving trajectories based on enhanced points and edges in the enhanced graph using a trajectory cost function. Compared to generating autonomous driving trajectories based solely on spatiotemporal semantic corridors, this method can more efficiently generate autonomous driving trajectories in driving scenarios, and the resulting autonomous driving trajectories are globally optimal. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A flow chart of a path generation method proposed in one embodiment of the present application is shown;

[0011] Figure 2 A schematic diagram showing an initial solution space of a path generation method proposed in an embodiment of the present application is shown;

[0012] Figure 3 A flow chart of a path generation method proposed in one embodiment of the present application is shown;

[0013] Figure 4 An enhanced diagram of a path generation method proposed in an embodiment of the present application is shown;

[0014] Figure 5 A structural block diagram of a path generation method proposed in another embodiment of the present application is shown;

[0015] Figure 6 A structural block diagram of a vehicle for executing the path generation method of an embodiment of the present application in real time is shown;

[0016] Figure 7 The present invention shows a storage unit in real time for storing or carrying program codes for implementing the path generation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] 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. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, and may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] In the embodiments of the present application, the inventors propose a path generation method, device, vehicle and storage medium. The method includes: obtaining a spatiotemporal semantic corridor; discretizing and scattering multiple basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space; based on the multiple basic trajectory points, determining an enhanced graph corresponding to the initial solution space, the enhanced graph including multiple enhanced points and multiple enhanced edges, where the enhanced edge is an edge that associates two enhanced points; based on the enhanced graph and the trajectory cost function, determining the autonomous driving trajectory at the current moment from the initial solution space. The enhanced graph is determined based on the spatiotemporal semantic corridor, and the autonomous driving trajectory is generated according to the enhanced points and enhanced edges in the enhanced graph in combination with the trajectory cost function. Compared with generating the autonomous driving trajectory only through the spatiotemporal semantic corridor, the present solution generates the autonomous driving trajectory through the enhanced graph, which can more efficiently obtain the autonomous driving trajectory in the driving scene, and the obtained autonomous driving trajectory is the global optimal solution.

[0020] See also Figure 1 , an embodiment of the present application provides a path generation method, the method comprising:

[0021] Step S110: Obtain a spatiotemporal semantic corridor, wherein the spatiotemporal semantic corridor is obtained by processing initial trajectory points based on the vehicle information and traffic environment information of the target vehicle obtained at the current moment, wherein the initial trajectory points are determined based on the behavior prediction result of the target vehicle at the current moment.

[0022] In an embodiment of the present application, the target vehicle itself is equipped with a perception module, through which the current moment target vehicle self-vehicle information and traffic environment information are obtained. The self-vehicle information includes vehicle speed information, heading angle information, length information and width information, etc. The traffic environment information includes the number of lanes in the lane where the target vehicle is located, lane width, weather conditions, the position of surrounding obstacles, the position of surrounding vehicles and the speed of surrounding vehicles, etc., which are not specifically limited here. Based on the self-vehicle information and traffic environment information, multiple semantic elements are generated, and the semantic elements are projected into the three-dimensional configuration space to obtain semantic boundaries. At the same time, the driving condition of the target vehicle is simulated according to the self-vehicle information of the target vehicle, and multiple initial trajectory points are determined according to the simulation results. Then, the multiple initial trajectory points are expanded to obtain multiple constraint spaces, and finally, the multiple constraint spaces are relaxed by soft constraints, so as to obtain a spatiotemporal semantic corridor in the three-dimensional configuration space according to the multiple relaxed constraint spaces. Among them, the perception module includes a camera, a lidar, etc., which are not specifically limited here. The three-dimensional configuration space is composed of a horizontal dimension, a vertical dimension and a time dimension. Among them, the behavior prediction result of the target vehicle is to predict the possible trajectory of the target vehicle in the future. After obtaining the behavior prediction result, the behavior prediction result is projected into the three-dimensional configuration space to obtain multiple initial trajectory points.

[0023] Step S120: Discretizing and scattering multiple basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the multiple basic trajectory points are used for dynamic path planning.

[0024] In an embodiment of the present application, a plurality of basic trajectory points are discretized and scattered in the spatiotemporal semantic corridor according to a preset resolution, and the space formed by the plurality of basic trajectory points after the discretization is used as the initial solution space. The preset resolution refers to the interval between the basic trajectory points. If a more accurate autonomous driving trajectory is desired, the preset resolution can be set to a larger value, but the consumption of computing resources will increase. If excessive computing resources are not desired, the preset resolution can be set to a smaller value, but this will result in an inaccurate autonomous driving trajectory. The basic trajectory points do not need to be obtained based on any parameters or information. The points scattered in the spatiotemporal semantic corridor can be considered as the basic trajectory points.

[0025] The preset resolution is expressed differently in the horizontal, vertical, and time dimensions. In the horizontal and vertical dimensions, the preset resolution is expressed as a preset distance; in the time dimension, the preset resolution is expressed as a sampling frequency. Specifically, in the time dimension, points can be discretized based on the sampling frequency. For example, if the sampling frequency is 0.5 seconds, a point will be scattered every 0.5 seconds in the time dimension. In both the horizontal and vertical dimensions, points can be discretized based on a preset distance. For example, if the preset distance is 0.2 meters in the horizontal dimension, a point will be scattered every 0.2 meters in the horizontal dimension.

