Vehicle motion planning method, device, equipment and medium

Abstract

The present disclosure relates to a vehicle motion planning method, device, equipment and medium, wherein the method comprises: obtaining the predicted motion trajectory of each obstacle of the vehicle in the current environment; determining the type of each obstacle based on the predicted motion trajectory; obtaining the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and a preset motion cost function; wherein the motion cost function includes cost items related to collision-type obstacles and cost items related to non-collision-type obstacles; and obtaining the current motion trajectory of the vehicle based on the optimal solution of the motion cost function. The present disclosure can effectively improve the rationality and reliability of motion planning results.

Description

Technical Field

[0001] The present disclosure relates to the field of autonomous driving technology, and in particular to a vehicle motion planning method, device, equipment and medium. Background Art

[0002] With the development of science and technology, vehicle autonomous driving has gradually emerged in the field of transportation. In the process of autonomous driving, it is mainly necessary to plan the vehicle's movement trajectory in the future based on the perceived vehicle environment and preset driving tasks. The rationality of the motion planning will directly affect the safety of the vehicle's autonomous driving.

[0003] However, the inventors have discovered through research that although the existing vehicle motion planning technology takes into account that obstacles such as parallel trucks that are not predicted to have a collision relationship with the vehicle may also pose potential safety hazards to the vehicle, and takes corresponding separate processing measures during motion planning, the processing method is not good, resulting in poor rationality of the vehicle motion planning results and unreliable motion planning results. Summary of the invention

[0004] In order to solve the above technical problems or at least partially solve the above technical problems, the present disclosure provides a vehicle motion planning method, device, equipment and medium.

[0005] In a first aspect, an embodiment of the present disclosure provides a vehicle motion planning method, comprising: obtaining a predicted motion trajectory of each obstacle of the vehicle in a current environment; determining the type of each obstacle based on the predicted motion trajectory; wherein the types include collision-type obstacles and non-collision-type obstacles, the predicted motion trajectory of the collision-type obstacles has at least partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision-type obstacles has no path overlap with the planned path of the vehicle; according to the type of each obstacle, the predicted motion trajectory and a preset motion cost function, obtaining an optimal solution of the motion cost function; wherein the motion cost function includes cost terms related to the collision-type obstacles and cost terms related to the non-collision-type obstacles; and obtaining the current motion trajectory of the vehicle based on the optimal solution of the motion cost function.

[0006] In a second aspect, an embodiment of the present disclosure provides a vehicle motion planning device, comprising: an obstacle trajectory acquisition module, used to acquire the predicted motion trajectory of each obstacle of the vehicle in the current environment; an obstacle type determination module, used to determine the type of each obstacle based on the predicted motion trajectory; wherein the types include collision-type obstacles and non-collision-type obstacles, the predicted motion trajectory of the collision-type obstacles has at least partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision-type obstacles has no path overlap with the planned path of the vehicle; a function optimal solution acquisition module, used to acquire the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and a preset motion cost function; wherein the motion cost function includes cost terms related to the collision-type obstacles and cost terms related to the non-collision-type obstacles; a vehicle trajectory acquisition module, used to obtain the current motion trajectory of the vehicle based on the optimal solution of the motion cost function.

[0007] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising: a processor; a memory for storing executable instructions of the processor; and the processor for reading the executable instructions from the memory and executing the instructions to implement the above-mentioned vehicle motion planning method.

[0008] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program is used to execute the above-mentioned vehicle motion planning method.

[0009] The above technical solution provided by the embodiment of the present disclosure can determine the type of each obstacle (collision obstacle or non-collision obstacle) based on the predicted motion trajectory of each obstacle in the current environment of the vehicle, and obtain the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and the preset motion cost function (including the cost items related to the collision obstacle and the cost items related to the non-collision obstacle). Finally, the current motion trajectory of the vehicle can be obtained based on the optimal solution of the motion cost function. The above method uniformly performs motion planning for non-collision obstacles and collision obstacles based on the same motion cost function, and obtains the vehicle motion trajectory according to the optimal solution of the function, which can effectively improve the rationality and reliability of the motion planning results.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0013] Figure 1 A schematic diagram of a process flow of a vehicle motion planning method provided by an embodiment of the present disclosure;

[0014] Figure 2 A schematic diagram of a scene recognition result provided by an embodiment of the present disclosure;

[0015] Figure 3 A schematic diagram of parallel time calculation for dangerous parallel vehicles provided in an embodiment of the present disclosure;

[0016] Figure 4 A decision-making schematic diagram provided for an embodiment of the present disclosure;

[0017] Figure 5 A motion planning schematic diagram provided for an embodiment of the present disclosure;

[0018] Figure 6 A motion planning schematic diagram provided for an embodiment of the present disclosure;

[0019] Figure 7 A schematic diagram of the structure of a vehicle motion planning device provided in an embodiment of the present disclosure;

[0020] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0022] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0023] In the relevant technology, during the process of automatic driving of the vehicle, the vehicle movement decision and planning are mainly carried out for obstacles that are predicted to have a collision relationship (or path occupancy relationship) with the vehicle. The above-mentioned types of obstacles can be referred to as collision obstacles. For obstacles that theoretically have no collision relationship with the vehicle (such as dangerous parallel vehicles, vehicles with no predicted trajectory to cut into the lane but with the intention to cut in, etc., which can be collectively referred to as non-collision obstacles), they will not be involved in the decision and planning together with the collision obstacles, but will be processed separately for non-collision obstacles. Specifically, there are two mainstream processing solutions, which are explained below:

[0024] The first existing solution is to divide the scene for vehicle motion decision-making and planning, divide non-collision obstacles into multiple types (also called multiple scenes, each type corresponds to a scene), and adopt different decision-making and planning strategies or parameters for different scenes, so as to achieve the expected processing of different types of non-collision obstacles. Although this solution can handle different scenes accordingly, it is necessary to switch different solutions and parameters according to the corresponding scenes when encountering different types of non-collision obstacles, which is easy to cause inconsistent vehicle behavior, thus affecting the smoothness and rationality of the vehicle. When multiple scenes appear together, it will cause confusion in system logic, destroy the stability of the system, make the final vehicle motion planning result unreasonable, and even cause safety problems.

[0025] The second existing solution is to limit the speed for non-collision obstacles. Different types of non-collision obstacles (or different scenarios) will correspond to different path speed limits. Although the second solution avoids the problems that may exist in the switching process of the first solution, the second solution directly limits the speed for non-collision obstacles, which directly compresses the space for decision-making and planning, making it impossible to guarantee the optimal motion planning result. In severe cases, it will lead to planning failure and cause the vehicle to brake suddenly.

[0026] In summary, the existing method of handling non-collision obstacles separately is not good, resulting in poor rationality of vehicle motion planning results. The motion planning results are unreliable and easily affect the safety of vehicle autonomous driving.

[0027] The defects in the treatment measures for non-collision obstacles adopted in the relevant technologies are the results obtained by the inventor after practice and careful research. Therefore, the process of discovering the above defects and the solutions to the above defects proposed in the embodiments of the present disclosure below should be recognized as the contributions made by the inventor to the present disclosure.

[0028] In order to improve the above-mentioned problems existing in the related art, the embodiments of the present disclosure provide a vehicle motion planning method, device, equipment and medium, which are explained in detail below.

[0029] Figure 1 The present invention provides a flow chart of a vehicle motion planning method provided by an embodiment of the present invention. The method can be executed by a vehicle motion planning device, wherein the device can be implemented by software and / or hardware and can generally be integrated into an electronic device. Figure 1 As shown, the method mainly includes the following steps S102 to S108:

[0030] Step S102, obtaining the predicted motion trajectory of each obstacle of the vehicle in the current environment.

[0031] In actual applications, the vehicle can collect environmental detection data through designated sensors during driving, wherein the designated sensors include but are not limited to one or more of ultrasonic sensors, laser radars, and visual sensors, and the environmental detection data can be images, point clouds, etc., which are not limited here. The vehicle can analyze the environmental detection data (the analysis process can be regarded as a scene recognition process), identify each obstacle from the environmental detection data according to the preset obstacle features, and analyze and predict the movement trajectory of each obstacle based on the continuously collected environmental detection data to obtain the predicted movement trajectory of each obstacle.

