Vehicle path planning method and path planning system based on recursive sampling

Abstract

The invention provides a vehicle path planning method and system based on recursive sampling, and the method comprises the steps: determining an initial sampling boundary under an own vehicle coordinate system; according to the initial sampling boundary and vehicle parameters of the target vehicle, processing the initial sampling boundary in combination with a first sampling requirement to obtain a first sampling queue; according to the first sampling queue, an initial planning set is obtained in combination with the second sampling requirement, and the initial planning set comprises a plurality of initial planning end points; and performing cost calculation on all initial planning end points in the initial planning set to obtain an optimal planning end point. According to the method and the device, the technical problem that in the process of executing trajectory planning by the intelligent vehicle in an unknown environment, trajectory point sampling is random, and rationality is lacked is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle path planning. Specifically, the present invention relates to a vehicle path planning method based on recursive sampling and a vehicle path planning system based on recursive sampling. Background Art

[0002] With the continuous improvement of the intelligent driving intelligence requirements, the intelligent parking function needs to explore forward and backward to help the intelligent vehicle find a suitable empty parking space before finding the parking space. Currently, the relatively common one is the memory parking (Home-zone Parking Assist, HPA) function. This function requires the intelligent vehicle to use the SLAM technology to record the parking lot map in advance and store it locally. When the user needs automatic parking subsequently, the corresponding parking lot map can be selected, but it cannot cope with the unknown parking lot environment; while other exploration methods in unknown environments, such as the Rapidly-exploring Random Tree (RRT), require a clear end point and sample randomly; the Dynamic Window Approach (DWA) needs to sample multiple regular trajectories according to requirements, and needs to evaluate and score all trajectories to select the optimal one; its proximal sampling is dense, but the distal sampling is sparse, and it may not be able to sample a complete collision-free trajectory when encountering obstacles.

[0003] However, there are at least the following problems in the related art: During the trajectory planning process of intelligent vehicles in an unknown environment, the sampling of trajectory points is relatively random and lacks rationality. Summary of the Invention

[0004] The present invention solves the technical problem that in the trajectory planning process of the prior art for intelligent vehicles, the sampling of trajectory points is relatively random and lacks rationality.

[0005] To solve the above problems, the present invention provides a vehicle path planning method based on recursive sampling, including: determining an initial sampling boundary in the vehicle's own coordinate system; processing the initial sampling boundary according to the initial sampling boundary and the vehicle's own parameters of the target vehicle in combination with the first sampling requirement to obtain a first sampling queue; obtaining an initial planning set according to the first sampling queue in combination with the second sampling requirement, where the initial planning set includes: a plurality of initial planning end points; calculating the cost of all the initial planning end points in the initial planning set to obtain an optimal planning end point.

[0006] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: After obtaining the initial sampling boundary, the coordinates of the initial sampling boundary are judged once in combination with the first sampling requirement and the vehicle's own parameters, and the initial sampling boundary is processed in combination with the first judgment result to obtain the first sampling queue. At this time, the first sampling queue generally includes two first sampling boundary points; then, the two first sampling boundary points in the first sampling queue are judged twice in combination with the second sampling requirement, and the first sampling queue is processed in combination with the second judgment result to obtain the initial planning set. The initial planning set includes multiple initial planning end points, and the initial planning end points here can also be regarded as the first sampling boundary points that meet the second sampling requirement; then, cost calculations are performed on all the initial planning end points in the initial planning set to obtain the optimal planning end point. In an unknown environment, the present application can provide reasonable sampling boundary points as the initial planning end points of the planning method, and can guide the intelligent vehicle to explore forward and backward in the unknown environment; and can sample uniformly and collision-free sampling points to the greatest extent, only need to generate one planning trajectory, without the need to plan multiple trajectories and select the optimal one, reducing the calculation amount.

[0007] In an example of the present invention, the initial sampling boundary includes: a left initial sampling boundary point and a right initial sampling boundary point , and according to the initial sampling boundary and the vehicle's own parameters of the target vehicle, the initial sampling boundary is processed in combination with the first sampling requirement to obtain the first sampling queue, including: calculating the sampling orientation according to the initial sampling boundary; judging whether the initial sampling boundary meets the first sampling requirement in combination with the vehicle's own parameters and the sampling orientation; if so, storing the initial sampling boundary in the first sampling queue; and / or if not, processing the initial sampling boundary to obtain the first sampling boundary, and obtaining the first sampling queue according to the first sampling boundary; wherein, the first sampling requirement is that neither the left initial sampling boundary point nor the right initial sampling boundary point has a collision.