[0026] For example, the initial solution space obtained in step S120 includes multiple basic trajectory points as follows: Figure 2 shown.

[0027] Step S130: Based on the multiple basic trajectory points, determine an enhanced graph corresponding to the initial solution space, wherein the enhanced graph includes multiple enhanced points and multiple enhanced edges, where the enhanced edge is an edge that associates two enhanced points, and the enhanced point is determined based on the multiple basic trajectory points.

[0028] In an embodiment of the present application, the enhanced graph is a graph used to determine the autonomous driving trajectory based on the trajectory cost function. According to the connection relationship between multiple basic trajectory points in the initial solution space, the enhancement point corresponding to each basic trajectory point is determined. Among them, an enhancement point includes the basic trajectory point corresponding to the enhancement point and up to three basic trajectory points before the basic trajectory point, that is, an enhancement point includes at least one basic trajectory point. After obtaining multiple enhancement points, based on the connection relationship of the basic trajectory points included in each enhancement point, the enhancement edge linking the two enhancement points is determined, so that multiple enhancement edges can be obtained. According to the multiple enhancement points and the enhancement edge linking the two enhancement points, the enhanced graph corresponding to the initial solution space is determined.

[0029] Step S140: Based on the enhanced graph and the trajectory cost function, determine the autonomous driving trajectory at the current moment from the initial solution space to control the target vehicle to perform autonomous driving based on the autonomous driving trajectory.

[0030] In this embodiment of the present application, the trajectory cost function is a function that calculates the cost of enhancement points. After obtaining an enhanced graph, the cost of each of the multiple enhancement points included in the enhanced graph is calculated using the trajectory cost function. This allows the current autonomous driving trajectory to be determined from the initial solution space based on the costs of each of the enhancement points, and the target vehicle is controlled to autonomously execute according to this autonomous driving trajectory.

[0031] The trajectory cost function includes horizontal and vertical parts. The horizontal part includes dl, ddl, dddl, obstacle distance cost, and reference line distance cost. The vertical part also includes smoothing cost v(ds), a(dds), jerk(ddds) and endpoint distance to constrain. The trajectory cost function can be expressed as:

[0032]

[0033] , where ω1 is the path angle cost weight, ω2 is the curvature cost weight, ω3 is the curvature change rate cost weight, ω collision is the collision cost weight with the obstacle, ω ref is the cost weight of the degree of deviation from the reference line, ω4 is the speed cost weight, ω5 is the acceleration cost weight, ω6 is the jerk cost weight, ω spatialpotential is the distance cost weight to the end point, t i is the t coordinate of the scattered points in the solution space, s i is the s coordinate of the scattered points in the solution space, l i is the l coordinate of the scattered points in the solution space, s end is the coordinate of the end point basic trajectory point s, f(s i ) is s i The l coordinate of the corresponding point, f′(s i) is s i The angle of the point, f″(s i ) is s i Point curvature, f″′(s i ) is s i The curvature change rate of the point, g[(s i -s) 2 +(l i -l) 2 ] is the distance between the scattering point and the obstacle, f(t i ) is t i The s coordinate of the corresponding point, f′(t i ) is t i The speed at time, f″(t i ) is t i The acceleration at time t i ) is t i The impact degree (jerk) when

[0034] An embodiment of the present application provides a path generation method that determines an enhanced graph based on spatiotemporal semantic corridors and generates autonomous driving trajectories based on enhanced points and edges in the enhanced graph in combination with a trajectory cost function. Compared to generating autonomous driving trajectories using only spatiotemporal semantic corridors, the autonomous driving trajectories generated by the enhanced graph in this solution can more efficiently obtain autonomous driving trajectories in driving scenarios, and the obtained autonomous driving trajectories are the global optimal solutions.

[0035] See also Figure 3 , an embodiment of the present application provides a path generation method, the method comprising:

[0036] Step S201: Obtain a spatiotemporal semantic corridor, wherein the spatiotemporal semantic corridor is obtained by processing initial trajectory points based on the vehicle information and traffic environment information of the target vehicle obtained at the current moment, wherein the initial trajectory points are determined based on the behavior prediction result of the target vehicle at the current moment.