[0032] Step S104, determining the type of each obstacle based on the predicted motion trajectory; wherein the types include collision type obstacles and non-collision type obstacles, the predicted motion trajectory of the collision type obstacle has at least partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision type obstacle has no path overlap with the planned path of the vehicle.

[0033] The disclosed embodiment can determine whether each obstacle is a collision-type obstacle or a non-collision-type obstacle based on the predicted motion trajectory. It is understandable that the planned path of the vehicle in a certain period of time in the future is known. If the predicted motion trajectory of the obstacle has at least a partial path overlap with the planned path of the vehicle, then the obstacle is a collision-type obstacle, that is, the obstacle has a collision relationship with the vehicle, and it can also be said that the obstacle has a path occupancy relationship with the vehicle. If there is no path overlap between the predicted motion trajectory of the obstacle and the planned path of the vehicle, then the obstacle is a non-collision-type obstacle, that is, the obstacle has no collision relationship with the vehicle, or in theory, the obstacle will not occupy the path of the vehicle.

[0034] Although in theory non-collision obstacles will not collide with the vehicle, they will still bring certain safety hazards to the vehicle. For example, if the non-collision obstacle is a truck (which can be called a parallel truck) running on the lane next to the vehicle, the truck may suddenly merge and collide with the vehicle sideways, and the cargo on the truck may also fall, thereby affecting the safety of the vehicle; for example, if the non-collision obstacle is a stationary vehicle, the vehicle may also suddenly start and affect the safety of the vehicle. The above is only an exemplary description and should not be regarded as a limitation. It can be seen from the above examples that non-collision obstacles will also affect the safety of vehicle driving. However, it can be understood that the impact of non-collision obstacles on the vehicle is different from the impact of collision obstacles on the vehicle. Therefore, the embodiment of the present disclosure will also perform corresponding processing according to the type of obstacle in the subsequent motion planning. In practical applications, identifying the type of obstacle can also be attributed to the scene recognition process.

[0035] Step S106, obtaining an optimal solution of a motion cost function according to the type of each obstacle, the predicted motion trajectory and a preset motion cost function; wherein the motion cost function includes cost items related to collision obstacles and cost items related to non-collision obstacles.

[0036] In practical applications, the specific form of the motion cost function can be set according to the needs. The motion cost function includes cost items related to collision obstacles and cost items related to non-collision obstacles. According to the type of each obstacle and the corresponding predicted motion trajectory, the cost paid by the vehicle for driving along different motion trajectories can be calculated. Compared with the motion cost function in the related art that only includes cost items related to collision obstacles, in order to ensure that non-collision obstacles have a reasonable impact on the behavior of the vehicle, the embodiment of the present disclosure innovatively introduces cost items related to non-collision obstacles into the motion cost function, so as to unify the motion planning of non-collision obstacles and collision obstacles based on the same motion cost function and find the most reasonable motion trajectory. Among them, the motion trajectory can be represented by the corresponding relationship between time and waypoints, and is reflected in the ST space-time diagram (also called ST space-time occupancy diagram).

[0037] In practical applications, the minimum value of the motion cost function can be obtained, and the motion parameters of the vehicle corresponding to the minimum motion cost function value can be used as the optimal solution of the motion cost function. The motion parameters include but are not limited to vehicle parameters such as speed, angular velocity, acceleration, etc., which are the same as the parameters required for conventional vehicle motion planning, and will not be described one by one here.

[0038] Step S108, obtaining the current motion trajectory of the vehicle based on the optimal solution of the motion cost function.

[0039] Specifically, the motion trajectory corresponding to the optimal solution of the motion cost function can be used as the current motion trajectory of the vehicle. The current motion trajectory is the most reasonable and reliable motion planning result obtained from the current motion planning, and the motion cost required by the vehicle is the smallest.

[0040] The above method performs motion planning for non-collision obstacles and collision obstacles based on the same motion cost function, and obtains the vehicle motion trajectory according to the optimal solution of the function, which can effectively improve the rationality and reliability of the motion planning results.

[0041] In some implementations, according to the types of obstacles, the predicted motion trajectories, and the preset motion cost function, the step of obtaining the optimal solution of the motion cost function can be implemented by referring to the following steps A to C:

[0042] Step A: When the obstacles include non-collision obstacles, a target non-collision obstacle is selected from the non-collision obstacles, and first spatiotemporal occupancy information corresponding to the target non-collision obstacle is obtained based on a first predicted motion trajectory corresponding to the target non-collision obstacle.

[0043] In practical applications, non-collision obstacles that pose a greater risk to the vehicle can be used as target non-collision obstacles (also called non-collision obstacles of interest), and only the target non-collision obstacles are processed, while non-collision obstacles that pose a smaller risk to the vehicle are not processed. This method can effectively save system computing power while ensuring safety. The disclosed embodiments provide the following two examples of selecting target non-collision obstacles from non-collision obstacles:

[0044] Example 1: When a non-collision obstacle includes a parallel moving body, the closest distance between the parallel moving body and the vehicle is estimated according to the first predicted motion trajectory corresponding to the parallel moving body; when the closest distance is less than a preset distance threshold, the estimated parallel time between the parallel moving body and the vehicle is obtained; when the estimated parallel time is greater than the preset time threshold, the parallel moving body is selected as the target non-collision obstacle.

[0045] The above-mentioned parallel moving body is an object moving on a path parallel to the driving path of the vehicle, which can be a vehicle, a person, etc. It should be noted that the parallel moving body mainly refers to the moving path of the moving body being parallel to the path of the vehicle, and should not be understood as the moving body and the vehicle traveling side by side at the same time. For example, the parallel moving body can have a chronological relationship with the vehicle. For example, the parallel moving body can be a large vehicle traveling in a lane next to the lane where the vehicle is located, and the large vehicle may be located in front of, behind, or in parallel with the vehicle.

[0046] Although the moving path of the parallel moving body is parallel to the moving path of the vehicle, theoretically it will not affect the vehicle. However, there are still risks such as the parallel moving body suddenly changing lanes, objects on the parallel moving body suddenly falling and hitting the vehicle or falling onto the path of the vehicle and affecting the smooth passage of the vehicle. Therefore, when encountering a parallel moving body, it is necessary to evaluate the safety risk it brings to the vehicle. The safety risk assessment method can first be based on the closest distance between the vehicle and the parallel moving body. If the closest distance is greater than the preset distance threshold, it means that the distance between the vehicle and the parallel moving body is far, and the safety risk of the parallel moving body to the vehicle is low, so the parallel moving body can be ignored. If the closest distance is less than the preset distance threshold, it means that If the distance between the vehicle and the parallel moving body is relatively close, and the parallel moving body poses a certain safety risk to the vehicle, it can be further evaluated through the estimated parallel time. It can be understood that the longer the parallel time, the greater the safety risk of the parallel moving body relative to the vehicle, and the shorter the parallel time, the smaller the safety risk of the parallel moving body relative to the vehicle. Therefore, the time threshold can be set in advance. When the estimated parallel time is greater than the preset time threshold, the parallel moving body is selected as the target non-collision obstacle that needs to be processed. When the estimated parallel time is less than the preset time threshold, it means that the vehicle can pass the parallel moving body faster. In this case, the safety risk of the parallel moving body to the vehicle is relatively small, and there is no need to process the parallel moving body at this time.

[0047] Example 2: When a non-collision obstacle includes an immovable object, it is determined whether the immovable object has a tendency to affect the movement of the vehicle; if so, the immovable object is selected as a target non-collision obstacle.