[0008] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: The first sampling requirement is that neither the left initial sampling boundary point nor the right initial sampling boundary point has a collision. According to the length and width of the target vehicle and the sampling orientation, it can be judged whether a collision will occur when the target vehicle travels to the position of the initial sampling boundary; if there is no collision (that is, the initial sampling boundary meets the first sampling requirement), the initial sampling boundary is directly stored in the first sampling queue as a basis for subsequent steps; if there is a collision (that is, the initial sampling boundary does not meet the first sampling requirement), the initial sampling boundary needs to be processed until the processed initial sampling boundary meets the first sampling requirement. The present application can initially narrow the range of the initial sampling boundary, improve the accuracy of sampling, and initially reduce the computing power.

[0009] In an example of the present invention, if not, process the initial sampling boundary to obtain a first sampling boundary, and obtain a first sampling queue according to the first sampling boundary, including: determining whether the left initial sampling boundary point and / or the right initial sampling boundary point meet the first sampling requirement; in the case where the left initial sampling boundary point does not meet the first sampling requirement, shift the left initial sampling boundary point to the right and update it until the left initial sampling boundary point meets the first sampling requirement; and / or in the case where the right initial sampling boundary point does not meet the first sampling requirement, shift the right initial sampling boundary point to the left and update it until the right initial sampling boundary point meets the first sampling requirement; store the updated left initial sampling boundary point and right initial sampling boundary point in the first sampling queue; wherein, the first sampling queue includes: a first left sampling boundary point and a first right sampling boundary point .

[0010] Compared with the prior art, the technical effect achieved by adopting this technical solution: The purpose of this step is to adjust and update the initial sampling boundary that does not meet the first sampling requirement; in the present application, the initial sampling boundary that does not meet the first sampling condition can be effectively avoided from colliding with obstacles by translation.

[0011] In an example of the present invention, obtain an initial planning set according to the first sampling queue in combination with the second sampling requirement, including: obtaining a first center point according to the first sampling queue; updating the first sampling queue in combination with the second sampling requirement, the first center point, and the first sampling queue; storing the first sampling queue in the initial planning set.

[0012] Compared with the prior art, the technical effect achieved by adopting this technical solution: Determine whether a collision will occur when the target vehicle travels to the position of the first center point; update the first sampling queue according to the judgment result in this step until the first center point meets the second sampling requirement.

[0013] In an example of the present invention, update the first sampling queue in combination with the second sampling requirement, the first center point, and the first sampling queue, including: determining whether the first center point meets the second sampling requirement; if so, store the first center point in the first sampling queue; and / or if not, process the first center point to obtain a second center boundary, and obtain a first sampling queue according to the second center boundary; wherein, the second sampling requirement is: no collision occurs at the first center point.

[0014] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: If no collision occurs (that is, the first center point meets the second sampling requirement), the first center point is directly stored in the first sampling queue, that is, a new left and right boundary point formed by the first center point, the first left sampling boundary point, and the first right sampling boundary point is stored in the first sampling queue; If a collision occurs (that is, the first center point does not meet the second sampling requirement), and there is an obstacle at the position of the first center point at this time, the first center point needs to be processed until the processed first center point meets the first sampling requirement.

[0015] In an example of the present invention, if not, the first center point is processed to obtain a second center boundary, and the first sampling queue is obtained according to the second center boundary, including: taking the first center point as a reference point, moving to both sides respectively to generate a second left center point and a second right center point; moving the second left center point to the left and updating it until the second left center point meets the third sampling requirement; moving the second right center point to the right and updating it until the second right center point meets the third sampling requirement; storing the updated second left center point and second right center point in the first sampling queue; wherein, the third sampling requirement is that neither the second left center point nor the second right center point has a collision.

[0016] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: The purpose of this step is to adjust and update the first center point that does not meet the second sampling requirement.

