An automatic parking path planning method and device, electronic equipment and storage medium

Abstract

The application provides an automatic parking path planning method and device, electronic equipment and a storage medium. The method comprises the following steps: obtaining target information for parking planning; abstractly modeling obstacles and parking spaces based on the target information to obtain a target vehicle driving environment; determining a current position of the target vehicle as a parking planning starting point, determining a position where the parking space is located as a parking planning ending point, performing discrete point sampling in the target vehicle driving environment, planning a parking path based on the sampling result, flexibly determining the parking path based on the sampling result, so that a combination of geometric figures corresponding to the planned parking path can be theoretically infinite, and the parking path is not limited by a starting attitude and an ending attitude of the target vehicle, and the success rate of an arbitrary attitude is high. Planning the parking path based on the sampling result can make the parking path closer to a global optimal value, meet vehicle dynamics constraints, and ensure the smoothness of the parking path.

Description

Technical Field

[0001] This invention relates to the field of vehicle networking, specifically to an automatic parking path planning method, device, electronic device, and storage medium. Background Technology

[0002] With the rapid development of artificial intelligence technology, autonomous driving technology has gradually received more attention, and vehicles equipped with autonomous driving functions are becoming increasingly popular. Automated parking path planning algorithms are one of the most crucial methods in autonomous driving systems, and the success rate of path planning in automatic parking assist (APA) systems is a key factor determining the user experience.

[0003] Currently, related technologies generally limit the size of the feasible solution space for path planning by pre-setting four combinations of straight lines and arcs. If the planning start and end points are not ideal, it is easy to fail to find a feasible path among these four combinations. Furthermore, these technologies directly use the combination of straight lines and arcs as the output of the planning algorithm. Discontinuities in curvature at the connection points of geometric shapes can cause oscillations in the lateral control of downstream vehicles. In the case of geometric shape combinations, when the starting heading and the ending heading of the parking space do not meet the pre-assumpted conditions, it is very difficult to establish the geometric equations. When two arcs are connected, in order to facilitate the solution, the radius of the arcs is set to be the same, which further limits the size of the feasible solution space. When the planning start and end points are not ideal, it is difficult to determine the parking path. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the present invention provides an automatic parking path planning method, device, electronic device and storage medium to solve the above technical problems.

[0005] This invention provides an automatic parking path planning method, which includes: acquiring target information for parking planning; wherein the target information includes: obstacles, parking spaces, target vehicle position, and target vehicle posture; abstractly modeling obstacles and parking spaces based on the target information to obtain the target vehicle's driving environment; taking the current position of the target vehicle as the starting point of parking planning and the location of the parking space as the ending point of parking planning; performing discrete point sampling on the target vehicle's driving environment, searching for a parking path from the starting point to the ending point of parking planning based on the sampling results, and feeding back the parking path to the target vehicle.

[0006] In one embodiment of the present invention, discrete point sampling is performed on the driving environment of the target vehicle, and a parking path from the starting point to the ending point of the parking plan is searched based on the sampling results. This includes: performing discrete point sampling on the driving environment of the target vehicle to obtain multiple sampling points; forming a target expansion tree based on the multiple sampling points, and taking the starting point of the parking plan as the root node of the target expansion tree; determining whether the path search time is less than a preset time threshold; if the path search time is greater than or equal to the preset time threshold, calculating the cost of the multiple candidate paths to obtain multiple candidate path costs, performing a weighted average calculation on the multiple candidate costs, and determining the candidate path with the lowest cost as the parking path; if... If the path search time is less than a preset time threshold, then state points are randomly sampled in the state space and constrained according to preset conditions to obtain constrained state points; wherein, the state point is a sample point in the target expansion tree, and the state space is generated based on the target vehicle position and target vehicle attitude; other sample points in the target expansion tree are used as potential connection points, starting from the constrained state point, the potential connection points in the target expansion tree are traversed, the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point is calculated, and the parking path from the parking planning start point to the parking planning endpoint is searched based on the heuristic cost calculation results of all potential connection points.

[0007] In one embodiment of the present invention, a parking path from the starting point to the ending point of parking planning is searched based on the heuristic cost calculation results of all potential connection points. A heuristic cost table is generated based on the heuristic cost calculation results of all potential connection points. All potential connection points are sorted according to the heuristic cost table to form a potential connection point list. A directed circular arc connection is performed with the potential connection point corresponding to each heuristic cost as the starting point and the current state point as the ending point to obtain a first circular arc connection line. If the radius of the first circular arc connection line is less than a first preset radius threshold, collision detection is performed on the first circular arc connection line to obtain a first collision detection result. Based on the first collision detection result and the first circular arc connection line, a parking path from the starting point to the ending point of parking planning is searched. If the radius of the first circular arc connection line is greater than or equal to the first preset radius threshold, the constraint state point and potential connection point are reacquired. A new first circular arc connection line is generated by the reacquired constraint state point and potential connection point until the radius of the new first circular arc connection line is less than the first preset radius threshold.

[0008] In one embodiment of the present invention, searching for a parking path from the starting point to the ending point of parking planning based on the first collision detection result and the first arc connecting line includes: if the first collision detection result is a collision detection failure, obtaining a new constraint state point in the state space, generating a new heuristic cost table based on the new constraint state point, and re-performing collision detection based on the new heuristic cost table until the collision detection result corresponding to the new heuristic cost table passes the collision detection; if the first collision detection result is a collision detection success, or the collision detection result corresponding to the new heuristic cost table passes the collision detection; then creating a new state point, adding the new state point to the target expansion tree, and setting the parent node of the new state point as the current potential connection point; when there is a tangent directed arc on the ray from the current potential connection point to the parking space direction of the parking planning ending point, determining the directed arc as the second arc connecting line, and searching for a parking path from the starting point to the ending point of parking planning based on the second arc connecting line and the first arc connecting line.