[0037] The method for generating spatiotemporal semantic corridors is as follows: After acquiring vehicle information and traffic environment information through the perception module, these information is converted into semantic elements, resulting in multiple semantic elements. Semantic elements are represented by cubes, with the cube's boundaries serving as constraints. Semantic elements include obstacle-related and restriction-related elements. Obstacle-related semantic elements are determined based on obstacles and represent areas of the three-dimensional configuration space where entry is prohibited. Obstacle-related semantic elements can be categorized as static and dynamic. Static obstacle semantic elements are determined based on static obstacles, such as immovable obstacles like stone pillars, trees, and road bollards. Dynamic obstacle semantic elements are determined based on dynamic obstacles, such as movable obstacles like moving vehicles, pedestrians, and bicycles. Restriction semantic elements are determined based on motion constraints or time constraints. For example, a 60 km / h speed limit on a road would have a motion constraint of [0,60]. Each semantic element has its own corresponding constraints. For example, the constraints for the semantic element corresponding to the speed limit are the numbers on the speed limit sign, and the constraints for the semantic element corresponding to the road bollard are the length, width, and height of the bollard. Therefore, based on the constraints corresponding to each of the multiple semantic elements, the semantic elements are projected into the three-dimensional configuration space, and the first semantic boundary can be determined in the three-dimensional configuration space. For the dynamic obstacle semantic elements among the obstacle semantic elements, since these dynamic obstacle semantic elements move in the time dimension, the corresponding predicted trajectory must also be calculated. This calculation process requires calculating the lane change probability of the surrounding vehicles based on their orientation offset, lateral position offset, the positional relationship between the vertex of the surrounding vehicle box and the lane boundary, the degree of intrusion of the surrounding vehicle box into the current lane, and the lateral speed. Based on the obtained lane change probability, the corresponding predicted trajectory is generated and projected into the three-dimensional configuration space as the second semantic boundary. The first and second semantic boundaries serve as semantic boundaries corresponding to a three-dimensional configuration space. This three-dimensional configuration space is pre-constructed based on the horizontal, vertical, and time dimensions. In the three-dimensional configuration space, the horizontal and vertical dimensions are related to the Frenet framework, and each semantic element is represented as a cube in the three-dimensional configuration space. A multi-strategy decision model is used to perform a forward simulation based on the target vehicle's current vehicle information to predict the target vehicle's driving conditions for a preset time period in the future. The predicted driving conditions for the target vehicle in the future are used as the behavior prediction results for the target vehicle at the current moment. The preset time period is a pre-set time period, which can generally be set to 8 seconds. After obtaining the behavior prediction results, the prediction results are projected into the three-dimensional configuration space to obtain multiple initial trajectory points.The target vehicle's current position has an initial trajectory point, which is considered the first initial trajectory point. After obtaining multiple initial trajectory points, the initial trajectory point corresponding to the target vehicle's current position is expanded until it intersects the semantic boundary. The expansion of this initial trajectory point is determined to have ended, and the constraint space corresponding to this initial trajectory point is obtained. The expansion process then begins for the next initial trajectory point. After all initial trajectory points have been expanded, multiple constraint spaces corresponding to the multiple initial trajectory points are obtained. These multiple constraint spaces are then relaxed according to the soft constraints to obtain the spatiotemporal semantic corridor.

[0038] Step S202: Discretizing and scattering multiple basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the multiple basic trajectory points are used for dynamic path planning.

[0039] For details of step S202 , please refer to the detailed explanation in the above embodiment, so it will not be described in detail in this embodiment.

[0040] Step S203: Determine the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory point between the starting basic trajectory point and the ending basic trajectory point in each plane included in the initial solution space, where the starting basic trajectory point is the current position of the target vehicle and the ending basic trajectory point is the predicted position of the target vehicle.

[0041] In an embodiment of the present application, after obtaining the initial solution space, since the initial solution space is determined based on the space-time semantic corridor, the initial solution space is the same as the space-time semantic corridor, and is also a three-dimensional space, including a horizontal dimension, a vertical dimension, and a time dimension, and any two dimensions of the horizontal dimension, the vertical dimension, and the time dimension are perpendicular to each other, so that the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory point between the starting basic trajectory point and the ending basic trajectory point in each plane in the initial solution space can be determined.

[0042] As a way, since the final autonomous driving trajectory can be determined through two planes, we can also only obtain the basic trajectory points included in the plane composed of the longitudinal dimension and the time dimension, and the basic trajectory points included in the plane composed of the transverse dimension and the longitudinal dimension, without considering the basic trajectory points included in the plane composed of the transverse dimension and the time dimension, thereby saving computing resources and improving the computing speed.

[0043] Step S204: Based on the starting basic trajectory point, the ending basic trajectory point, and the connection relationship between the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point corresponding to each plane, determine multiple enhancement points corresponding to each plane, wherein one basic trajectory point corresponds to at least one enhancement point, and each enhancement point includes at least one basic trajectory point.

[0044] In an embodiment of the present application, in a plane, based on the starting basic trajectory point, the ending basic trajectory point, and the connection relationship between the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, the enhancement point corresponding to each basic trajectory point in the plane is determined to obtain multiple enhancement points corresponding to the plane, thereby obtaining multiple enhancement points corresponding to each plane. The enhancement point includes the basic trajectory point corresponding to the enhancement point and up to three basic trajectory points preceding the basic trajectory point.

[0045] Specifically, step S204 may be as described in steps S2041 and S2042.

[0046] Step S2041: For each basic trajectory point on each plane, based on the connection relationship between the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, trace back up to three basic trajectory points in the direction of the starting basic trajectory point to obtain the traced basic trajectory point corresponding to each basic trajectory point on each plane.