[0017] In an example of the present invention, after storing the first sampling queue in the initial planning set, it further includes: repeatedly executing obtaining the first center point according to the first sampling queue; repeatedly executing combining the second sampling requirement with the first center point and the first sampling queue to update the first sampling queue; repeatedly executing storing the first sampling queue in the initial planning set.

[0018] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: Recursively obtain a new first sampling queue until the sampling point density reaches the requirement, increasing the sampling quantity of the initial planning end points in the initial planning set, facilitating the selection of the optimal planning end point by increasing the sampling quantity; At the same time, through the binary search recursive sampling method, it is possible to flexibly sample reasonable and effective first center points for different first sampling queues.

[0019] In an example of the present invention, after calculating the cost of all initial planning end points in the initial planning set to obtain the optimal planning end point, the vehicle path planning method further includes: generating an initial planning trajectory based on the improved HybirdA* algorithm and Reeds-Shepp curve according to the optimal planning end point; optimizing the initial planning trajectory through an osqp solver to generate a target planning trajectory.

[0020] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: In this application, the improved HybirdA* algorithm and Reeds-Shepp curve can sample specifically according to the forward and backward exploration scenarios, reduce the calculation amount, and plan a more reasonable trajectory.

[0021] On the other hand, the embodiment of the present invention further provides a vehicle path planning system based on recursive sampling, which uses the vehicle path planning method based on recursive sampling in any one of the above embodiments, including: an initial module for determining an initial sampling boundary in the vehicle coordinate system of the host vehicle; a first processing module for processing the initial sampling boundary according to the initial sampling boundary and the vehicle own parameters of the target vehicle in combination with the first sampling requirement to obtain a first sampling queue; a second processing module for obtaining an initial planning set according to the first sampling queue in combination with the second sampling requirement, where the initial planning set includes: a plurality of initial planning end points; an evaluation module for calculating the cost of all the initial planning end points in the initial planning set to obtain an optimal planning end point.

[0022] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: The vehicle path planning system based on recursive sampling in this embodiment is used to implement the vehicle path planning method based on recursive sampling in any one of the embodiments of the present invention. Therefore, it has all the beneficial effects of the vehicle path planning method based on recursive sampling in any one of the embodiments of the present invention, which will not be elaborated here.

[0023] After adopting the technical solution of the present invention, the following technical effects can be achieved: (1) In an unknown environment, this application can provide reasonable sampling boundary points as the initial planning end points of the planning method, and can guide the intelligent vehicle to explore forward and backward in the unknown environment; (2) It can sample uniformly and collision-free sampling points to the maximum extent, only need to generate one planning trajectory, without the need to plan multiple ones and select the optimal one, reducing the calculation amount; (3) It can effectively avoid collisions with obstacles by means of translation; (4) It can flexibly sample reasonable and effective first center points for different first sampling queues by means of dichotomy recursive sampling; (5) The improved HybirdA* algorithm and Reeds-Shepp curve can sample specifically according to the forward and backward exploration scenarios, reduce the calculation amount, and plan a more reasonable trajectory. Description of the Drawings

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings to be used in the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings; Figure 1 It is a flowchart of a vehicle path planning method based on recursive sampling provided by Embodiment 1 of the present invention; Figure 2 It is a coordinate schematic diagram of the vehicle itself and the initial sampling boundary in the vehicle's own coordinate system; Figure 3 It is a coordinate schematic diagram of the vehicle itself and the first sampling queue when the first center point meets the second sampling requirement in the vehicle's own coordinate system; Figure 4 It is a coordinate schematic diagram of the vehicle itself and the second center boundary when the first center point does not meet the second sampling requirement in the vehicle's own coordinate system; Figure 5 It is a structural schematic block diagram of a vehicle path planning system based on recursive sampling provided by Embodiment 2 of the present invention.

[0025] Explanation of reference numerals: 100 - Vehicle path planning system; 101 - Initial module; 102 - First processing module; 103 - Second processing module; 104 - Evaluation module. Detailed implementation manners

[0026] To make the above objects, features, and advantages of the present application more clearly understood, the technical solutions in the embodiments of the present application are clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0027]

Embodiment 1

[0028] In a specific embodiment, the initial sampling boundary is represented by two initial sampling boundary points, specifically as Figure 2 shown as: the left initial sampling boundary point and the right initial sampling boundary point , in the vehicle's own coordinate system ( , with the length direction of the target vehicle as the X-axis and the width direction as the Y-axis), set the ordinate of the left initial sampling boundary point to be positive and the ordinate of the right initial sampling boundary point to be negative, that is . The vehicle's own parameters include: the length and width of the target vehicle.