[0009] In one embodiment of the present invention, searching for a parking path from the starting point to the ending point of the parking plan based on the second arc connecting line and the first arc connecting line includes: when the radius of the second arc connecting line is greater than or equal to a second preset radius threshold, and the path search time is less than a preset time threshold, generating a new second arc connecting line based on the next potential connection point to the parking plan ending point, until the radius of the new second arc connecting line is less than the second preset radius threshold; when the radius of the second arc connecting line is less than the second preset radius threshold, performing collision detection on the second arc connecting line to obtain a second collision detection result; if the second collision... If the collision detection result is a failure, and the path search time is less than a preset time threshold, a new second arc connection line is generated based on the next potential connection point to the parking planning endpoint. Collision detection is then performed again based on the radius of the new second arc connection line until the collision detection result corresponding to the new second arc connection line passes the collision detection. If the second collision detection result is a success, or the collision detection result corresponding to the new second arc connection line passes the collision detection, then the first arc connection line and the second arc connection line form a potential parking planning path, and dynamic constraints are applied to the potential parking planning path to obtain the parking path.

[0010] In one embodiment of the present invention, dynamic constraints are applied to the potential parking planning path to obtain a parking path, including: establishing a dynamic model for the target vehicle; starting from the starting point of the potential path, performing a tracking simulation on the potential path, and determining candidate paths based on the simulation results; performing collision detection on the candidate paths to obtain a third collision detection result; if the third collision detection result is a successful collision detection, then adding the candidate path to the candidate path list; if the third collision detection result is a failed collision detection, and the path search time is less than a preset time threshold, dynamic constraints are applied to the new potential parking planning path to obtain a new parking path, and the new parking path is added to the candidate path list; cost calculation is performed on multiple candidate paths in the candidate path list to obtain multiple candidate path costs; the multiple candidate costs are weighted and averaged, and the candidate path with the lowest cost is determined as the parking path.

[0011] In one embodiment of the present invention, acquiring target information for parking planning includes: sensing the obstacle and the parking space through a sensing sensor; wherein the sensing sensor includes an ultrasonic sensor, a millimeter-wave sensor, and a camera; and locating the target vehicle position and target vehicle attitude through a positioning sensor.

[0012] This invention provides an automatic parking path planning device, comprising: an information acquisition module for acquiring target information for parking planning; wherein the target information includes: obstacles, parking spaces, target vehicle position, and target vehicle posture; a driving environment modeling module for abstractly modeling obstacles and parking spaces based on the target information to obtain the target vehicle's driving environment; and a path planning module for using the target vehicle's current position as the parking planning start point and the parking space's location as the parking planning end point; and performing discrete point sampling on the target vehicle's driving environment, searching for a parking path from the parking planning start point to the parking planning end point based on the sampling results, and feeding the parking path back to the target vehicle.

[0013] The beneficial effects of this invention are as follows: By acquiring target information for parking planning; abstractly modeling obstacles and parking spaces based on the target information to obtain the target vehicle's driving environment; by determining the current position of the target vehicle as the starting point of parking planning and the location of the parking space as the ending point, discrete point sampling is performed in the target vehicle's driving environment, and a parking path is planned based on the sampling results. This parking path is then fed back to the target vehicle. Thus, the parking path can be flexibly determined based on the sampling results, theoretically allowing for an infinite number of combinations of geometric shapes corresponding to the planned parking path. Furthermore, the planning of the parking path is not limited by the target vehicle's initial and final postures, making the path planning process flexible and achieving a high success rate for any posture. It also supports multi-threaded planning, improving planning efficiency. Moreover, the parking path planned based on the sampling results is closer to the global optimum, satisfies vehicle dynamics constraints, and ensures the smoothness of the parking path.

[0014] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0016] Figure 1 This is a system architecture diagram illustrating an automatic parking path planning method according to an exemplary embodiment of this application;

[0017] Figure 2 This is a flowchart illustrating an exemplary embodiment of the automatic parking path planning method of this application;

[0018] Figure 3 This is a schematic diagram illustrating a potential path for smooth parking planning, as shown in an exemplary embodiment of this application.

[0019] Figures 4(a) and 4(b) are schematic diagrams illustrating automatic parking path planning at two different times, as shown in an exemplary embodiment of this application;

[0020] Figure 5 This is a flowchart illustrating an exemplary embodiment of the automatic parking path planning method of this application;

[0021] Figure 6 This is a block diagram illustrating an automatic parking path planning device in an exemplary embodiment of this application;

[0022] Figure 7 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown. Detailed Implementation

[0023] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0024] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0025] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0026] First, it's important to clarify that the cl-RRT (Contraction-based Rapidly-exploring Random Trees) algorithm is a contraction-based rapid exploration random tree algorithm used for path planning and robot motion planning. It's an improvement upon the Rapidly-exploring Random Trees (RRT) algorithm. The main idea of ​​cl-RRT is to introduce a contraction operation into the RRT algorithm. This algorithm improves the quality and feasibility of the path by locally contracting the nodes between the tree nodes. Specifically, the cl-RRT algorithm includes the following steps: Initialization: Select a starting point as the root node of the tree and add it to the tree. Expansion: Randomly sample a point in the state space and find the nearest node in the tree. Starting from this node, expand along the direction of the sampled point until a certain expansion distance is reached. Contraction: Perform a contraction operation on the newly expanded path, iteratively adjusting the positions of nodes on the path to gradually bring it closer to obstacle boundaries or other constraints. The contraction operation aims to improve the feasibility and quality of the path. Connection: Connect the contracted path to the nearest node in the tree, forming a new tree structure. Repeat: Repeat the expansion step to the connection step until the target point is reached or the specified number of iterations is reached. Path extraction: Finding the path connecting the starting point and the target point from the tree, typically selecting the shortest path or the path with the best quality as the final planned path. The cl-RRT algorithm, by introducing a shrinking operation, can further improve the feasibility and quality of the path based on the rapid exploration of the random tree. It has wide applications in robot path planning, autonomous navigation, and other fields, and can be improved and optimized according to specific needs.