[0048] The above-mentioned immobile body can be, for example, a stationary vehicle, a person, etc. In theory, the immobile body will not affect the vehicle, but there is still the possibility that the immobile body suddenly moves and affects the driving safety of the vehicle. Therefore, when encountering an immobile body, it is necessary to evaluate the safety risk it brings to the vehicle. In the disclosed embodiment, it can be determined whether the immobile body has a moving trend that affects the vehicle. Only when it has a moving trend that affects the vehicle will it be used as a target non-collision obstacle that needs to be processed. If the immobile body does not have a moving trend that affects the vehicle, it does not need to be processed. In some implementation examples, if the predicted moving trend of the immobile body indicates that there is an overlap between the predicted trajectory of the immobile body and the planned path of the vehicle, it means that the immobile body has a moving trend that affects the vehicle. For example, assuming that the vehicle is driving forward, and a vehicle A is parked on the roadside, assuming that the front of the vehicle A is tilted toward the lane, if the vehicle A starts, there is a safety risk of side-impacting the vehicle, so it can be considered that the vehicle A has a moving trend that affects the vehicle. For another example, the front of vehicle B at the roadside stop is facing away from the lane where the host vehicle is located. If vehicle B starts, it will move away from the lane where the host vehicle is located. At this time, it can be considered that vehicle B has no tendency to affect the movement of the host vehicle.

[0049] It should be noted that the above descriptions are exemplary and should not be regarded as limiting. In practical applications, the target non-collision obstacles to be processed can be flexibly selected from the determined non-collision obstacles according to needs.

[0050] On this basis, the embodiment of the present disclosure further provides an implementation method for obtaining the first space-time occupancy information corresponding to the target non-collision obstacle based on the first predicted motion trajectory corresponding to the target non-collision obstacle: projecting the first predicted motion trajectory corresponding to the target non-collision obstacle onto the planned path of the vehicle, and determining the first space-time occupancy information of the target non-collision obstacle on the planned path of the vehicle based on the projection result.

[0051] It is understandable that, under normal circumstances, it is believed that non-collision obstacles have no time-space occupation relationship with the vehicle, and only collision obstacles that have a time-space occupation relationship with the vehicle are processed. Specifically, the collision obstacles are converted into path time-space occupation individuals corresponding to the vehicle, and then decisions and planning are made on the time-space occupation individuals in time and space, so as to plan the time-space motion trajectory of the vehicle that does not overlap with the occupied time and space of the collision obstacles and satisfies the decision results. However, the disclosed embodiment not only takes into account the safety risks of non-collision obstacles to the vehicle, but also converts non-collision obstacles that have no time-space occupation relationship with the vehicle into virtual time-space occupation individuals. Specifically, the predicted motion trajectory of the non-collision obstacles is projected onto the planned path of the vehicle, thereby forming the time-space occupation of the vehicle path. In specific implementation, non-collision obstacles can be abstracted into a unified data structure with collision obstacles for simultaneous decision-making and planning. This method can not only improve the efficiency of motion planning, but also improve the rationality and reliability of motion planning.

[0052] Step B: when the obstacles include collision-type obstacles, each collision-type obstacle is used as a target collision-type obstacle, and second space-time occupancy information corresponding to the target collision-type obstacle is obtained based on a second predicted motion trajectory corresponding to the target collision-type obstacle.

[0053] In the disclosed embodiment, in order to ensure the safety of motion planning, each collision-type obstacle can be used as a target collision-type obstacle to be processed to ensure that the motion planning trajectory obtained by the vehicle does not overlap with the time and space of all collision-type obstacles, thereby ensuring that the vehicle can avoid collision with the collision-type obstacles. The method of obtaining the second time and space occupancy information corresponding to the target collision-type obstacle according to the second predicted motion trajectory corresponding to the target collision-type obstacle can be specifically implemented with reference to the relevant technology, and will not be described in detail here.

[0054] Step C, obtaining the optimal solution of the motion cost function according to the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and the preset motion cost function. In practical applications, vehicle motion planning can be performed when the spatiotemporal occupancy information corresponding to non-collision obstacles (i.e., the first spatiotemporal occupancy information), the spatiotemporal occupancy information corresponding to collision obstacles (i.e., the second spatiotemporal occupancy information), and the motion cost function pre-set to include cost items related to collision obstacles and cost items related to non-collision obstacles are known. Specifically, the vehicle motion planning can be constrained by the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and the motion cost, so that the motion cost required for the vehicle motion planning is minimized while not causing the vehicle's motion planning trajectory to overlap with the space and time corresponding to the collision obstacle. It should also be noted that the second spatiotemporal occupancy information has a strong spatiotemporal constraint on the vehicle motion planning, that is, the vehicle is not allowed to occupy the spatiotemporal area corresponding to the collision obstacle during the motion planning process; while the first spatiotemporal occupancy information has a soft spatiotemporal constraint on the vehicle motion planning, that is, under certain circumstances, the vehicle can be allowed to occupy the spatiotemporal area corresponding to the non-collision obstacle during the motion planning process. For example, in order to ensure the minimum motion cost, the vehicle can be allowed to quickly pass through the spatiotemporal area corresponding to the parallel vehicle or exit the area at a low speed when entering the spatiotemporal area.

[0055] When planning vehicle motion, it is necessary to first make a scene decision, that is, to make a decision on each obstacle in turn according to the order of the time and distance of space-time occupation, and then obtain the decision result of each obstacle (such as rushing or giving way). In some specific implementation examples, the above step C can be implemented with reference to the following steps C1 and C2:

[0056] Step C1, determining a first decision result corresponding to a non-collision obstacle and a second decision result corresponding to a collision obstacle according to the first time-space occupancy information and the second time-space occupancy information.

[0057] The first decision result and the second decision result may both include overtake or follow. The specific decision method can refer to the relevant technology and is not limited here. Compared with the relevant technology, the embodiment of the present disclosure mainly abstracts non-collision obstacles into a unified data structure with collision obstacles for simultaneous decision-making and planning, which improves the efficiency of decision-making and planning while also improving the accuracy and reliability of planning results.

[0058] Step C2, according to the first decision result, the second decision result, the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and the preset motion cost function, obtain the optimal solution of the motion cost function. In some specific implementation examples, the above step C2 can be implemented with reference to the following steps C2.1 to C2.3:

[0059] Step C2.1, determining a first space-time occupied area of ​​the target non-collision obstacle in the ST space-time diagram based on the first space-time occupancy information, and determining a second space-time occupied area of ​​the target collision obstacle in the ST space-time diagram based on the second space-time occupancy information.

[0060] In the ST space-time diagram, S represents the longitudinal displacement of the vehicle, and T represents time. The first space-time occupancy information can be mapped to the distance / time space of the vehicle to form the space-time occupancy area of ​​the non-collision type obstacle relative to the vehicle (i.e., the first space-time occupancy area). Accordingly, the second space-time occupancy information can be mapped to the distance / time space of the vehicle to form the space-time occupancy area of ​​the collision type obstacle relative to the vehicle (i.e., the second space-time occupancy area), so as to facilitate subsequent search on the ST space-time diagram and determine the motion planning path of the vehicle (which can be represented by lines in the ST space-time diagram).

[0061] Step C2.2, obtaining the expected speed of the vehicle in the first space-time occupied area according to the first decision result. Since the first space-time occupied area is the space-time occupied area of ​​non-collision type obstacles, the vehicle can be allowed to enter. On this basis, in order to minimize the impact of non-collision type obstacles on the vehicle, when the vehicle reaches the first space-time occupied area, the expected speed of the vehicle in the first space-time occupied area can be obtained, so that the vehicle can travel at the expected speed and achieve the effect of overtaking non-collision type obstacles at high speed or giving way to non-collision type obstacles at low speed. Exemplarily, the above step C2.2 can be implemented with reference to the following steps 1 to 3:

[0062] Step 1: determine the current speed of the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle.

[0063] Step 2: Determine the escape speed according to the current speed of the vehicle and the current speed of the target non-collision type obstacle. For example, the escape speed can be determined according to the difference between the current speed of the vehicle and the current speed of the target non-collision type obstacle.

[0064] Step 3, determining the expected speed of the vehicle in the first space-time occupied area according to the first decision result, the current speed of the target non-collision obstacle and the escape speed.