[0029] Specifically, after obtaining the initial sampling boundary and , combine the first sampling requirement and the vehicle's own parameters to make a first judgment on the coordinates of the initial sampling boundary, and process the initial sampling boundary according to the first judgment result to obtain the first sampling queue. At this time, the first sampling queue generally includes two first sampling boundary points; then, combine the second sampling requirement to make a second judgment on the two first sampling boundary points in the first sampling queue, and process the first sampling queue according to the second judgment result to obtain the initial planning set. The initial planning set includes multiple initial planning end points. The initial planning end points here can also be regarded as the first sampling boundary points that meet the second sampling requirement; then, calculate the cost for all the initial planning end points in the initial planning set to obtain the optimal planning end points. In an unknown environment, the present application can provide reasonable sampling boundary points as the initial planning end points of the planning method, and can guide the intelligent vehicle to explore forward and backward in the unknown environment; and can sample the most uniform and collision-free sampling points to the greatest extent. Only one planning trajectory needs to be generated, without the need to plan multiple trajectories and select the optimal one, reducing the computational amount.

[0030] Preferably, before determining the initial sampling boundary, first determine the initial sampling range as , , , obtain the initial sampling boundary according to the initial sampling range, and set the sampling density as k. In the present application, the sampling density represents the distance threshold of the distance between the initial sampling boundaries on the Y-axis.

[0031] Further, S200 includes: S210: Calculate the sampling orientation according to the initial sampling boundary; S220: Combine the vehicle's own parameters and the sampling orientation to determine whether the initial sampling boundary meets the first sampling requirement; S221: If so, store the initial sampling boundary in the first sampling queue; and / or S222: If not, process the initial sampling boundary to obtain the first sampling boundary, and obtain the first sampling queue according to the first sampling boundary.

[0032] Specifically, calculate the sampling orientation according to the initial sampling range or the initial sampling boundary. The sampling orientation ; The first sampling requirement is that neither the left initial sampling boundary point nor the right initial sampling boundary point has a collision. According to the length and width of the target vehicle and the sampling orientation, it can be judged whether a collision will occur when the target vehicle travels to the position of the initial sampling boundary; if there is no collision (that is, the initial sampling boundary meets the first sampling requirement), then directly store the initial sampling boundary in the first sampling queue as a basis for subsequent steps; if there is a collision (that is, the initial sampling boundary does not meet the first sampling requirement), then the initial sampling boundary needs to be processed until the processed initial sampling boundary meets the first sampling requirement. Through S200 of the present application, the range of the initial sampling boundary can be initially reduced, the accuracy of sampling can be improved, and the computing power can be initially reduced.

[0033] Further, S222 includes: Judge whether the left initial sampling boundary point and / or the right initial sampling boundary point meets the first sampling requirement; In the case where the left initial sampling boundary point does not meet the first sampling requirement, shift the left initial sampling boundary point to the right and update it until the left initial sampling boundary point meets the first sampling requirement; and / or In the case where the right initial sampling boundary point does not meet the first sampling requirement, shift the right initial sampling boundary point to the left and update it until the right initial sampling boundary point meets the first sampling requirement; Store the updated left initial sampling boundary point and right initial sampling boundary point in the first sampling queue; Among them, the first sampling queue includes: the first left sampling boundary point and the first right sampling boundary point .

[0034] Specifically, the purpose of this step is to adjust and update the initial sampling boundary that does not meet the first sampling requirement. S222 specifically includes: when the left initial sampling boundary point collides (that is, the left initial sampling boundary point does not meet the first sampling requirement), the Y-axis coordinate of the left initial sampling boundary point is translated to the right (that is, translated to the negative half-axis of the Y-axis) until the left initial sampling boundary point has no collision (or the Y-axis coordinate of the updated left initial sampling boundary point < the Y-axis coordinate of the right initial sampling boundary point + k); define the left initial sampling boundary point that meets the first sampling requirement after update as: the first left sampling boundary point , L represents the left side, then = , .