[0027] Please see Figure 1 , Figure 1 A schematic diagram of an exemplary system architecture to which the technical solutions of embodiments of the present invention can be applied is shown. For example... Figure 1As shown, the system architecture may include a positioning module 110, a perception module 120, an environment reconstruction module 130, and a path planning module 140. The environment reconstruction module 130, based on the location information of parking spaces, obstacles, and the target vehicle provided by the positioning module 110, and combined with the obstacle information provided by the perception module 120, abstracts and models the environment surrounding the vehicle, abstracting polygonal obstacles and convex polygonal parking space information, and calculates the parking planning endpoint within the parking space. The path planning module 140 uses the current position of the target vehicle as the starting point of the parking planning and the parking endpoint output by the environment reconstruction module 130 as the planning endpoint. It searches for a path by sampling discrete points in the environment and performs collision detection based on the environmental information output by the environment reconstruction module 130, ultimately outputting a smooth, collision-free path that satisfies vehicle dynamics constraints. In this way, by abstractly modeling obstacles and parking spaces based on location information, the driving environment of the target vehicle is obtained. By determining the current position of the target vehicle as the starting point of parking planning and the location of the parking space as the ending point, discrete point sampling is performed in the target vehicle's driving environment. Based on the sampling results, a parking path is planned and fed back to the target vehicle. This allows for flexible determination of the parking path based on the sampling results, theoretically resulting in an infinite number of possible geometric combinations for the planned parking path. Furthermore, the planning process is not limited by the target vehicle's initial and final postures, offering flexibility and a high success rate for any posture. It also supports multi-threaded planning, improving planning efficiency. The parking path planned based on sampling results is closer to the global optimum, satisfies vehicle dynamics constraints, and ensures the smoothness of the parking path.

[0028] Please see Figure 2 , Figure 2 This is a flowchart illustrating an exemplary embodiment of the automatic parking path planning method of this application. Figure 2 As shown, in an exemplary embodiment, the automatic parking path planning method includes at least steps S210 to S250, which are described in detail below:

[0029] Step S210: Obtain target information for parking planning; wherein, the target information includes: obstacles, parking spaces, target vehicle position and target vehicle attitude.

[0030] In one embodiment of this application, obtaining target information for parking planning includes: sensing obstacles and parking spaces through sensing sensors; wherein the sensing sensors include ultrasonic sensors, millimeter-wave sensors, and cameras; and locating the target vehicle position and target vehicle attitude through positioning sensors.

[0031] Step S220: Based on the target information, abstract modeling of obstacles and parking spaces is performed to obtain the target vehicle's driving environment.

[0032] In one embodiment of this application, the environment reconstruction module abstracts and models the environment surrounding the vehicle based on obstacles provided by the perception module, abstracting polygonal obstacles. These obstacles can be polygons of various geometric shapes, such as rectangles, triangles, pentagons, etc. In this embodiment, each obstacle is abstracted as a polygon, and its shape can be represented using the vertex coordinates of the polygon. Furthermore, other attributes of the obstacles, such as size and orientation, can be recorded by abstracting and modeling them. By abstracting polygonal obstacles into geometric shape and positional information, the spatial layout and characteristics of polygonal obstacles can be represented in environment modeling. These abstract models can be used for tasks such as path planning and collision detection to ensure the vehicle can move and park safely.

[0033] In one embodiment of this application, the environment reconstruction module provides the perception module with an abstract model of the environment surrounding the vehicle, based on the parking space information. This abstracts polygonal obstacles, specifically identifying and extracting the shape and location information of the convex polygonal parking space based on the perception system. Convex polygonal parking spaces are commonly used in parking scenarios, and their shape can be a rectangle, square, or other convex polygons. Abstracting each parking space as a convex polygon allows its shape to be represented using the vertex coordinates. Similarly, by abstracting the parking space, other attributes such as size and orientation can be recorded. By abstracting the convex polygonal parking space into geometric shape and location information, the spatial layout and characteristics of the convex polygonal parking space can be represented in the environment model. These abstract models can be used for tasks such as path planning and collision detection to ensure the safe movement and parking of vehicles.

[0034] Step S230: The current position of the target vehicle is taken as the starting point of the parking planning, and the position of the parking space is taken as the ending point of the parking planning.

[0035] Step S240: Discrete point sampling is performed on the target vehicle's driving environment. Based on the sampling results, a parking path from the starting point to the ending point of the parking planning is searched, and the parking path is fed back to the target vehicle.

[0036] In one embodiment of this application, discrete point sampling is performed on the target vehicle's driving environment to generate a set of discrete points as candidate points for path search. These sampling points can be selected according to the actual scenario; they can be discrete points in the state space or key locations in the environment.

[0037] exist Figure 2The technical solution involves acquiring target information for parking planning; abstractly modeling obstacles and parking spaces based on the target information to obtain the target vehicle's driving environment; determining the target vehicle's current position as the starting point of parking planning and the parking space's location as the ending point; sampling discrete points in the target vehicle's driving environment; planning a parking path based on the sampling results; and feeding the parking path back to the target vehicle. This allows for flexible determination of the parking path based on the sampling results, theoretically enabling an infinite number of geometric combinations for the planned parking path. Furthermore, the planning process is not limited by the target vehicle's initial and final postures, offering flexibility and a high success rate for arbitrary postures. It also supports multi-threaded planning, improving planning efficiency. Planning a parking path based on sampling results ensures the path is closer to the global optimum, satisfies vehicle dynamics constraints, and guarantees the smoothness of the parking path.

[0038] In one embodiment of this application, discrete point sampling is performed on the target vehicle's driving environment, and a parking path from the parking planning start point to the parking planning end point is searched based on the sampling results. This includes: performing discrete point sampling in the target vehicle's driving environment to obtain multiple sampling points; forming a target expansion tree based on the multiple sampling points, with the parking planning start point as the root node of the target expansion tree; determining whether the path search time is less than a preset time threshold; if the path search time is greater than or equal to the preset time threshold, calculating the cost of the multiple candidate paths to obtain multiple candidate path costs, performing a weighted average calculation on the multiple candidate costs, and determining the candidate path with the lowest cost as the parking path. Parking path; if the path search time is less than a preset time threshold, state points are randomly sampled in the state space and constrained according to preset conditions to obtain constrained state points; where each state point is a sample point in the target expansion tree, and the state space is generated based on the target vehicle position and attitude; other sample points in the target expansion tree are used as potential connection points. Starting from the constrained state point, the potential connection points in the target expansion tree are traversed, and the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point is calculated. Based on the heuristic cost calculation results of all potential connection points, a parking path from the parking planning start point to the parking planning endpoint is searched. During the path search process, the connection relationship and feasibility between discrete sample points are considered to generate a path from the starting point to the target point. The search algorithm can be adjusted and optimized according to actual needs to find the optimal or reasonable path, which not only ensures the safety and feasibility of path planning, but also enables vehicles to navigate and move safely in complex environments.