[0065] Specifically, when the first decision result indicates that the vehicle needs to give way to the target non-collision type obstacle, the difference between the current speed of the target non-collision type obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area; and when the first decision result indicates that the vehicle needs to rush over the target non-collision type obstacle, the sum of the current speed of the target non-collision type obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area. In other words, the expected speed is expressed as v sceneThe escape velocity is represented by ΔV, and the current velocity of the target non-collision obstacle is represented by scene_vel. For example, when the first decision result is that the vehicle yields, v scene = scene_vel - ΔV; and when the first decision result is that the vehicle cuts in, then v scene =scene_vel+ΔV.

[0066] The method for determining the above-mentioned expected speed provided in the embodiment of the present disclosure is relatively reasonable and reliable. By pre-setting the expected speed, it is helpful to further reasonably plan the vehicle's movement trajectory in the future, so that even if the vehicle is planned to enter the first space-time occupied area, the vehicle is driven to travel at the expected speed as much as possible, so as to achieve the effect of quickly overtaking non-collision obstacles or avoiding non-collision obstacles at a low speed, thereby minimizing the movement cost caused by non-collision obstacles.

[0067] It should also be noted that the collision type obstacle does not have an expected speed, which can also be understood as the expected speed of the collision type obstacle is zero. The reason is that during the motion planning process, the vehicle is not allowed to enter the second space-time occupied area corresponding to the collision type obstacle to avoid a collision between the vehicle and the collision type obstacle.

[0068] Step C2.3, according to the first space-time occupied area and the expected speed, the second space-time occupied area, the second decision result and the preset motion cost function, obtain the optimal solution of the motion cost function. Exemplarily, it can be implemented with reference to the following steps (1) to (4):

[0069] (1) Determine the first generation value of the cost item corresponding to the target non-collision obstacle based on the first space-time occupied area and the expected speed.

[0070] In some specific implementation examples, the first generation value C of the cost item corresponding to the target non-collision obstacle can be determined according to the following formula: scene (s):

[0071]

[0072] Among them, s is the planned waypoint of the vehicle in the ST space-time graph, t is the time it takes for the vehicle to move to the planned waypoint, and E scene is the first space-time occupied area, s' is the first derivative of s with respect to time, and is also the speed of the vehicle at point s; v scene is the expected speed. The above is the cost item corresponding to the pre-set non-collision obstacle. Substituting the speed of the vehicle at point s into the cost item, the first generation value can be obtained.

[0073] It can be seen from the above formula that if the vehicle does not enter the first space-time occupied area, the first cost value is zero, and if the vehicle enters the first space-time occupied area, the greater the difference between the vehicle speed and the expected speed, the higher the first generation value, the smaller the difference between the vehicle speed and the expected speed, the lower the first generation value, and when the vehicle speed is equal to the expected speed, the first cost value is zero. Through the constraints of the cost items corresponding to the above non-collision obstacles, in order to make the first generation value as small as possible, the vehicle can be prevented from entering the first space-time occupied area or enter the first space-time occupied area at a speed as close to the expected speed as possible during the motion planning process, thereby minimizing the impact of non-collision obstacles on the vehicle during the motion planning process.

[0074] (2) Determine the second generation value of the cost item corresponding to the target collision obstacle based on the second space-time occupied area and the second decision result.

[0075] The embodiments of the present disclosure do not limit the specific form of the cost item corresponding to the collision type obstacle. The cost item corresponding to the collision type obstacle used in any related technology can be used, or new cost items corresponding to the collision type obstacles can be additionally created according to needs, which is not limited here.

[0076] (3) Obtaining the motion cost function value of the vehicle according to the first generation value and the second generation value. In some specific implementation examples, the motion cost function value may be determined according to the weighted sum of the first generation value and the second cost value. In other specific implementation examples, other costs may be introduced, such as a road network speed limit cost value, a smoothness cost value, etc.

[0077] In some specific implementation examples, when setting the motion cost function, various types of influencing factors such as obstacles, road network speed limits, and vehicle ride comfort can be comprehensively considered, and these influencing factors can be quantitatively measured. Generally, different types of influences need to be abstracted into specific and calculable cost items, so as to transform the task of finding the best planning result into finding the planning result with the minimum cost function value. The present disclosure embodiment provides a specific example of a motion cost function, which can be implemented with reference to the following formula:

[0078] C total (s)=w1C obj (s)+w2 C max_vel (s)+w3C smooth (s)+w4C scene (s)

[0079] Among them, C total (s) is the total motion cost function, C obj (s) is the cost item corresponding to the collision obstacle, C max_vel (s) is the cost item corresponding to the speed limit of the road network, C smooth(s) is the cost term corresponding to the vehicle's smooth motion, C scene (s) is the cost item corresponding to the non-collision obstacle, w1~w4 are the weight coefficients corresponding to each cost item, and the above total motion cost function specifically represents the cost of the planning step when the vehicle plans to move to the corresponding waypoint s at a certain time t.

[0080] (4) The motion parameter corresponding to the minimum motion cost function value is taken as the optimal solution of the motion cost function.

[0081] In practical applications, multiple groups of motion parameters can be tried in advance and substituted into the motion cost function respectively to search for the optimal motion parameters (the corresponding motion cost function value is the smallest). Specifically, multiple candidate motion planning paths can be set in advance, each candidate motion planning path corresponds to a group of motion parameters, and then the motion cost function is used to evaluate each candidate motion planning path to obtain the total motion cost value of each candidate motion planning path, and then the motion planning path with the smallest total motion cost value is used as the optimal motion planning path, and the corresponding motion parameters are used as the optimal solution of the motion cost function.

[0082] In practical applications, by simultaneously mapping the space-time occupied areas of collision obstacles and non-collision obstacles in the same ST space-time graph, it is convenient to directly perform space-time search in the ST space-time graph, and finally find the optimal motion planning path according to the constraints of the cost function.

[0083] In summary, the vehicle motion planning method provided by the embodiment of the present disclosure performs motion planning for both non-collision obstacles and collision obstacles based on the same motion cost function, and obtains the vehicle motion trajectory according to the optimal solution of the function, which can effectively improve the rationality and reliability of the motion planning results.

[0084] To facilitate understanding of the above content provided in the embodiment of the present disclosure, the embodiment of the present disclosure provides a specific application example applied to the above vehicle planning method, including the following three main links:

[0085] 1. Scene Recognition

[0086] During the driving process, the vehicle can target the scene contained in the current environment (such as within a designated area centered on the vehicle). In some specific implementation examples, the scene can be determined by identifying obstacles in the current environment, such as scene feature extraction and scene matching through the environmental detection data of the current environment collected by the vehicle. First, the scene features contained in the current environment are extracted, and the scene features are combined with the road network and environmental perception information to match the target scene to be processed, so as to determine whether the target scene to be processed does exist in the current scene, and then the scene information of the target scene can be output. The above-mentioned target scene can be, for example, a parallel large vehicle scene, a junction vehicle scene, etc., which are not limited here. In some specific implementation examples, each obstacle can be regarded as a sub-scene. The scene information may include: the waypoint scene_start_s where the scene starts to act on the vehicle in the Frenet coordinate system based on the vehicle path, the time scene_start_t when the scene starts, the time scene_end_t when the scene ends, the length scene_occupy_dist occupied by the scene on the vehicle path, the scene movement speed scene_vel, the scene type scene_type corresponding to the scene, etc. The predicted movement trajectory of each obstacle can also be reflected through the above scene information.

[0087] For easier understanding, see Figure 2 The diagram of a scene recognition result shown in FIG. 1 illustrates the obstacles identified after scene recognition, where the black arrow ego_car represents the ego vehicle, i.e., the vehicle that needs to be motion planned; the dark gray arrow obj_car_1 represents the vehicle that is about to merge into the ego vehicle's lane; the dark gray arrow obj_car_2 represents the vehicle in front of the ego vehicle that is blocking the ego vehicle's travel; the light gray solid arrow long_vehicle represents the parallel large vehicle traveling in the right lane of the ego vehicle's lane; the light gray dotted arrow sense_1 represents the impact of the parallel large vehicle on the ego vehicle projected onto the ego vehicle's lane, where obj_car_1 and obj_car_2 are collision obstacles, and long_vehicle is a non-collision obstacle. In summary, Figure 2 The scenario shown can be summarized as follows: the vehicle is traveling straight, there is a dangerous large parallel vehicle long_vehicle traveling in the same direction in the right lane, an obstacle vehicle obj_car_1 on the left is about to merge into the vehicle's lane, and there is an obstacle vehicle obj_car_2 in front of the merging intersection in front of the vehicle that blocks the vehicle's travel.