[0047] In the embodiment of the present application, a plurality of basic trajectory points are obtained by discretizing points according to a preset resolution in the spatiotemporal semantic corridor. According to the acceleration formula:

[0048]

[0049] And the impact formula:

[0050]

[0051] It can be seen that when calculating the acceleration and impact corresponding to a basic trajectory point, a maximum of three basic trajectory points will be traced back. The acceleration and impact are calculated based on the status of the three traced basic trajectory points. If there are fewer than three basic trajectory points, the impact cannot be calculated; if there are more than three basic trajectory points, the status of one basic trajectory point is not used, resulting in a waste of resources. On a plane, the direction from the starting basic trajectory point to the ending basic trajectory point is set as the positive direction. For a basic trajectory point, based on the connection relationship between the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, a maximum of three basic trajectory points are traced back in the reverse direction. The basic trajectory points traced back are used as the corresponding traced basic trajectory points for the basic trajectory point, thereby obtaining the traced basic trajectory points corresponding to each basic trajectory point on each plane. In the process of tracing back in the reverse direction, if it is determined based on the connection relationship that there are only two basic trajectory points in the reverse direction, then these two basic trajectory points are used as the corresponding traced basic trajectory points; if there is only one basic trajectory point, then this basic trajectory point is used as the corresponding traced basic trajectory point.

[0052] Step S2042: Based on each basic trajectory point on each plane and the traced basic trajectory point corresponding to each basic trajectory point on each plane, determine the enhancement point corresponding to each basic trajectory point on each plane to obtain multiple enhancement points corresponding to each plane.

[0053] In an embodiment of the present application, as described in step S2041, calculating the acceleration and impact of a basic trajectory point requires tracing back three basic trajectory points. However, when determining the autonomous driving trajectory, if the basic trajectory points are used alone, it is obvious that the state of the previous basic trajectory point will affect the selection of the next basic trajectory point. When performing dynamic programming, we need to avoid this situation. Therefore, on a plane, after obtaining a basic trajectory point and the traced basic trajectory point corresponding to the basic trajectory point, the basic trajectory point and the traced basic trajectory point corresponding to the basic trajectory point are packaged to obtain the enhancement point corresponding to the basic trajectory point, thereby obtaining the enhancement point corresponding to each basic trajectory point on the plane, and thus obtaining multiple enhancement points corresponding to each plane. Through the above processing, when subsequently determining the autonomous driving trajectory, it can be determined directly based on the enhancement point, avoiding the situation where the state of the previous basic trajectory point affects the selection of the next basic trajectory point.

[0054] Step S205: Determine the input edges and output edges corresponding to the multiple enhancement points corresponding to each plane, where the input edges are edges formed by the first specified basic trajectory points included in each enhancement point, and the output edges are edges formed by the second basic trajectory points included in each enhancement point, wherein the first specified basic trajectory point is a basic trajectory point arranged at a first specified position among the at least one basic trajectory point included in each enhancement point, and the second specified basic trajectory point is a basic trajectory point arranged at a second specified position among the at least one basic trajectory point included in each enhancement point.

[0055] In an embodiment of the present application, after obtaining multiple enhancement points corresponding to each plane, for the multiple enhancement points corresponding to a plane, the connection relationship of at least one basic trajectory point included in each enhancement point is determined, the edge formed by the basic trajectory points arranged at a first specified position among the at least one basic trajectory point included in the enhancement point is determined as the input edge, and the edge formed by the basic trajectory points arranged at a second specified position among the at least one basic trajectory point included in the enhancement point is determined as the output edge, thereby obtaining the input edge and output edge corresponding to each of the multiple enhancement points corresponding to the plane, and further obtaining the input edge and output edge corresponding to each of the multiple enhancement points corresponding to each plane.

[0056] As a method, for the input edge, if the enhancement point includes two basic trajectory points, then the basic trajectory points arranged at the first specified position in the enhancement point are these two basic trajectory points, and the basic trajectory points arranged at the second specified position are also these two basic trajectory points, that is, the edge formed by these two basic trajectory points is both the input edge and the output edge; if the enhancement point includes three basic trajectory points, then the basic trajectory points arranged at the first specified position in the enhancement point are these three basic trajectory points, and the basic trajectory points arranged at the second specified position are also these three basic trajectory points, that is, the edge formed by these three basic trajectory points is both the input edge and the output edge; if the enhancement point includes four basic trajectory points, then the basic trajectory points arranged at the first specified position in the enhancement point are the first three basic trajectory points among the four basic trajectory points, and the basic trajectory points arranged at the second specified position are the last three basic trajectory points among the four basic trajectory points, that is, among the four basic trajectory points included in the enhancement point, the edge formed by the first three basic trajectory points serves as the input edge of the enhancement point, and the edge formed by the last three basic trajectory points serves as the output edge of the enhancement point. If the enhancement point includes more basic track points, the input edges and output edges are determined in the same way as when the enhancement point includes four basic track points.

[0057] Step S206: Determine multiple enhancement edges corresponding to each plane based on the input edges and output edges corresponding to the multiple enhancement points corresponding to each plane, wherein, for two enhancement points connected by the enhancement edge, the output edge of one enhancement point is the input edge of the other enhancement point.

[0058] In an embodiment of the present application, after obtaining multiple enhancement points corresponding to each plane, for the multiple enhancement points corresponding to a plane, the input edge and output edge of each enhancement point are compared, and two enhancement points with the same input edge and output edge are selected from the multiple enhancement points. It is determined that there is an enhancement edge for connection between the two enhancement points, and multiple enhancement edges corresponding to the plane can be obtained, thereby obtaining multiple enhancement edges corresponding to multiple planes.