[0039] In one embodiment of the application, constrained state points are obtained by constraining state points according to preset conditions, including hard constraints and probabilistic constraints on the state points.

[0040] In one embodiment of the application, hard constraints define upper and lower bounds for each state in the state space, preventing the search algorithm from performing wide-area searches of large state spaces. Hard constraints can increase the speed of finding a path and reduce the path length. Probabilistic constraints refer to artificially increasing the sampling probability values ​​around certain state points based on the parking preferences of general car owners and the rationality of the path, when all environmental information and vehicle positioning status are known. This can guide the path planning algorithm to be more likely to select these areas as path connection points, thus taking into account specific preferences and rationality when generating the path. In this embodiment, hard constraints are applied to the state space by setting boundary values ​​for the horizontal and vertical coordinates and heading angle, which can increase the speed of finding a path and reduce the path length. In this embodiment, the midpoint of the line connecting the parking planning start point and the parking planning end point is taken. The probability of the state variables in the state space at the midpoint can be set. Based on the midpoint and the variance, a Gaussian distribution function is generated as the probability distribution function for sampling near the midpoint. By adjusting the size of the variance, the probability of sampling near the midpoint can be increased. This can guide path planning algorithms to be more likely to select these areas as path connection points, thus taking into account specific preferences and rationality when generating paths.

[0041] In one embodiment of this application, a parking path from the starting point to the ending point of parking planning is searched based on the heuristic cost calculation results of all potential connection points. This includes: generating a heuristic cost table based on the heuristic cost calculation results of all potential connection points; sorting all potential connection points according to the heuristic cost table to form a potential connection point list; performing directed circular arc connections with each potential connection point corresponding to a heuristic cost as the starting point and the current state point as the ending point to obtain a first circular arc connection line; if the radius of the first circular arc connection line is less than a first preset radius threshold, performing collision detection on the first circular arc connection line to obtain a first collision detection result; and searching for a parking path from the starting point to the ending point of parking planning based on the first collision detection result and the first circular arc connection line; if the radius of the first circular arc connection line is greater than or equal to the first preset radius threshold, re-acquiring constraint state points and potential connection points, and generating a new first circular arc connection line through the re-acquiring constraint state points and potential connection points, until the radius of the new first circular arc connection line is less than the first preset radius threshold.

[0042] In one embodiment of this application, for each potential connection point corresponding to the heuristic cost, a directed arc connection is made starting from the potential connection point and with the current sampling point as the destination. The arc satisfies the position constraints of the two points and the direction constraints of the potential connection point. At the same time, considering that the central angle of the arc does not exceed 180 degrees, a directed arc can be uniquely solved, and the directed arc is determined as the first arc connection line.

[0043] In one embodiment of the application, the heuristic cost is calculated for each node (potential connection point) in the RRT extension tree. The heuristic cost can be selected from Manhattan distance, Euclidean distance, weighted distance of each state variable, or arc connection length, depending on actual needs. This step does not require traversing all nodes in the tree; instead, it can calculate and sort the heuristic cost only for potential connection points within a certain range near the current sampling point. Nodes are then sorted according to their heuristic cost, either from smallest to largest or from largest to smallest. Nodes with lower heuristic costs are placed at the front to facilitate subsequent path selection. By limiting the range of traversed nodes, the traversal of all nodes in the entire tree is reduced, thereby reducing computational load and improving efficiency.

[0044] In one embodiment of this application, the SAT (Separate Axis Theorem) algorithm is used to perform projection collision detection on abstracted polygonal obstacles and convex polygons of parking spaces. If the environment modeling is abstracted as a raster map, the FCC (Fast Collision Check) algorithm can be used. Additionally, some basic geometric collision detection algorithms can be used for points and lines. In this embodiment, the separate axis algorithm is a commonly used algorithm for collision detection, used to determine whether two convex polygons intersect. If two convex polygons do not intersect, then there must exist a separate axis between them, meaning that the projections on this axis will not overlap. Fast collision detection algorithms are a class of algorithms for efficient collision detection of complex geometries, which can reduce computation and improve the speed of collision detection to adapt to application scenarios with high real-time requirements. Combining the separate axis algorithm with the fast collision detection algorithm can achieve efficient collision detection, thereby making the final parking route the globally optimal route.

[0045] In one embodiment of this application, searching for a parking path from the starting point to the ending point of parking planning based on a first collision detection result and a first arc connection line includes: if the first collision detection result is a collision detection failure, obtaining a new constraint state point in the state space, generating a new heuristic cost table based on the new constraint state point, and re-performing collision detection based on the new heuristic cost table until the collision detection result corresponding to the new heuristic cost table passes the collision detection; if the first collision detection result is a collision detection pass, or the collision detection result corresponding to the new heuristic cost table passes the collision detection; then creating a new state point, adding the new state point to the target expansion tree, and setting the parent node of the new state point as the current potential connection point; when there is a tangent directed arc on the ray from the current potential connection point to the parking space direction of the parking planning ending point, determining the directed arc as a second arc connection line, and searching for a parking path from the starting point to the ending point of parking planning based on the second arc connection line and the first arc connection line. If the collision detection fails, the sampling points are redefined in the target driving environment, and collision detection continues based on the new sampling points until a collision-free arc connection line is generated. This not only inherits the arc features of geometric planning and is compatible with the straight path of geometric planning (the arc radius can be infinitely large), but also facilitates the addition of vehicle kinematic constraints, can plan multiple routes, and can determine the globally optimal route among multiple routes.