[0035] When the right initial sampling boundary point collides (that is, the right initial sampling boundary point does not meet the first sampling requirement), the Y-axis coordinate of the right initial sampling boundary point is translated to the left (that is, translated to the positive half-axis of the Y-axis) until the right initial sampling boundary point has no collision (or the Y-axis coordinate of the updated right initial sampling boundary point > the Y-axis coordinate of the left initial sampling boundary point - k); define the right initial sampling boundary point that meets the first sampling requirement after update as: the first right sampling boundary point , R represents the right side, then = , , at this time . In this application, the initial sampling boundary that does not meet the first sampling condition can effectively avoid collision with obstacles by translation.

[0036] Preferably, , represents the moving distance of each translation of the left initial sampling boundary point, represents the number of times the left initial sampling boundary point needs to move to meet the first sampling requirement; , represents the moving distance of each translation of the right initial sampling boundary point, represents the number of times the right initial sampling boundary point needs to move to meet the first sampling requirement.

[0037] Further, S300 includes: S310: Obtain the first center point according to the first sampling queue; S320: Combine the second sampling requirement with the first center point and the first sampling queue to update the first sampling queue; S330: Store the first sampling queue into the initial planning set; wherein, the second sampling requirement is: the first center point does not collide.

[0038] Specifically, after obtaining the first sampling queue, the first center point ( can be calculated based on the first left sampling boundary point and the first right sampling boundary point ), where m represents the center, , . The second sampling requirement is that there is no collision at the first center point. Determine whether a collision will occur when the target vehicle travels to the position of the first center point; update the first sampling queue according to the judgment result in this step until the first center point meets the second sampling requirement.

[0039] Further, S320 includes: S321: Determine whether the first center point meets the second sampling requirement; S322: If so, store the first center point in the first sampling queue; and / or S323: If not, process the first center point to obtain the second center boundary, and obtain the first sampling queue according to the second center boundary.

[0040] Specifically, determine whether a collision will occur when the target vehicle travels to the position of the first center point; refer to Figure 3 . If there is no collision (that is, the first center point meets the second sampling requirement), and there is no obstacle at the position of the first center point at this time, directly store the first center point in the first sampling queue, that is, form new left and right boundary points with the first left sampling boundary point and the first right sampling boundary point and store them in the first sampling queue; as in Figure 3 , regard the first center point as two boundary points: the new first right sampling boundary point and the new first left sampling boundary point. The first left sampling boundary point, the new first right sampling boundary point, the new first left sampling boundary point, and the first right sampling boundary point together form the new first sampling queue.

[0041] Refer to Figure 4 . If a collision occurs (that is, the first center point does not meet the second sampling requirement), and there is an obstacle at the position of the first center point at this time, the first center point needs to be processed until the processed first center point meets the second sampling requirement, and generate the second center boundary according to the updated first center point that meets the second sampling requirement.

[0042] Further, S323 includes: Taking the first center point as the reference point, move to both sides respectively to generate the second left center point and the second right center point; Move the second left center point to the left and update it until the second left center point meets the third sampling requirement; Shift the second right center point to the right and update it until the second right center point meets the third sampling requirement; Store the updated second left center point and second right center point in the first sampling queue; Among them, the third sampling requirement is that neither the second left center point nor the second right center point has a collision.

[0043] Specifically, the purpose of this step is to adjust and update the first center point that does not meet the second sampling requirement. S323 specifically includes: regarding the first center point as two boundary points: the second left center point and the second right center point; moving the two boundary points to the left and right sides of the first center point respectively until the left-shifted second left center point and the right-shifted second right center point meet the third sampling requirement (or the Y-axis coordinate of the updated second left center point > the Y-axis coordinate of the first left sampling boundary point - k, or the Y-axis coordinate of the updated second right center point < the Y-axis coordinate of the first right sampling boundary point + k); generating a second center boundary based on the second left center point and the second right center point that meet the third sampling requirement.

[0044] Specifically, if the updated second left center point is ( ), then , , represents the moving distance of each translation of the second left center point, represents the number of times the second left center point needs to move to meet the third sampling requirement; if the updated second right center point is ( ), then , , represents the moving distance of each translation of the second right center point, represents the number of times the second right center point needs to move to meet the third sampling requirement.