[0059] Step S207: Determine an enhanced graph corresponding to each plane based on the multiple enhanced points corresponding to each plane and the multiple enhanced edges corresponding to each plane.

[0060] In an embodiment of the present application, after obtaining multiple enhancement points corresponding to a plane and multiple enhancement edges corresponding to the plane, the enhancement points connected to each enhancement edge are determined, thereby determining the connection relationship between each enhancement point, and the enhancement graph corresponding to the plane can be determined, thereby determining the enhancement graph corresponding to each plane. In the enhancement graph, the enhancement points to which each enhancement point is connected through the enhancement edge can be viewed.

[0061] For example, steps S204 to S207 may be as follows: Figure 4 As shown, on the plane formed by the vertical dimension and the time dimension, the distribution of basic trajectory points is as follows Figure 4 , where the black basic trajectory point close to the coordinate origin is used as the starting basic trajectory point, and the black basic trajectory point far from the coordinate origin is used as the ending basic trajectory point. The time corresponding to the ending basic trajectory point is determined according to how long the autonomous driving trajectory is expected to be predicted. For example, if you want to predict an 8-second autonomous driving trajectory, the ending basic trajectory point is located in the column of t=8. Since the target vehicle cannot exist at two basic trajectory points at the same time, there is no connection between basic trajectory points at the same time. For each basic trajectory point, trace back up to three basic trajectory points in the direction of the starting basic trajectory point, and package each basic trajectory point and the basic trajectory points traced back from each basic trajectory point to obtain the enhancement point corresponding to each basic trajectory point. For example, in Figure 4 For the point above the second column in the starting basic track point, only one basic track point can be traced back. Then the basic track point above the second column and the basic track point obtained by tracing back are used as the enhancement point corresponding to the basic track point above the second column. Similarly, for Figure 4The basic trajectory points above the fourth column in the graph can be traced back to the starting basic trajectory point in four ways. Therefore, the basic trajectory points above the fourth column correspond to four enhancement points. Similarly, the enhancement points corresponding to each basic trajectory point can be obtained. After obtaining the enhancement points corresponding to each basic trajectory point, the input edges and output edges corresponding to each enhancement point are determined. For example, for the first enhancement point in the third column, since this enhancement point only has three basic trajectory points, the input edges and output edges of this enhancement point are both edges formed by these three basic trajectory points. For the first enhancement point in the fourth column, this enhancement point includes four basic trajectory points, of which the first three basic trajectory points are input edges and the last three basic trajectory points are output edges. At the same time, it can be determined that the output edge of the first enhancement point in the third column is the same as the input edge of the first enhancement point in the fourth column. Therefore, it can be determined that there is an enhancement edge between the first enhancement point in the third column and the first enhancement point in the fourth column. Similarly, the input edges and output edges included in each enhancement point can be determined, and then the enhancement edges connecting each enhancement point can be determined, thereby obtaining the enhanced graph corresponding to the plane formed by the longitudinal dimension and the time dimension.

[0062] Step S208: Determine the enhanced graph corresponding to the initial solution space based on the enhanced graph corresponding to each plane.

[0063] In the embodiment of the present application, after obtaining the enhanced graph corresponding to each plane, since the initial solution space is composed of three planes, the enhanced graphs corresponding to the three planes are used as the enhanced graphs corresponding to the initial solution space.

[0064] Step S209: In the current calculation round, determine at least one enhanced edge of the enhanced point corresponding to the current calculation round.

[0065] In an embodiment of the present application, in the current calculation round, the enhancement point corresponding to the current round on each plane is determined. Since each enhancement point is connected to at least one enhancement edge, at least one enhancement edge of the enhancement point corresponding to the current round on each plane can be determined.

[0066] Step S210: Determine at least one to-be-processed enhancement point connected to the at least one enhancement edge in the current calculation round.

[0067] In the embodiment of the present application, since two enhancement points are connected by an enhancement edge, after obtaining at least one enhancement edge corresponding to the current calculation round on each plane, at least one enhancement point to be processed connected to the at least one enhancement edge corresponding to the current calculation round on each plane can be determined based on the connection relationship.

[0068] Step S211: Based on the trajectory cost function, the cost of at least one to-be-processed enhancement point corresponding to the current calculation round is calculated.

[0069] In the embodiment of the present application, based on the trajectory cost function, the cost of at least one to-be-processed enhancement point corresponding to the current calculation round on each plane is calculated.

[0070] Step S212: Selecting the to-be-processed enhancement point with the lowest cost from the at least one to-be-processed enhancement point corresponding to the current calculation round as the planned enhancement point corresponding to the current calculation round.

[0071] In an embodiment of the present application, in a plane, the to-be-processed enhancement point with the lowest cost among at least one to-be-processed enhancement point corresponding to the current calculation round is selected as the planned enhancement point corresponding to the current calculation round on the plane, so that the planned enhancement point corresponding to the current calculation round on each plane can be obtained.

[0072] Step S213: taking the planned enhancement points corresponding to the current calculation round as enhancement points corresponding to the next calculation round.

[0073] In an embodiment of the present application, after obtaining the planning enhancement point corresponding to the current calculation round, the current calculation round is completed, and the planning enhancement point obtained in the current calculation round is used as the enhancement point corresponding to the next calculation round, and the next calculation round is subsequently continued.