[0046] In one embodiment of this application, searching for a parking path from the starting point to the ending point of the parking plan based on a second arc connecting line and a first arc connecting line includes: when the radius of the second arc connecting line is greater than or equal to a second preset radius threshold, and the path search time is less than a preset time threshold, generating a new second arc connecting line based on the next potential connection point to the parking plan ending point, until the radius of the new second arc connecting line is less than the second preset radius threshold; when the radius of the second arc connecting line is less than the second preset radius threshold, performing collision detection on the second arc connecting line to obtain a second collision detection result; if the second collision detection result is less than the second preset radius threshold, performing collision detection on the second arc connecting line to obtain a second collision detection result; if the second collision detection result is less than the second preset radius threshold, performing collision detection on the second arc connecting line to obtain a second collision detection result. If the collision detection result is a failure, and the path search time is less than a preset time threshold, a new second circular arc connection line is generated based on the next potential connection point to the parking planning endpoint. Collision detection is then performed again based on the radius of this new second circular arc connection line until the collision detection result corresponding to the new second circular arc connection line passes. If the second collision detection result passes, or the collision detection result corresponding to the new second circular arc connection line passes, then the first and second circular arc connection lines form a potential parking planning path. Dynamic constraints are then applied to this potential parking planning path to obtain the parking path. Collision detection, by considering environmental information, can generate collision-avoiding paths, ensuring the safety and feasibility of path planning and enabling vehicles to navigate and move safely in complex environments.

[0047] In one embodiment of this application, dynamic constraints are applied to potential parking planning paths to obtain parking paths, including: establishing a dynamic model for the target vehicle, performing a tracking simulation of the potential path starting from the starting point of the potential path, and determining candidate paths based on the simulation results; performing collision detection on the candidate paths to obtain a third collision detection result; if the third collision detection result is a successful collision detection, the candidate path is added to the candidate path list; if the third collision detection result is a failed collision detection, and the path search time is less than a preset time threshold, dynamic constraints are applied to a new potential parking planning path to obtain a new parking path, and the new parking path is added to the candidate path list; the cost of multiple candidate paths in the candidate path list is calculated to obtain multiple candidate path costs, and a weighted average of the multiple candidate costs is calculated to determine the candidate path with the lowest cost as the parking path. By establishing a dynamic model of the target vehicle, the kinematics and dynamic characteristics of the vehicle can be taken into account. After simulation, the target vehicle can follow the path according to the selected controller without violating the dynamic constraints of the vehicle. This ensures that the generated candidate path conforms to the actual physical constraints, so that the planned path satisfies the vehicle dynamic constraints and guarantees the smoothness of the path. Furthermore, the path with the minimum cost is the local optimum, making the parking path closer to the global optimum.

[0048] In one embodiment of this application, Figure 3 This is a schematic diagram of a smooth parking planning potential path as shown in this application. The parking planning potential path is formed by connecting the first and second circular arcs. In this embodiment, the starting point is the parking planning start point, and the ending point is the parking planning end point. The potential path is triggered by the child node corresponding to the parking planning end point. The process traces back from the parent node of the node corresponding to the state point to the child node corresponding to the parking planning end point. The connecting lines in the tracing process are extracted to form the parking planning potential path. The parking planning potential path is smoothed in a closed loop using a virtual controller. A dynamic model of the vehicle is established, and a simple controller, such as a pure-persuit controller, is used to trigger the potential path from the starting point. The potential path is then simulated to obtain the sequence of state variables output by the vehicle model during the simulation, such as position, orientation, and speed. This sequence is used as a candidate path. Closed-loop simulation smoothing is performed using a simulated controller to ensure that the planned path satisfies the vehicle dynamics constraints and guarantees the smoothness of the path. Furthermore, these paths satisfy the vehicle dynamics constraints, meaning that the vehicle can follow the selected controller during the simulation without violating the vehicle's dynamics limitations.

[0049] In one embodiment of this application, the cost of multiple candidate paths in the path list is calculated. The cost can take into account factors such as path length, number of gear shifts, curvature, etc. Finally, a weighted average of the cost of each path is calculated, and the path with the best cost is selected as the parking path output.

[0050] In one embodiment of this application, Figures 4(a) and 4(b) are schematic diagrams of automatic parking path planning at two different times in an exemplary embodiment of this application. As shown in Figure 4(a), the target expansion tree is the expansion tree, the parking planning start point is the start point, and the parking planning end point is the end point. The state space of the driving environment is established within a preset range based on the vehicle position and vehicle posture. The obstacle in the coordinate system is abstracted into a long polygon (4) and the parking space is abstracted into a convex polygon (3) through the environment reconstruction module. The current position of the target vehicle is taken as the parking planning start point (1) and the position of the parking space is taken as the parking planning end point (2). Discrete point sampling is performed in the driving environment to obtain sampling points (5). The target expansion tree is formed based on the sampling points (6). The state points in the state space are constrained to form the path envelope. Using the RRT algorithm, the heuristic cost is calculated for potential connection points within a preset range near sampling point (5). That is, the heuristic cost of potential connection points on the extended tree reaching the parking planning endpoint via the current sampling point is calculated. Based on the heuristic cost calculation results of all potential connection points, the parking path from the parking planning starting point to the parking planning endpoint is searched. As shown in Figure 4(b), the parking path is a planned path. A coordinate system for the driving environment is established within a preset range based on the vehicle position and vehicle posture. The obstacles in the coordinate system are abstracted into long polygons (4) and the parking spaces are abstracted into convex polygons (3) through the environment reconstruction module. The current position of the target vehicle is taken as the starting point of the parking planning (1) and the position of the parking space is taken as the ending point of the parking planning (2). Discrete point sampling is performed in the driving environment to obtain sampling points (5). A target extension tree (6) is formed based on the sampling points. After the potential parking planning path is formed by connecting the first and second circular arcs, the potential parking planning path is simulated in a closed loop through the virtual controller. A snapshot of the vehicle's movement (8) is obtained, and multiple candidate paths are obtained. The weighted average of the multiple candidate paths is calculated to obtain the planned path (7). By acquiring target information for parking planning, and abstractly modeling obstacles and parking spaces based on this information, the driving environment of the target vehicle is obtained. The current position of the target vehicle is determined as the starting point of parking planning, and the location of the parking space is taken as the ending point. Discrete point sampling is performed within the target vehicle's driving environment. A parking path is planned based on the sampling results, and this path is fed back to the target vehicle. This allows for flexible determination of the parking path based on the sampling results, theoretically resulting in an infinite number of possible geometric combinations for the planned parking path. Furthermore, the planning process is not limited by the initial or final posture of the target vehicle, offering flexibility and a high success rate for any posture. Multi-threaded planning is supported, improving planning efficiency. The parking path planned based on sampling results is closer to the global optimum, satisfies vehicle dynamics constraints, and ensures the smoothness of the parking path.