[0045] Preferably, .

[0046] Further, after S330, it further includes: Loop to execute S310; loop to execute S320; loop to execute S330.

[0047] Specifically, a loop is executed to obtain the first center point according to the first sampling queue; a loop is executed to update the first sampling queue in combination with the second sampling requirement, the first center point and the first sampling queue; a loop is executed to store the first sampling queue in the initial planning set, and a new first sampling queue can be recursively obtained until the sampling point density reaches the requirement, and the sampling number of the initial planning end point in the initial planning set is increased, so that the optimal planning end point can be selected by increasing the sampling number; at the same time, a reasonable and effective first center point can be flexibly sampled for different first sampling queues through binary recursive sampling.

[0048] Furthermore, after S400, the vehicle path planning method further includes: Based on the improved HybridA* algorithm and Reeds-Shepp curve, the initial planning trajectory is generated according to the optimal planning end point; The initial planning trajectory is optimized through the osqp solver to generate the target planning trajectory.

[0049] Specifically, the improved HybirdA* algorithm and Reeds-Shepp curve are used to generate the initial planning trajectory according to the optimal planning end point; that is, in the forward exploration scenario, the HybirdA* algorithm only generates forward sampling points (that is, only samples in the direction of the front of the target vehicle), and the Reeds-Shepp curve does not consider the curve with backward conditions; in the backward exploration scenario, the HybirdA* algorithm only generates backward sampling points (that is, only samples in the direction of the rear of the target vehicle), and the Reeds-Shepp curve does not consider the curve with forward conditions; after generating the rough initial planning trajectory, the trajectory and speed are optimized through the osqp solver, and finally the forward and backward target planning trajectory is generated. In this application, the improved HybirdA* algorithm and Reeds-Shepp curve can perform targeted sampling according to the forward and backward exploration scenarios, reduce the amount of calculation, and plan a more reasonable trajectory.

[0050] Preferably, S400 specifically includes: setting the evaluation function to: ,in, and is the weight value, is the absolute value of the Y-axis coordinate of the initial planning end point, d is the maximum effective distance between the vehicle and the obstacle, is the distance between the vehicle and the obstacle. The initial planning end point with the minimum cost is selected in the initial planning set as the optimal planning end point for subsequent planning.

[0051] [Example 2] See also Figure 5, this embodiment also provides a vehicle path planning system 100 based on recursive sampling, which uses the vehicle path planning method as in the first embodiment. The vehicle path planning system 100 includes: an initial module 101, a first processing module 102, a second processing module 103, and an evaluation module 104. The initial module 101 is used to determine an initial sampling boundary in the vehicle's own coordinate system. The first processing module 102 is used to process the initial sampling boundary according to the initial sampling boundary and the vehicle's own parameters of the target vehicle, in combination with the first sampling requirement, to obtain a first sampling queue. The second processing module 103 is used to obtain an initial planning set according to the first sampling queue, in combination with the second sampling requirement, where the initial planning set includes: a plurality of initial planning end points. The evaluation module 104 is used to calculate the cost of all the initial planning end points in the initial planning set to obtain the optimal planning end point.

[0052] In a specific embodiment, the initial module 101, the first processing module 102, the second processing module 103, and the evaluation module 104 of the vehicle path planning system 100 cooperate to implement the vehicle path planning method based on recursive sampling as in the first embodiment above, which will not be elaborated here.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the present invention in each embodiment.

Claims

1. A vehicle path planning method based on recursive sampling, characterized in that, The vehicle path planning method includes: Determine an initial sampling boundary in the vehicle's own coordinate system; Process the initial sampling boundary according to the initial sampling boundary and the vehicle's own parameters of the target vehicle, and combine the first sampling requirement to obtain a first sampling queue; Obtain an initial planning set according to the first sampling queue and combine the second sampling requirement, where the initial planning set includes: a plurality of initial planning end points; Perform cost calculation on all the initial planning end points in the initial planning set to obtain an optimal planning end point.