[0088] Furthermore, the disclosed embodiment can determine, based on the identified obstacles, that obj_car_1 and obj_car_2 are collision-type obstacles (there is at least partial path overlap between the predicted motion trajectory and the planned path of the vehicle), and long_vehicle is a non-collision-type obstacle (there is no path overlap between the predicted motion trajectory and the planned path of the vehicle). Moreover, it can be further identified whether the parallel cart long_vehicle is the target non-collision-type obstacle to be processed. Exemplarily, it can be determined based on the predicted motion trajectory of the parallel cart whether the closest distance between the parallel cart and the vehicle is less than a preset distance threshold (also referred to as a safety distance threshold). If it is greater, the parallel cart is treated as a safety obstacle and is no longer processed. If it is less, the estimated parallel time between the vehicle and the parallel cart is calculated. The parallel time T_p between the vehicle and the dangerous vehicle can be preliminarily calculated when the vehicle is traveling at a constant speed at the current vehicle speed (v_ego). For ease of understanding, it can be combined with Figure 3 The schematic diagram of parallel time calculation of a dangerous parallel vehicle is shown in FIG. Figure 3 The black slash in the figure represents the trajectory of the vehicle, and the gray area is the space-time occupied area of ​​the parallel vehicle mapped in the ST space-time diagram. Assuming that the distance between the rear of the parallel vehicle and the vehicle is s1, the distance between the front of the parallel vehicle and the vehicle is s2, and the speed of the parallel vehicle is v_obj, then T_p can be expressed as:

[0089] T_p=t2-t1=(s2-s1) / (v_ego-v_obj).

[0090] Assuming that the preset time threshold is T_p_thresh, if T_p>T_p_thresh, the parallel vehicles are target non-collision obstacles that need to be processed, and the scenario corresponding to the parallel vehicles is a non-collision scenario that needs to be considered in decision-making and planning.

[0091] (II) Obstacle Decision

[0092] The disclosed embodiment can convert obstacles into space-time occupied individuals of the path corresponding to the vehicle, and then make decisions on the space-time occupied individuals in space and time. For collision-type obstacles that have a space-time occupied relationship with the vehicle, they can be directly converted into space-time occupied individuals, and for non-collision-type obstacles that have no space-time occupied relationship with the vehicle, they can also be converted into virtual space-time occupied individuals. Specifically, the predicted motion trajectory of the non-collision-type obstacle is projected onto the planned path of the vehicle, thereby forming the space-time occupation of the path of the vehicle, so that the collision-type obstacle and the non-collision-type obstacle can participate in the overall decision-making and planning at the same time. However, the difference is that the space-time area of ​​the collision-type obstacle does not allow the vehicle to enter (it can be constrained by the motion cost. For example, if entering will result in infinite motion cost), while the space-time area of ​​the non-collision-type obstacle allows the vehicle to enter under certain conditions, such as, assuming that the total motion cost value obtained after the vehicle enters the space-time area of ​​the non-collision-type obstacle is the smallest in combination with other factors, then the space-time area can be entered.

[0093] Before making an obstacle decision (also called a scene decision), each obstacle can be first abstracted into a data structure of a specified type. The data structure of non-collision obstacles can be unified with the data structure of collision obstacles, which is more convenient for subsequent unified analysis and processing of non-collision obstacles and collision obstacles. Exemplarily, the impact of non-collision obstacles on the vehicle can be abstracted into a time-space occupancy data structure based on the predicted motion trajectory of the non-collision obstacles. The exemplary data structure may include the following data: the time when the non-collision scene (the scene corresponding to the non-collision obstacle) starts to act; the time when the non-collision scene ends; the position where the non-collision scene starts on the path of the vehicle; the speed at which the non-collision scene moves forward longitudinally along the path of the vehicle; the length of the non-collision scene affecting the path of the vehicle at a single time point; the sequence of lanes affected by the non-collision scene; the specific scene classification corresponding to the non-collision scene; the decision result corresponding to the non-collision scene; the scene expected speed, which can be used to calculate the scene expected speed based on the scene movement speed, the scene decision result and the corresponding scene configuration parameters when planning the non-collision scene. It should be noted that the time-space occupancy data structure abstracted from non-collision obstacles and collision obstacles can be consistent, but there is no scene expected speed in the time-space occupancy data of collision obstacles. The above are all exemplary descriptions and should not be regarded as limiting.

[0094] After knowing the types of obstacles, decisions can be made based on the vehicle description information (vehicle speed, vehicle size, etc.) and scene description information and obstacle information such as the data structure corresponding to each obstacle. Specifically, each obstacle is taken as an object to be decided, and the two types of objects to be decided are decided in sequence according to the order of the time and distance of space-time occupation, and the decision results of each object to be decided can be obtained (such as whether the vehicle should rush or give way to each obstacle).

[0095] In the aforementioned Figure 2 On the basis of Figure 4 A decision diagram is shown, in which the decision results for each obstacle are indicated by black bold arrows, such as requiring the ego_car to give way to sence_1, obj_car_1 and obj_car_2 in the current environment.

[0096] After the decision result is obtained, the decision result can be input into the subsequent longitudinal planning module for subsequent motion planning.

[0097] (III) Movement Planning

[0098] After the decision-making stage, collision obstacles and non-collision obstacles can be mapped to the same ST space-time diagram. Due to the particularity of non-collision obstacles, the space-time occupied area corresponding to them can allow vehicles to enter under certain circumstances. Therefore, the expected speed of the vehicle entering the space-time occupied area of ​​the non-collision obstacles can be obtained. The space-time occupied area corresponding to the non-collision obstacles in the ST space-time diagram can be associated with the expected speed. Finally, spatial search can be performed based on the ST space-time diagram, and the best planning result can be obtained through cost function design and optimization solution.

[0099] On the basis of the above, we can further refer to Figure 5 A motion planning schematic diagram is shown, the curve in the ST space-time diagram is the motion planning path of the vehicle, the dark area represents the space-time occupied area of ​​the collision type obstacle, and the light area represents the space-time occupied area of ​​the non-collision type obstacle. It can be understood that when performing vehicle motion planning, it is necessary to avoid the motion planning path of the vehicle passing through the space-time occupied area of ​​the collision type obstacle, that is, to avoid the vehicle from entering the space-time occupied area of ​​the collision type obstacle during driving; although the space-time occupied area of ​​the non-collision type obstacle allows the vehicle to enter, the cost item of the non-collision type obstacle will drive the vehicle to move at a speed close to the expected speed in the space-time occupied area of ​​the non-collision type obstacle, or prevent the vehicle from entering the space-time occupied area of ​​the non-collision type obstacle. In the above manner, the impact of non-collision type obstacles on the vehicle can be minimized.

[0100] On the basis of the above, for ease of understanding, the present disclosure also provides a Figure 6 The motion planning diagram shown in FIG. 1 can more vividly and intuitively combine the actual scene with the ST space-time diagram. Figure 6 To elaborate in detail:

[0101] The figure shows the vehicle ego_car whose path is to be planned, the collision obstacle obj_car_1 that is about to merge into the lane of the vehicle, the non-collision obstacle obj_car_2 that is running in parallel in the lane beside the vehicle, and the collision obstacle obj_car_3 that is about to enter the intersection from another lane. The space-time occupied area of ​​each obstacle to the vehicle is mapped into the ST diagram. Figure 6 In the ST diagram shown, the horizontal axis is the distance axis starting from the front of the vehicle, and the vertical axis is the time axis starting from the current moment. The filled area in the ST diagram can reflect the occupation of the corresponding vehicle obstacle in the space-time space of the vehicle. The longitudinal planning curve of the vehicle should avoid passing through the area occupied by the collision-type obstacle, so as to avoid arriving at the same position at the same time as the collision-type obstacle to avoid collision.