[0051] In one embodiment of this application, Figure 5 This application illustrates a flowchart of the automatic parking route planning process. Figure 5 As shown, it includes the following steps:

[0052] Step S501: Obtain target information for parking planning. Target information includes obstacles, target vehicle position, and target vehicle attitude. Then, proceed to step S502.

[0053] Step S502: Establish the driving environment of the target vehicle based on the target information, perform discrete point sampling in the driving environment to obtain multiple sampling points, form a target extension tree based on the multiple sampling points, and then execute step S503.

[0054] Step S503: Determine whether the path planning time is less than a preset time threshold. If it is less, proceed to step S504. If it is greater than or equal to the preset time threshold, proceed to step S515.

[0055] Step S504: Randomly sample a state point in the state space, and use other sampled points in the target expansion tree as potential connection points. Starting from the constrained state point, traverse all potential connection points in the target expansion tree, calculate the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point, and generate a heuristic cost table. Then execute step S505.

[0056] Step S505: Sort all potential connection points according to the heuristic cost table to form a potential connection point list; take the potential connection point corresponding to each heuristic cost as the starting point and the current state point as the ending point, perform a directed circular arc connection to obtain the first circular arc connection line, and then execute step S506.

[0057] Step S506: Determine whether the radius of the first arc connecting line is less than the first preset radius. If it is satisfied, proceed to step S507; otherwise, proceed to step S504.

[0058] Step S507: Perform collision detection on the first arc connecting line and determine whether the collision detection passes. If the collision detection passes, proceed to step S508; if the collision detection fails, proceed to step S505.

[0059] Step S508: Add the new state point to the target expansion tree and set its parent node as the potential connection point of the current loop. Then execute step S509.

[0060] Step S509: Determine whether the target expansion tree has been updated. If it has been updated, proceed to step S510. If the target expansion tree has not been updated, proceed to step S503.

[0061] Step S510: From the newly added sampling point to the parking space direction of the planned endpoint, determine whether there is a tangent directed arc on the ray, i.e. the second arc connection line, and then execute step S511.

[0062] Step S511: Determine whether the radius of the second arc connecting line is less than the second preset radius threshold. If it is less, proceed to step S512. If it is greater than or equal to the threshold, proceed to step S503.

[0063] Step S512: Perform collision detection on the second arc connecting line and determine whether the collision detection passes. If the collision detection passes, proceed to step S513; if the collision detection fails, proceed to step S503.

[0064] Step S513: The parking planning potential path is formed by connecting the first and second circular arcs. The virtual controller performs closed-loop smoothing on the parking planning potential path to obtain candidate paths. Then, step S514 is executed.

[0065] Step S514: Perform collision detection on the candidate path and determine whether the collision detection passes. If it passes, proceed to step S515; if the collision result fails, proceed to step S503.

[0066] Step S515: Add the candidate path to the candidate path list, calculate the cost of multiple candidate paths in the candidate path list, and output the parking path.

[0067] exist Figure 5 In the illustrated technical solution, obstacles and parking spaces are abstractly modeled based on target information to obtain the target vehicle's driving environment. By determining the target vehicle's current position as the starting point of parking planning and the parking space's location as the ending point, discrete point sampling is performed within the target vehicle's driving environment. A parking path is planned based on the sampling results and fed back to the target vehicle. This allows for flexible determination of the parking path based on the sampling results, theoretically enabling an infinite number of geometric combinations for the planned parking path. Furthermore, the planning process is not limited by the target vehicle's initial and final postures, offering flexibility and a high success rate for arbitrary postures. It also supports multi-threaded planning, improving planning efficiency. Planning a parking path based on sampling results ensures the path is closer to the global optimum, satisfies vehicle dynamics constraints, and guarantees the smoothness of the parking path.

[0068] Figure 6 This is a block diagram illustrating an automatic parking path planning device according to an exemplary embodiment of this application. Figure 6 As shown, this exemplary automated parking path planning system includes:

[0069] The system comprises an information acquisition module 610, a driving environment modeling module 620, and a path planning module 630. The information acquisition module 610 acquires target information for parking planning, including obstacles, parking spaces, the target vehicle's position, and its attitude. The driving environment modeling module 620 abstractly models obstacles and parking spaces based on the target information to obtain the target vehicle's driving environment. The path planning module 630 uses the target vehicle's current position as the starting point of parking planning and the parking space's location as the ending point; it also performs discrete point sampling of the target vehicle's driving environment, searches for a parking path from the starting point to the ending point based on the sampling results, and feeds the parking path back to the target vehicle. By acquiring target information for parking planning, and abstractly modeling obstacles and parking spaces based on this information, the driving environment of the target vehicle is obtained. The current position of the target vehicle is determined as the starting point of parking planning, and the location of the parking space is taken as the ending point. Discrete point sampling is performed within the target vehicle's driving environment. A parking path is planned based on the sampling results, and this path is fed back to the target vehicle. This allows for flexible determination of the parking path based on the sampling results, theoretically resulting in an infinite number of possible geometric combinations for the planned parking path. Furthermore, the planning process is not limited by the initial or final posture of the target vehicle, offering flexibility and a high success rate for any posture. Multi-threaded planning is supported, improving planning efficiency. The parking path planned based on sampling results is closer to the global optimum, satisfies vehicle dynamics constraints, and ensures the smoothness of the parking path.

[0070] In one embodiment of the present invention, the path planning module 630 is used to sample discrete points in the target vehicle's driving environment to obtain multiple sampling points; form a target extension tree based on the multiple sampling points, and take the parking planning start point as the root node of the target extension tree; determine whether the path search time is less than a preset time threshold; if the path search time is greater than or equal to the preset time threshold, calculate the cost of the multiple candidate paths searched, obtain the cost of multiple candidate paths, calculate the weighted average of the multiple candidate costs, and determine the candidate path with the minimum cost as the parking path; if the path search time is less than the preset time threshold, randomly sample state points in the state space, and constrain the state points according to preset conditions to obtain constrained state points; wherein, a state point is a sampling point in the target extension tree, and the state space is generated based on the target vehicle's position and attitude; take other sampling points in the target extension tree as potential connection points, start from the constrained state point, traverse the potential connection points in the target extension tree, calculate the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point, and search for the parking path from the parking planning start point to the parking planning endpoint based on the heuristic cost calculation results of all potential connection points.