2. The vehicle path planning method according to claim 1, characterized in that The initial sampling boundary includes: the left initial sampling boundary point and the right initial sampling boundary point , The step of processing the initial sampling boundary according to the initial sampling boundary and the vehicle's own parameters of the target vehicle, and combining the first sampling requirement to obtain a first sampling queue includes: Calculate a sampling orientation according to the initial sampling boundary; Combine the vehicle's own parameters and the sampling orientation to determine whether the initial sampling boundary meets the first sampling requirement; If so, store the initial sampling boundary in the first sampling queue; and / or If not, process the initial sampling boundary to obtain a first sampling boundary, and obtain the first sampling queue according to the first sampling boundary; Wherein, the first sampling requirement is that neither the left initial sampling boundary point nor the right initial sampling boundary point has a collision.

3. The vehicle path planning method according to claim 2, wherein The step of, if not, processing the initial sampling boundary to obtain a first sampling boundary, and obtaining the first sampling queue according to the first sampling boundary includes: Determine whether the left initial sampling boundary point and / or the right initial sampling boundary point meets the first sampling requirement; In the case where the left initial sampling boundary point does not meet the first sampling requirement, shift the left initial sampling boundary point to the right and update it until the left initial sampling boundary point meets the first sampling requirement; and / or In the case where the right initial sampling boundary point does not meet the first sampling requirement, shift the right initial sampling boundary point to the left and update it until the right initial sampling boundary point meets the first sampling requirement; Store the updated left initial sampling boundary point and the right initial sampling boundary point in the first sampling queue; Among them, the first sampling queue includes: a first left sampling boundary point and a first right sampling boundary point .

4. The vehicle path planning method according to claim 1, wherein The step of obtaining an initial planning set according to the first sampling queue and combining the second sampling requirement includes: Obtain a first center point according to the first sampling queue; Combine the second sampling requirement, the first center point, and the first sampling queue to update the first sampling queue; Store the first sampling queue in the initial planning set.

5. The vehicle path planning method according to claim 4, wherein The step of combining the second sampling requirement, the first center point, and the first sampling queue to update the first sampling queue includes: Determine whether the first center point meets the second sampling requirement; If so, store the first center point in the first sampling queue; and / or Otherwise, process the first center point to obtain a second center boundary, and obtain the first sampling queue according to the second center boundary; Wherein, the second sampling requirement is that the first center point does not collide.

6. The vehicle path planning method according to claim 5, characterized in that, The "otherwise, process the first center point to obtain a second center boundary, and obtain the first sampling queue according to the second center boundary" includes: Taking the first center point as a reference point, move to both sides respectively to generate a second left center point and a second right center point; Move the second left center point to the left and update it until the second left center point meets the third sampling requirement; Move the second right center point to the right and update it until the second right center point meets the third sampling requirement; Store the updated second left center point and the second right center point into the first sampling queue; Wherein, the third sampling requirement is that neither the second left center point nor the second right center point collides.

7. The vehicle path planning method according to claim 4, characterized in that, After storing the first sampling queue into the initial planning set, it further includes: Loop to execute obtaining a first center point according to the first sampling queue; Loop to execute combining the second sampling requirement, the first center point, and the first sampling queue to update the first sampling queue; Loop to execute storing the first sampling queue into the initial planning set.

8. The vehicle path planning method according to claim 1, characterized in that, After calculating the cost of all the initial planning end points in the initial planning set to obtain an optimal planning end point, the vehicle path planning method further includes: Based on the improved HybirdA* algorithm and the Reeds-Shepp curve, generate an initial planning trajectory according to the optimal planning end point; Optimize the initial planning trajectory through an osqp solver to generate a target planning trajectory.

9. A vehicle path planning system based on recursive sampling, the vehicle path planning system using the recursive sampling-based vehicle path planning method according to any one of claims 1-8, characterized in that, It includes: An initial module for determining an initial sampling boundary in the vehicle's own coordinate system; A first processing module for processing the initial sampling boundary according to the initial sampling boundary and the vehicle's own parameters of the target vehicle, and combining the first sampling requirement to obtain a first sampling queue; A second processing module for obtaining an initial planning set according to the first sampling queue and combining the second sampling requirement, wherein the initial planning set includes: multiple initial planning end points; An evaluation module for calculating the cost of all the initial planning end points in the initial planning set to obtain an optimal planning end point.

Images