[0074] Step S214: Based on the enhancement points corresponding to the next calculation round, continue to execute the next calculation round until the calculation end condition is met, and obtain the planning enhancement points corresponding to each calculation round.

[0075] In an embodiment of the present application, the planning enhancement point corresponding to the current calculation round on each plane is used as the enhancement point corresponding to the next calculation round, and the next calculation round is continued until the planning enhancement point corresponding to the end basic trajectory point on each plane is obtained, and it is determined that the calculation end condition is met, and the planning enhancement point corresponding to each calculation round on each plane is obtained.

[0076] Step S215: Based on the planning enhancement points corresponding to each calculation round, determine the automatic driving trajectory at the current moment from the initial solution space to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

[0077] In an embodiment of the present application, after obtaining the planning enhancement points corresponding to each calculation round, the basic trajectory points included in the planning enhancement points corresponding to each calculation round are determined in each plane, so that the planned path at the current moment is determined in each plane based on the basic trajectory points determined in each plane. Obviously, the planned path in each plane is the projection of the autonomous driving trajectory on the plane. Therefore, combined with the planned path corresponding to each plane, the autonomous driving trajectory at the current moment can be planned in the initial solution space, so as to control the target vehicle to perform autonomous driving according to the autonomous driving trajectory.

[0078] An embodiment of the present application provides a path generation method that determines an enhanced graph based on spatiotemporal semantic corridors and generates autonomous driving trajectories based on enhanced points and edges in the enhanced graph in combination with a trajectory cost function. Compared to generating autonomous driving trajectories using only spatiotemporal semantic corridors, the autonomous driving trajectories generated by the enhanced graph in this solution can more efficiently obtain autonomous driving trajectories in driving scenarios, and the obtained autonomous driving trajectories are the global optimal solutions.

[0079] See also Figure 5 , an embodiment of the present application provides a path generation device 300, the device 300 comprising:

[0080] The spatiotemporal semantic corridor acquisition unit 310 is used to obtain a spatiotemporal semantic corridor, which is obtained by processing the initial trajectory points based on the vehicle information and traffic environment information of the target vehicle obtained at the current moment, wherein the initial trajectory points are determined based on the behavior prediction result of the target vehicle at the current moment.

[0081] The initial solution space determining unit 320 is configured to discretize and scatter a plurality of basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the plurality of basic trajectory points are used for dynamic path planning.

[0082] The enhanced graph determination unit 330 is used to determine the enhanced graph corresponding to the initial solution space based on the multiple basic trajectory points, wherein the enhanced graph includes multiple enhanced points and multiple enhanced edges, where the enhanced edges are edges that associate two enhanced points, and the multiple enhanced edges are determined based on the multiple basic trajectory points.

[0083] As a method, the enhanced graph determination unit 330 is further used to determine the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory point between the starting basic trajectory point and the ending basic trajectory point in each plane included in the initial solution space, wherein the starting basic trajectory point is the current position of the target vehicle, and the ending basic trajectory point is the position predicted to be reached by the target vehicle; based on the connection relationship between the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory point between the starting basic trajectory point and the ending basic trajectory point corresponding to each plane, a plurality of enhanced points corresponding to each plane are determined, wherein one basic trajectory point corresponds to at least one enhanced point, and each enhanced point includes at least one basic trajectory point; and the input edges and output edges corresponding to the plurality of enhanced points corresponding to each plane are determined, wherein the input edges are the first specified basic trajectory points included in each enhanced point. The edge is composed of trajectory points, and the output edge is an edge composed of the second basic trajectory points included in each enhancement point, wherein the first specified basic trajectory point is a basic trajectory point arranged at a first specified position among the at least one basic trajectory point included in each enhancement point, and the second specified basic trajectory point is a basic trajectory point arranged at a second specified position among the at least one basic trajectory point included in each enhancement point; based on the input edges and output edges corresponding to the multiple enhancement points corresponding to each plane, the multiple enhancement edges corresponding to each plane are determined, wherein, among the two enhancement points connected by the enhancement edge, the output edge of one enhancement point is the input edge of the other enhancement point; based on the multiple enhancement points corresponding to each plane and the multiple enhancement edges corresponding to each plane, the enhanced graph corresponding to each plane is determined; based on the enhanced graph corresponding to each plane, the enhanced graph corresponding to the initial solution space is determined.

[0084] Optionally, the enhanced graph determination unit 330 is further configured to, for each basic trajectory point, trace back up to three basic trajectory points in the direction of the starting basic trajectory point based on the starting basic trajectory point, the ending basic trajectory point, and the connection relationship between the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, to obtain a traced basic trajectory point corresponding to each basic trajectory point; and determine an enhancement point corresponding to each basic trajectory point based on each basic trajectory point and the traced basic trajectory point corresponding to each basic trajectory point, to obtain the multiple enhancement points.