[0102] exist Figure 6 In the ST diagram, the longitudinal planning curve of the vehicle is indicated by the black oblique line. The inverse of the slope of each point on the black oblique line corresponds to the point speed planned by the vehicle at the corresponding position. Figure 6 If the obstacle is a straight line, it means that the vehicle will keep driving at a constant speed in the future. The above is just a simple example. In actual applications, if the vehicle is not driving at a constant speed, it will be indicated by a curve. According to the obstacle prediction trajectory, the obstacle car obj_car1 will converge on the path of the vehicle in the future. The starting time of its occupation of the planned path of the vehicle is about 4 seconds in the future. At this time, it occupies the interval of 25 to 30 meters of the planned path of the vehicle (the length of the vehicle has been considered). At this time, the vehicle has traveled to 38 meters and has achieved the intersection of obj_car1. The obstacle car obj_car2 is located in the parallel lane of the vehicle, and according to its predicted trajectory, it will continue to drive in front of the planned path of the vehicle in the future, and its speed is lower than that of the vehicle. In the future, the distance between the vehicle and the obstacle car will gradually narrow. The obstacle car obj_car3 will intersect the planned path of this vehicle in the future. According to the predicted trajectory of the obstacle car obj_car3, it can be known that the car will start to occupy the planned path of this vehicle in about 1 second in the future. At this time, it will occupy the interval of 49 to 54 meters of the planned path of this vehicle. According to its speed and relative position, it will end occupying the planned path of this vehicle in about 3 seconds in the future. According to the results of the uniform speed planning of this vehicle, when this vehicle passes through the 49 to 54 meter interval of the planned path, the corresponding time is 5 to 5.8 seconds. Therefore, the current planning result of this vehicle can achieve avoidance of this vehicle.

[0103] In summary, the vehicle motion planning method provided by the embodiment of the present disclosure can be well applied to the field of autonomous driving that requires vehicle motion planning. Non-collision obstacles and collision obstacles are uniformly motion planned based on the same motion cost function, and the vehicle motion trajectory is obtained according to the optimal solution of the function, which can effectively improve the rationality and reliability of the motion planning results. Furthermore, compared with the related art, the embodiment of the present disclosure has the following characteristics:

[0104] 1) The related art processes non-collision obstacles by directly ignoring, switching methods or parameters, limiting vehicle speed, etc., which has poor reliability and safety. The embodiment of the present disclosure abstracts non-collision obstacles into data structures similar to collision obstacles, and uniformly transmits them to the decision-making and planning layers for processing, thereby effectively improving processing efficiency and processing reliability.

[0105] 2) When making decisions, the related technology cannot put the two types of obstacles into the same decision-making framework, and the separate decision-making scheme cannot guarantee the rationality of the final decision result. The disclosed embodiment makes decisions based on the type of obstacles, and the order of the time and distance of the two types of obstacles can be comprehensively considered in the decision-making, and the mutual influence of the decision results of the two types of obstacles is considered to ensure the rationality of the decision, which helps to further ensure the rationality of subsequent motion planning.

[0106] 3) When planning, the related technology only considers the cost of collision-type obstacles to the vehicle in the motion cost function, while the disclosed embodiment can introduce the cost of non-collision obstacles to the vehicle into the motion cost function, and perform strong spatiotemporal constraints on the vehicle trajectory through the spatiotemporal occupancy of collision-type obstacles, and project the impact of non-collision obstacles on the vehicle onto the planned path of the vehicle, so that the spatiotemporal occupancy of non-collision obstacles can perform soft spatiotemporal constraints on the vehicle trajectory. The two types of obstacles can affect the behavior of the vehicle at the same time. During motion planning, a reasonable spatiotemporal search can be uniformly performed through the same ST spatiotemporal graph, thereby obtaining a more reasonable and reliable motion trajectory.

[0107] Corresponding to the above-mentioned vehicle motion planning method, the embodiment of the present disclosure provides a vehicle motion planning device, Figure 7 FIG. 1 is a schematic diagram of a vehicle motion planning device provided in an embodiment of the present disclosure. The device can be implemented by software and / or hardware and can generally be integrated into an electronic device, such as Figure 7 As shown, including:

[0108] The obstacle trajectory acquisition module 702 is used to obtain the predicted motion trajectory of each obstacle of the vehicle in the current environment;

[0109] The obstacle type determination module 704 is used to determine the type of each obstacle based on the predicted motion trajectory; wherein the types include collision type obstacles and non-collision type obstacles, the predicted motion trajectory of the collision type obstacle has at least a partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision type obstacle has no path overlap with the planned path of the vehicle;

[0110] The function optimal solution acquisition module 706 is used to acquire the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and the preset motion cost function; wherein the motion cost function includes the cost item related to the collision type obstacle and the cost item related to the non-collision type obstacle;

[0111] The vehicle trajectory acquisition module 708 is used to obtain the current motion trajectory of the vehicle based on the optimal solution of the motion cost function.

[0112] The above device performs motion planning for non-collision obstacles and collision obstacles based on the same motion cost function, and obtains the vehicle motion trajectory according to the optimal solution of the function, which can effectively improve the rationality and reliability of the motion planning results.

[0113] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: when the obstacles include non-collision obstacles, select a target non-collision obstacle from the non-collision obstacles, and obtain first space-time occupancy information corresponding to the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle; when the obstacles include collision obstacles, take each of the collision obstacles as a target collision obstacle, and obtain second space-time occupancy information corresponding to the target collision obstacle based on a second predicted motion trajectory corresponding to the target collision obstacle; and obtain the optimal solution of the motion cost function based on the first space-time occupancy information, the second space-time occupancy information and a preset motion cost function.

[0114] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: when the non-collision obstacle includes a parallel moving body, estimate the shortest distance between the parallel moving body and the vehicle according to the first predicted motion trajectory corresponding to the parallel moving body; when the shortest distance is less than a preset distance threshold, obtain the expected parallel time between the parallel moving body and the vehicle; when the expected parallel time is greater than the preset time threshold, select the parallel moving body as the target non-collision obstacle.

[0115] In some implementations, the function optimal solution acquisition module 706 is specifically used to: when the non-collision obstacle includes an immovable body, determine whether the immovable body has a tendency to affect the movement of the vehicle; if so, select the immovable body as a target non-collision obstacle.

[0116] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: project the first predicted motion trajectory corresponding to the target non-collision obstacle onto the planned path of the vehicle, and determine the first spatiotemporal occupancy information of the target non-collision obstacle on the planned path of the vehicle based on the projection result.

[0117] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: determine a first decision result corresponding to the non-collision obstacle and a second decision result corresponding to the collision type obstacle based on the first space-time occupancy information and the second space-time occupancy information; and obtain the optimal solution of the motion cost function based on the first decision result, the second decision result, the first space-time occupancy information, the second space-time occupancy information and a preset motion cost function.

[0118] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: determine the first space-time occupied area of ​​the target non-collision obstacle in the ST space-time diagram based on the first space-time occupancy information, and determine the second space-time occupied area of ​​the target collision obstacle in the ST space-time diagram based on the second space-time occupancy information; obtain the expected speed of the vehicle in the first space-time occupied area according to the first decision result; obtain the optimal solution of the motion cost function according to the first space-time occupied area and the expected speed, the second space-time occupied area, the second decision result and a preset motion cost function.

[0119] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: determine the current speed of the target non-collision obstacle based on the first predicted motion trajectory corresponding to the target non-collision obstacle; determine the escape speed according to the current speed of the vehicle and the current speed of the target non-collision obstacle; determine the expected speed of the vehicle in the first space-time occupied area according to the first decision result, the current speed of the target non-collision obstacle and the escape speed.