[0071] In one embodiment of the present invention, the path planning module 630 is used to search for a parking path from the parking planning start point to the parking planning end point based on the heuristic cost calculation results of all potential connection points. It generates a heuristic cost table based on the heuristic cost calculation results of all potential connection points, sorts all potential connection points according to the heuristic cost table, and forms a potential connection point list. It then performs directed circular arc connections with each potential connection point corresponding to a heuristic cost as the start point and the current state point as the end point to obtain a first circular arc connection line. If the radius of the first circular arc connection line is less than a first preset radius threshold, it performs collision detection on the first circular arc connection line to obtain a first collision detection result. Based on the first collision detection result and the first circular arc connection line, it searches for a parking path from the parking planning start point to the parking planning end point. If the radius of the first circular arc connection line is greater than or equal to the first preset radius threshold, it re-acquires constraint state points and potential connection points, and generates a new first circular arc connection line using the re-acquired constraint state points and potential connection points, until the radius of the new first circular arc connection line is less than the first preset radius threshold.

[0072] In one embodiment of the present invention, the path planning module 630 is configured to: if the first collision detection result is a collision detection failure, obtain a new constraint state point in the state space, generate a new heuristic cost table based on the new constraint state point, and re-perform collision detection based on the new heuristic cost table until the collision detection result corresponding to the new heuristic cost table passes the collision detection; if the first collision detection result is a collision detection success, or the collision detection result corresponding to the new heuristic cost table passes the collision detection; create a new state point, add the new state point to the target expansion tree, and set the parent node of the new state point as the current potential connection point; when there is a tangent directed arc on the ray from the current potential connection point to the parking space direction of the parking planning endpoint, determine the directed arc as the second arc connection line, and search for the parking path from the parking planning start point to the parking planning endpoint based on the second arc connection line and the first arc connection line.

[0073] In one embodiment of the present invention, the path planning module 630 is used to generate a new second arc connecting line from the next potential connection point to the parking planning endpoint when the radius of the second arc connecting line is greater than or equal to a second preset radius threshold and the path search time is less than a preset time threshold, until the radius of the new second arc connecting line is less than the second preset radius threshold; when the radius of the second arc connecting line is less than the second preset radius threshold, collision detection is performed on the second arc connecting line to obtain a second collision detection result; if the second collision detection result is a collision failure, then when the path search time is less than the preset time threshold, a new second arc connecting line is generated from the next potential connection point to the parking planning endpoint, and collision detection is re-performed based on the radius of the new second arc connecting line until the collision detection result corresponding to the new second arc connecting line passes the collision detection; if the second collision detection result is a collision pass, or the collision detection result corresponding to the new second arc connecting line passes the collision detection; then a potential parking planning path is formed by the first arc connecting line and the second arc connecting line, and dynamic constraints are applied to the potential parking planning path to obtain the parking path.

[0074] In one embodiment of the present invention, the path planning module 630 establishes a dynamic model of the target vehicle, starts from the starting point of the potential path, performs a tracking simulation of the potential path, and determines candidate paths based on the simulation results; performs collision detection on the candidate paths to obtain a third collision detection result; if the third collision detection result is a successful collision detection, the candidate path is added to the candidate path list; if the third collision detection result is a failed collision detection and the path search time is less than a preset time threshold, dynamic constraints are applied to the new parking planning potential path to obtain a new parking path, and the new parking path is added to the candidate path list; the cost of multiple candidate paths in the candidate path list is calculated to obtain multiple candidate path costs, and the multiple candidate costs are weighted and averaged to determine the candidate path with the lowest cost as the parking path.

[0075] In one embodiment of the present invention, the information acquisition module is used to acquire target information for parking planning, including: sensing obstacles and parking spaces through sensing sensors; wherein the sensing sensors include ultrasonic sensors, millimeter-wave sensors, and cameras; and locating the target vehicle position and target vehicle attitude through positioning sensors.

[0076] It should be noted that the automatic parking path planning system and the automatic parking path planning method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the road condition refresh device provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0077] Embodiments of this application also provide an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the automatic parking path planning method provided in the above embodiments.

[0078] Figure 7 A schematic diagram of a computer system suitable for implementing the embodiments of this application is shown. It should be noted that... Figure 7 The computer system 700 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0079] As shown in Figure 4, the computer system 700 includes a Central Processing Unit (CPU) 701, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 702 or programs loaded from storage portion 708 into Random Access Memory (RAM) 703, such as performing the methods described in the above embodiments. The RAM 703 also stores various programs and data required for system operation. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An Input / Output (I / O) interface 705 is also connected to the bus 704.

[0080] The following components are connected to the I / O interface 705: an input section 606 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed.

[0081] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 711. When the computer program is executed by central processing unit (CPU) 401, it performs various functions defined in the system of this application.

[0082] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0083] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0084] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0085] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the automatic parking path planning method as described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not incorporated into the electronic device.

[0086] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the automatic parking path planning method provided in the various embodiments described above.

[0087] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. An automatic parking path planning method, characterized in that, The automatic parking path planning method includes: Obtain target information for parking planning; the target information includes: obstacles, parking spaces, target vehicle position, and target vehicle attitude. Based on the target information, the obstacles and parking spaces are abstractly modeled to obtain the target vehicle's driving environment; The current location of the target vehicle is used as the starting point for parking planning, and the location of the parking space is used as the ending point for parking planning. Discrete point sampling is performed on the target vehicle's driving environment. Based on the sampling results, a parking path from the parking planning start point to the parking planning end point is searched, and the parking path is fed back to the target vehicle. This process involves sampling multiple points in the target vehicle's driving environment; a target expansion tree is formed based on these multiple sampling points, with the parking planning start point as the root node; the path search time is determined to be less than a preset time threshold; if the path search time is greater than or equal to the preset time threshold, the cost of the multiple candidate paths is calculated, and a weighted average of these costs is calculated. The candidate path with the lowest cost is then determined as the parking path. If the path search time is less than a preset time threshold, then state points are randomly sampled in the state space and constrained according to preset conditions to obtain constrained state points; the state point is a sample point in the target expansion tree, and the state space is generated based on the target vehicle position and target vehicle attitude; other sample points in the target expansion tree are used as potential connection points, starting from the constrained state point, the potential connection points in the target expansion tree are traversed, the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point is calculated, and the parking path from the parking planning start point to the parking planning endpoint is searched based on the heuristic cost calculation results of all potential connection points.