[0085] The trajectory determination unit 340 is used to determine the automatic driving trajectory at the current moment from the initial solution space based on the enhanced graph and the trajectory cost function, so as to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

[0086] As a method, the trajectory determination unit 340 is also used to determine at least one enhancement edge of the enhancement point corresponding to the current calculation round in the current calculation round; determine at least one to-be-processed enhancement point connected by the at least one enhancement edge in the current calculation round; calculate the cost of the at least one to-be-processed enhancement point corresponding to the current calculation round based on the trajectory cost function; select the to-be-processed enhancement point with the lowest cost from the at least one to-be-processed enhancement point corresponding to the current calculation round as the planned enhancement point corresponding to the current calculation round; use the planned enhancement point corresponding to the current calculation round as the enhancement point corresponding to the next calculation round; based on the enhancement point corresponding to the next calculation round, continue to execute the next calculation round until the calculation end condition is met, and obtain the planned enhancement point corresponding to each calculation round; based on the planned enhancement point corresponding to each calculation round, determine the automatic driving trajectory at the current moment from the initial solution space to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

[0087] It should be noted that the device embodiment in the present application corresponds to the aforementioned method embodiment. The specific principles in the device embodiment can be found in the contents of the aforementioned method embodiment and will not be repeated here.

[0088] The following will be combined Figure 6 A vehicle provided in this application is described.

[0089] See also Figure 6 Based on the above-described path generation method and apparatus, embodiments of the present application further provide another vehicle 400 capable of executing the aforementioned path generation method. Vehicle 400 includes one or more (only one shown in the figure) processors 402, a memory 404, and a network module 406, which are coupled to each other. The memory 404 stores a program capable of executing the contents of the above-described embodiments, and the processor 402 can execute the program stored in the memory 404.

[0090] Processor 402 may include one or more processing cores. Using various interfaces and circuits, processor 402 connects to various components within vehicle 400. By running or executing instructions, programs, code sets, or instruction sets stored in memory 404, and accessing data stored in memory 404, processor 402 executes various functions and processes data within server 400. Optionally, processor 402 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). Processor 402 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understandable that the above-mentioned modem may not be integrated into the processor 402, but may be implemented separately through a communication chip.

[0091] Memory 404 may include random access memory (RAM) or read-only memory (ROM). Memory 404 may be used to store instructions, programs, codes, code sets, or instruction sets. Memory 404 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the various method embodiments described below, and the like. The data storage area may also store data created by vehicle 400 during use (such as a phone book, audio and video data, chat history data, etc.).

[0092] The network module 406 is used to receive and transmit electromagnetic waves, realize the mutual conversion between electromagnetic waves and electrical signals, and thus communicate with a communication network or other devices, such as communicating with an audio playback device. The network module 406 may include various existing circuit components for performing these functions, such as an antenna, a radio frequency transceiver, a digital signal processor, an encryption / decryption chip, a subscriber identity module (SIM) card, a memory, etc. The network module 406 can communicate with various networks such as the Internet, an intranet, a wireless network, or communicate with other devices via a wireless network. The above-mentioned wireless network may include a cellular telephone network, a wireless local area network, or a metropolitan area network. For example, the network module 406 can exchange information with a base station.

[0093] Please refer to Figure 7 , which shows a block diagram of a computer-readable storage medium provided in an embodiment of the present application. The computer-readable storage medium 500 stores program code, which can be called by a processor to execute the method described in the above method embodiment.

[0094] The computer-readable storage medium 500 can be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read-Only Memory), an EPROM, a hard disk, or a ROM. Alternatively, the computer-readable storage medium 500 includes a non-volatile computer-readable storage medium. The computer-readable storage medium 500 has storage space for program code 510 that executes any of the method steps in the above method. These program codes can be read from or written into one or more computer program products. The program code 510 can, for example, be compressed in an appropriate form.

[0095] The embodiments of the present application provide a path generation method, device, vehicle and storage medium. The method includes: obtaining a spatiotemporal semantic corridor; discretizing and scattering multiple basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space; based on the multiple basic trajectory points, determining an enhanced graph corresponding to the initial solution space, the enhanced graph including multiple enhanced points and multiple enhanced edges, where the enhanced edge is an edge that associates two enhanced points; based on the enhanced graph and the trajectory cost function, determining the autonomous driving trajectory at the current moment from the initial solution space. The enhanced graph is determined on the basis of the spatiotemporal semantic corridor, and the autonomous driving trajectory is generated according to the enhanced points and enhanced edges in the enhanced graph in combination with the trajectory cost function. Compared with generating the autonomous driving trajectory only through the spatiotemporal semantic corridor, the present solution generates the autonomous driving trajectory through the enhanced graph, which can more efficiently obtain the autonomous driving trajectory in the driving scene, and the obtained autonomous driving trajectory is the global optimal solution.

[0096] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are protected by the present invention.

Claims

1. A path generation method, characterized in that: The method comprises: Obtaining a spatiotemporal semantic corridor, wherein the spatiotemporal semantic corridor is obtained by processing initial trajectory points based on vehicle information and traffic environment information of the target vehicle obtained at the current moment, wherein the initial trajectory points are determined based on a behavior prediction result of the target vehicle at the current moment; Discretizing and scattering a plurality of basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the plurality of basic trajectory points are used for dynamic path planning; Determining an enhanced graph corresponding to the initial solution space based on the multiple basic trajectory points, wherein the enhanced graph includes multiple enhanced points and multiple enhanced edges, where the enhanced edge is an edge that associates two enhanced points, and the enhanced point is determined based on the multiple basic trajectory points; Based on the enhanced graph and the trajectory cost function, the autonomous driving trajectory at the current moment is determined from the initial solution space to control the target vehicle to perform autonomous driving based on the autonomous driving trajectory.