[0120] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: when the first decision result indicates that the vehicle needs to give way to the target non-collision obstacle, let the difference between the current speed of the target non-collision obstacle and the escape speed be the expected speed of the vehicle in the first time and space occupied area; when the first decision result indicates that the vehicle needs to overtake the target non-collision obstacle, let the sum of the current speed of the target non-collision obstacle and the escape speed be the expected speed of the vehicle in the first time and space occupied area.

[0121] In some embodiments, the function optimal solution acquisition module 706 is specifically used to: determine the first generation value of the cost item corresponding to the target non-collision obstacle based on the first space-time occupied area and the expected speed; determine the second generation value of the cost item corresponding to the collision obstacle based on the second space-time occupied area and the second decision result; obtain the motion cost function value of the vehicle based on the first generation value and the second generation value; and use the motion parameter corresponding to the minimum motion cost function value as the optimal solution of the motion cost function.

[0122] In some implementations, the function optimal solution acquisition module 706 is specifically used to determine the first generation value C of the cost item corresponding to the target non-collision obstacle according to the following formula: scene (s):

[0123]

[0124] Where s is the planned waypoint of the vehicle in the ST space-time graph, t is the time it takes for the vehicle to move to the planned waypoint, and E scene is the area occupied by the first space-time, s' is the first-order derivative of s with respect to time; v scene is the desired speed.

[0125] The vehicle motion planning device provided in the embodiments of the present disclosure can execute the vehicle motion planning method provided in any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0126] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device embodiment can refer to the corresponding process in the method embodiment, and will not be repeated here.

[0127] An embodiment of the present disclosure also provides an electronic device, which includes: a processor; a memory for storing instructions executable by the processor; and a processor for reading executable instructions from the memory and executing the instructions to implement any of the above-mentioned vehicle motion planning methods.

[0128] Figure 8 The structure diagram of an electronic device provided by the embodiment of the present disclosure is shown in FIG. Figure 8 As shown, the electronic device 800 includes one or more processors 801 and a memory 802 .

[0129] The processor 801 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 800 to perform desired functions.

[0130] The memory 802 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 801 may run the program instructions to implement the vehicle motion planning method of the embodiment of the present disclosure described above and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.

[0131] In one example, the electronic device 800 may further include: an input device 803 and an output device 804, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0132] In addition, the input device 803 may also include, for example, a keyboard, a mouse, and the like.

[0133] The output device 804 can output various information to the outside, including the determined distance information, direction information, etc. The output device 804 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0134] Of course, to simplify, Figure 8 Only some of the components related to the present disclosure in the electronic device 800 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 800 may also include any other appropriate components.

[0135] In addition to the above-mentioned method and device, the embodiment of the present disclosure may also be a computer program product, which includes computer program instructions, and when the computer program instructions are executed by a processor, the processor executes the vehicle motion planning method provided by the embodiment of the present disclosure.

[0136] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0137] In addition, the embodiment of the present disclosure may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the vehicle motion planning method provided by the embodiment of the present disclosure.

[0138] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0139] The embodiment of the present disclosure also provides a computer program product, including a computer program / instruction, which implements the vehicle motion planning method in the embodiment of the present disclosure when the computer program / instruction is executed by a processor.

[0140] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0141] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0142] In summary, the vehicle motion planning method provided in the embodiments of the present disclosure can be implemented as follows:

[0143] A1. A vehicle motion planning method, comprising:

[0144] Obtain the predicted motion trajectory of the vehicle in the current environment;

[0145] Determining the type of each obstacle based on the predicted motion trajectory; wherein the types include collision-type obstacles and non-collision-type obstacles, the predicted motion trajectory of the collision-type obstacle has at least a partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision-type obstacle has no path overlap with the planned path of the vehicle;

[0146] According to the type of each obstacle, the predicted motion trajectory and the preset motion cost function, an optimal solution of the motion cost function is obtained; wherein the motion cost function includes a cost term related to the collision type obstacle and a cost term related to the non-collision type obstacle;

[0147] The current motion trajectory of the vehicle is obtained based on the optimal solution of the motion cost function.

[0148] A2. According to the method described in A1, the step of obtaining the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and the preset motion cost function includes:

[0149] In the case where the obstacles include non-collision obstacles, selecting a target non-collision obstacle from the non-collision obstacles, and acquiring first spatiotemporal occupancy information corresponding to the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle;

[0150] In the case where the obstacles include collision-type obstacles, each of the collision-type obstacles is used as a target collision-type obstacle, and second spatiotemporal occupancy information corresponding to the target collision-type obstacle is acquired based on a second predicted motion trajectory corresponding to the target collision-type obstacle;

[0151] An optimal solution of the motion cost function is obtained according to the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function.

[0152] A3. According to the method described in A2, the step of selecting a target non-collision obstacle from the non-collision obstacles comprises:

[0153] When the non-collision obstacle includes a parallel moving body, estimating the shortest distance between the parallel moving body and the vehicle according to a first predicted motion trajectory corresponding to the parallel moving body;

[0154] When the closest distance is less than a preset distance threshold, obtaining an estimated parallel time between the parallel moving object and the vehicle;

[0155] When the estimated parallel duration is greater than a preset duration threshold, the parallel moving body is selected as a target non-collision obstacle.

[0156] A4. According to the method described in A2, the step of selecting a target non-collision obstacle from the non-collision obstacles comprises:

[0157] When the non-collision obstacle includes an immovable object, determining whether the immovable object has a tendency to affect the movement of the vehicle;

[0158] If so, the non-moving object is selected as the target non-collision obstacle.

[0159] A5. According to the method described in A2, the step of acquiring first spatiotemporal occupancy information corresponding to the target non-collision obstacle based on the first predicted motion trajectory corresponding to the target non-collision obstacle comprises:

[0160] A first predicted motion trajectory corresponding to the target non-collision obstacle is projected onto a planned path of the vehicle, and first spatiotemporal occupancy information of the target non-collision obstacle on the planned path of the vehicle is determined based on the projection result.

[0161] A6. The method according to A2, wherein the step of obtaining an optimal solution of the motion cost function according to the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function comprises:

[0162] Determining a first decision result corresponding to the non-collision obstacle and a second decision result corresponding to the collision obstacle according to the first space-time occupancy information and the second space-time occupancy information;

[0163] An optimal solution of the motion cost function is obtained according to the first decision result, the second decision result, the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function.

[0164] A7. The method according to A6, wherein the step of obtaining an optimal solution of the motion cost function according to the first decision result, the second decision result, the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function comprises:

[0165] Determine a first space-time occupied area of ​​the target non-collision type obstacle in the ST space-time diagram based on the first space-time occupied information, and determine a second space-time occupied area of ​​the target collision type obstacle in the ST space-time diagram based on the second space-time occupied information;

[0166] Acquire the expected speed of the vehicle in the first space-time occupied area according to the first decision result;

[0167] An optimal solution of the motion cost function is obtained according to the first space-time occupied area and the expected speed, the second space-time occupied area, the second decision result and a preset motion cost function.

[0168] A8. According to the method described in A7, the step of obtaining the expected speed of the vehicle in the first space-time occupied area according to the first decision result comprises:

[0169] Determining a current speed of the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle;

[0170] Determining an escape velocity based on a current velocity of the vehicle and a current velocity of the target non-collision obstacle;

[0171] An expected speed of the vehicle within the first space-time occupied area is determined according to the first decision result, the current speed of the target non-collision obstacle, and the escape speed.

[0172] A9. The method according to A8, wherein the step of determining the expected speed of the vehicle within the first space-time occupied area according to the first decision result, the current speed of the target non-collision obstacle and the escape speed comprises:

[0173] When the first decision result indicates that the vehicle needs to give way to the target non-collision obstacle, the difference between the current speed of the target non-collision obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area;

[0174] When the first decision result indicates that the vehicle needs to overtake the target non-collision obstacle, the sum of the current speed of the target non-collision obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area.

[0175] A10. The method according to A7, wherein the step of obtaining an optimal solution of the motion cost function according to the first spatiotemporal occupied area and the expected speed, the second spatiotemporal occupied area, the second decision result and a preset motion cost function comprises:

[0176] Determining a first generation value of a cost item corresponding to the target non-collision obstacle according to the first space-time occupied area and the expected speed;

[0177] Determining a second generation value of the cost item corresponding to the target collision obstacle according to the second space-time occupied area and the second decision result;

[0178] Obtaining a motion cost function value of the vehicle according to the first generation value and the second generation value;

[0179] The motion parameter corresponding to the minimum motion cost function value is taken as the optimal solution of the motion cost function.