2. The automatic parking path planning method according to claim 1, characterized in that, Based on the heuristic cost calculation results of all potential connection points, a parking path is searched from the starting point to the ending point of the parking planning process, including: A heuristic cost table is generated based on the heuristic cost calculation results of all potential connection points. All potential connection points are then sorted according to the heuristic cost table to form a potential connection point list. Starting from the potential connection point corresponding to each heuristic cost and ending at the current state point, a directed circular arc connection is formed to obtain the first circular arc connection line; If the radius of the first arc connecting line is less than the first preset radius threshold, then collision detection is performed on the first arc connecting line to obtain the first collision detection result, and, based on the first collision detection result and the first arc connecting line, a parking path from the parking planning start point to the parking planning end point is searched. If the radius of the first arc connecting line is greater than or equal to the first preset radius threshold, then the constraint state point and potential connection point are reacquired, and a new first arc connecting line is generated by the reacquired constraint state point and potential connection point until the radius of the new first arc connecting line is less than the first preset radius threshold.

3. The automatic parking path planning method according to claim 2, characterized in that, Based on the first collision detection result and the first arc connection line, a parking path is searched from the starting point to the ending point of the parking planning, including: If the first collision detection result is that the collision detection fails, a new constraint state point is obtained in the state space, and a new heuristic cost table is generated based on the new constraint state point. Then, the collision detection is performed again based on the new heuristic cost table until the collision detection result corresponding to the new heuristic cost table passes the collision detection. If the first collision detection result is a successful collision detection, or the collision detection result corresponding to the new heuristic cost table is a successful collision detection, then a new state point is created and added to the target expansion tree, and the parent node of the new state point is set as the current potential connection point; when there is a tangent directed arc on the ray from the current potential connection point to the parking space direction of the parking planning endpoint, the directed arc is determined as the second arc connection line, and the parking path from the parking planning start point to the parking planning endpoint is searched based on the second arc connection line and the first arc connection line.

4. The automatic parking path planning method according to claim 3, characterized in that, Based on the second and first circular arc connecting lines, a parking path is searched from the starting point to the ending point of the parking planning, including: When the radius of the second arc connecting line is greater than or equal to the second preset radius threshold and the path search time is less than the preset time threshold, a new second arc connecting line is generated based on the next potential connection point to the parking planning endpoint, until the radius of the new second arc connecting line is less than the second preset radius threshold. When the radius of the second arc connecting line is less than the second preset radius threshold, collision detection is performed on the second arc connecting line to obtain the second collision detection result; If the second collision detection result is a collision detection failure, then when the path search time is less than the preset time threshold, a new second arc connection line is generated based on the next potential connection point to the parking planning endpoint. The collision detection is then performed again based on the radius of the new second arc connection line until the collision detection result corresponding to the new second arc connection line passes the collision detection. If the second collision detection result is a successful collision detection, or the collision detection result corresponding to the new second arc connecting line is a successful collision detection, then a potential parking planning path is formed by the first arc connecting line and the second arc connecting line, and dynamic constraints are applied to the potential parking planning path to obtain the parking path.

5. The automatic parking path planning method according to claim 4, characterized in that, The parking planning potential path is subjected to dynamic constraints to obtain the parking path, including: A dynamic model is established for the target vehicle. Starting from the starting point of the potential path, a line-following simulation is performed on the potential path, and a candidate path is determined based on the simulation results. Collision detection is performed on the candidate paths to obtain a third collision detection result; If the third collision detection result is that the collision detection is successful, then the candidate path is added to the candidate path list; If the third collision detection result is a collision detection failure and the path search time is less than a preset time threshold, dynamic constraints are applied to the new parking planning potential path to obtain a new parking path, and the new parking path is added to the candidate path list. The cost of multiple candidate paths in the candidate path list is calculated to obtain the cost of multiple candidate paths. The multiple candidate costs are then weighted and averaged to determine the candidate path with the lowest cost as the parking path.

6. The automatic parking path planning method according to claim 1, characterized in that, Obtain target information for parking planning, including: The obstacle and the parking space are detected by sensing sensors; wherein the sensing sensors include ultrasonic sensors, millimeter-wave sensors, and cameras; The location and attitude of the target vehicle are determined by positioning sensors.

7. An automatic parking path planning method device, characterized in that, The automatic parking path planning method device includes: The information acquisition module is used to acquire target information for parking planning; the target information includes: obstacles, parking spaces, target vehicle position, and target vehicle attitude. The driving environment modeling module is used to abstractly model obstacles and parking spaces based on target information to obtain the driving environment of the target vehicle; The path planning module uses the current position of the target vehicle as the starting point and the location of the parking space as the ending point. It performs discrete point sampling on the target vehicle's driving environment, searches for a parking path from the starting point to the ending point based on the sampling results, and feeds the parking path back to the target vehicle. This process involves sampling multiple points in the target vehicle's driving environment, forming a target expansion tree based on these sampling points, and using the parking planning starting point as the root node of the target expansion tree. The module determines whether the path search time is less than a preset time threshold. If the path search time is greater than or equal to the preset time threshold, it calculates the cost of the multiple candidate paths, obtaining multiple candidate path costs, and then processes these multiple candidate costs. A weighted average calculation is used to determine the candidate path with the lowest cost as the parking path. If the path search time is less than a preset time threshold, state points are randomly sampled in the state space and constrained according to preset conditions to obtain constrained state points. Each state point is a sample point in the target expansion tree, and the state space is generated based on the target vehicle position and attitude. Other sample points in the target expansion tree are used as potential connection points. Starting from the constrained state point, the potential connection points in the target expansion tree are traversed, and the heuristic cost of all potential connection points reaching the parking planning endpoint through the current constrained state point is calculated. Based on the heuristic cost calculation results of all potential connection points, a parking path from the parking planning start point to the parking planning endpoint is searched.

8. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the automatic parking path planning method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by the computer's processor, causes the computer to perform the automatic parking path planning method as described in any one of claims 1 to 6.

Images