2. The method according to claim 1, characterized in that The step of determining an enhanced graph corresponding to the initial solution space based on the multiple basic trajectory points includes: Determine a starting basic trajectory point, an ending basic trajectory point, and a basic trajectory point between the starting basic trajectory point and the ending basic trajectory point in each plane included in the initial solution space, wherein the starting basic trajectory point is the current position of the target vehicle, and the ending basic trajectory point is the predicted position of the target vehicle; Determine a plurality of enhancement points corresponding to each plane based on the starting basic trajectory point, the ending basic trajectory point, and a connection relationship between the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, wherein one basic trajectory point corresponds to at least one enhancement point, and each enhancement point includes at least one basic trajectory point; Determine an input edge and an output edge corresponding to each of the plurality of enhancement points corresponding to each plane, wherein the input edge is an edge constituted by a first designated basic trajectory point included in each enhancement point, and the output edge is an edge constituted by a second designated basic trajectory point included in each enhancement point, wherein the first designated basic trajectory point is a basic trajectory point arranged at a first designated position among at least one basic trajectory point included in each enhancement point, and the second designated basic trajectory point is a basic trajectory point arranged at a second designated position among at least one basic trajectory point included in each enhancement point; Determine multiple enhancement edges corresponding to each plane based on the input edges and output edges corresponding to the multiple enhancement points corresponding to each plane, wherein, of the two enhancement points connected by the enhancement edge, the output edge of one enhancement point is the input edge of the other enhancement point; Determining an enhanced graph corresponding to each plane based on the multiple enhanced points corresponding to each plane and the multiple enhanced edges corresponding to each plane; Based on the enhanced graph corresponding to each plane, an enhanced graph corresponding to the initial solution space is determined.

3. The method according to claim 2, characterized in that The determining of a plurality of enhancement points corresponding to each plane based on the starting basic trajectory point, the ending basic trajectory point, and a connection relationship between the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, includes: For each basic trajectory point on each plane, based on the connection relationship between the starting basic trajectory point, the ending basic trajectory point, and the basic trajectory points between the starting basic trajectory point and the ending basic trajectory point, trace back up to three basic trajectory points in the direction of the starting basic trajectory point to obtain the traced basic trajectory point corresponding to each basic trajectory point on each plane; Based on each basic track point on each plane and the traced basic track point corresponding to each basic track point on each plane, an enhancement point corresponding to each basic track point on each plane is determined to obtain a plurality of enhancement points corresponding to each plane.

4. The method according to claim 1, wherein The method of determining the current autonomous driving trajectory from the initial solution space based on the enhanced graph and the trajectory cost function, so as to control the target vehicle to perform autonomous driving based on the autonomous driving trajectory, includes: In a current calculation round, determining at least one enhanced edge of the enhanced point corresponding to the current calculation round; determining at least one to-be-processed enhancement point connected by the at least one enhancement edge in a current calculation round; Calculating, based on the trajectory cost function, a cost of at least one to-be-processed enhancement point corresponding to the current calculation round; Selecting, from the at least one to-be-processed enhancement point corresponding to the current calculation round, the to-be-processed enhancement point with the lowest cost as the planned enhancement point corresponding to the current calculation round; Using the planned enhancement point corresponding to the current calculation round as the enhancement point corresponding to the next calculation round; Based on the enhancement points corresponding to the next calculation round, continue to execute the next calculation round until the calculation end condition is met, and obtain the planning enhancement points corresponding to each calculation round; Based on the planning enhancement points corresponding to each calculation round, the automatic driving trajectory at the current moment is determined from the initial solution space to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

5. A path generation device, characterized in that: The device comprises: a spatiotemporal semantic corridor acquisition unit, configured to acquire a spatiotemporal semantic corridor, wherein the spatiotemporal semantic corridor is obtained by processing initial trajectory points based on vehicle information and traffic environment information of the target vehicle acquired at the current moment, wherein the initial trajectory points are determined based on a behavior prediction result of the target vehicle at the current moment; an initial solution space determining unit, configured to discretize and scatter a plurality of basic trajectory points in the spatiotemporal semantic corridor to obtain an initial solution space, wherein the plurality of basic trajectory points are used for dynamic path planning; an enhanced graph determining unit, configured to determine an enhanced graph corresponding to the initial solution space based on the multiple basic trajectory points, wherein the enhanced graph includes multiple enhanced points and multiple enhanced edges, where the enhanced edges are edges that associate two enhanced points, and the multiple enhanced points are determined based on the multiple basic trajectory points; A trajectory determination unit is used to determine the automatic driving trajectory at the current moment from the initial solution space based on the enhanced graph and the trajectory cost function, so as to control the target vehicle to perform automatic driving based on the automatic driving trajectory.

6. A vehicle, characterized in that: The method comprises one or more processors and a memory, wherein one or more programs are stored in the memory and configured to execute the method according to any one of claims 1 to 4 by the one or more processors.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores program codes, wherein the program codes include instructions for executing the method according to any one of claims 1 to 4.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images