[0180] A11. According to the method described in A10, the step of determining the first generation value of the cost item corresponding to the target non-collision obstacle according to the first space-time occupied area and the expected speed includes:

[0181] Determine the first generation value C of the cost item corresponding to the target non-collision obstacle according to the following formula: scene (s):

[0182]

[0183] Where s is the planned waypoint of the vehicle in the ST space-time graph, t is the time it takes for the vehicle to move to the planned waypoint, and E scene is the area occupied by the first space-time, s' is the first-order derivative of s with respect to time; v scene is the desired speed.

Claims

1. A vehicle motion planning method, characterized in that: include: Obtain the predicted motion trajectory of the vehicle in the current environment; Determining the type of each obstacle based on the predicted motion trajectory; wherein the types include collision-type obstacles and non-collision-type obstacles, the predicted motion trajectory of the collision-type obstacle has at least a partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision-type obstacle has no path overlap with the planned path of the vehicle; According to the type of each obstacle, the predicted motion trajectory and the preset motion cost function, an optimal solution of the motion cost function is obtained; wherein the motion cost function includes a cost term related to the collision type obstacle and a cost term related to the non-collision type obstacle; Obtaining a current motion trajectory of the vehicle based on an optimal solution of the motion cost function; The step of obtaining the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and the preset motion cost function includes: In the case where the obstacles include non-collision obstacles, selecting a target non-collision obstacle from the non-collision obstacles, and acquiring first spatiotemporal occupancy information corresponding to the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle; In the case where the obstacles include collision-type obstacles, each of the collision-type obstacles is used as a target collision-type obstacle, and second spatiotemporal occupancy information corresponding to the target collision-type obstacle is acquired based on a second predicted motion trajectory corresponding to the target collision-type obstacle; An optimal solution of the motion cost function is obtained according to the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function.

2. The method according to claim 1, characterized in that The step of selecting a target non-collision obstacle from the non-collision obstacles comprises: When the non-collision obstacle includes a parallel moving body, estimating the shortest distance between the parallel moving body and the vehicle according to a first predicted motion trajectory corresponding to the parallel moving body; When the closest distance is less than a preset distance threshold, obtaining an estimated parallel time between the parallel moving object and the vehicle; When the estimated parallel duration is greater than a preset duration threshold, the parallel moving body is selected as a target non-collision obstacle.

3. The method according to claim 1, characterized in that The step of selecting a target non-collision obstacle from the non-collision obstacles comprises: When the non-collision obstacle includes an immovable object, determining whether the immovable object has a tendency to affect the movement of the vehicle; If so, the non-moving object is selected as the target non-collision obstacle.

4. The method according to claim 1, characterized in that: The step of acquiring first spatiotemporal occupancy information corresponding to the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle comprises: A first predicted motion trajectory corresponding to the target non-collision obstacle is projected onto a planned path of the vehicle, and first spatiotemporal occupancy information of the target non-collision obstacle on the planned path of the vehicle is determined based on the projection result.

5. The method according to claim 1, characterized in that The step of obtaining an optimal solution of the motion cost function according to the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function comprises: Determining a first decision result corresponding to the non-collision obstacle and a second decision result corresponding to the collision obstacle according to the first space-time occupancy information and the second space-time occupancy information; An optimal solution of the motion cost function is obtained according to the first decision result, the second decision result, the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function.

6. The method according to claim 5, characterized in that The step of obtaining an optimal solution of the motion cost function according to the first decision result, the second decision result, the first spatiotemporal occupancy information, the second spatiotemporal occupancy information and a preset motion cost function comprises: Determine a first space-time occupied area of ​​the target non-collision type obstacle in the ST space-time diagram based on the first space-time occupied information, and determine a second space-time occupied area of ​​the target collision type obstacle in the ST space-time diagram based on the second space-time occupied information; Acquire the expected speed of the vehicle in the first space-time occupied area according to the first decision result; An optimal solution of the motion cost function is obtained according to the first space-time occupied area and the expected speed, the second space-time occupied area, the second decision result and a preset motion cost function.

7. The method according to claim 6, characterized in that The step of obtaining the expected speed of the vehicle in the first space-time occupied area according to the first decision result includes: Determining a current speed of the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle; Determining an escape velocity based on a current velocity of the vehicle and a current velocity of the target non-collision obstacle; An expected speed of the vehicle within the first space-time occupied area is determined according to the first decision result, the current speed of the target non-collision obstacle, and the escape speed.

8. The method according to claim 7, characterized in that The step of determining the expected speed of the vehicle in the first space-time occupied area according to the first decision result, the current speed of the target non-collision obstacle and the escape speed comprises: When the first decision result indicates that the vehicle needs to give way to the target non-collision obstacle, the difference between the current speed of the target non-collision obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area; When the first decision result indicates that the vehicle needs to overtake the target non-collision obstacle, the sum of the current speed of the target non-collision obstacle and the escape speed is used as the expected speed of the vehicle in the first space-time occupied area.

9. The method according to claim 6, characterized in that The step of obtaining an optimal solution of the motion cost function according to the first space-time occupied area and the expected speed, the second space-time occupied area, the second decision result and a preset motion cost function comprises: Determining a first generation value of a cost item corresponding to the target non-collision obstacle according to the first space-time occupied area and the expected speed; Determining a second generation value of the cost item corresponding to the target collision obstacle according to the second space-time occupied area and the second decision result; Obtaining a motion cost function value of the vehicle according to the first generation value and the second generation value; The motion parameter corresponding to the minimum motion cost function value is taken as the optimal solution of the motion cost function.

10. The method according to claim 9, characterized in that The step of determining a first generation value of the cost item corresponding to the target non-collision obstacle according to the first space-time occupied area and the expected speed comprises: The first generation value C of the cost item corresponding to the target non-collision obstacle is determined according to the following formula: scene (s): Where s is the planned waypoint of the vehicle in the ST space-time graph, t is the time it takes for the vehicle to move to the planned waypoint, and E scene is the area occupied by the first space-time, s' is the first-order derivative of s with respect to time; v scene is the desired speed.

11. A vehicle motion planning device, characterized in that: include: The obstacle trajectory acquisition module is used to obtain the predicted motion trajectory of each obstacle in the current environment; an obstacle type determination module, configured to determine the type of each obstacle based on the predicted motion trajectory; wherein the types include collision-type obstacles and non-collision-type obstacles, the predicted motion trajectory of the collision-type obstacle having at least a partial path overlap with the planned path of the vehicle, and the predicted motion trajectory of the non-collision-type obstacle having no path overlap with the planned path of the vehicle; A function optimal solution acquisition module, used to acquire the optimal solution of the motion cost function according to the type of each obstacle, the predicted motion trajectory and the preset motion cost function; wherein the motion cost function includes the cost term related to the collision type obstacle and the cost term related to the non-collision type obstacle; A vehicle trajectory acquisition module, used to obtain the current motion trajectory of the vehicle based on the optimal solution of the motion cost function; Among them, the function optimal solution acquisition module is specifically used for: when the obstacles include non-collision obstacles, selecting a target non-collision obstacle from the non-collision obstacles, and acquiring first space-time occupancy information corresponding to the target non-collision obstacle based on a first predicted motion trajectory corresponding to the target non-collision obstacle; when the obstacles include collision obstacles, taking each of the collision obstacles as a target collision obstacle, and acquiring second space-time occupancy information corresponding to the target collision obstacle based on a second predicted motion trajectory corresponding to the target collision obstacle; and acquiring the optimal solution of the motion cost function according to the first space-time occupancy information, the second space-time occupancy information and a preset motion cost function.

12. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the vehicle motion planning method described in any one of claims 1-10 above.

13. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the vehicle motion planning method described in any one of claims 1-10.

Images