Lane line constraint-based skeleton-guided hierarchical parking path planning method and device

Abstract

The invention relates to a skeleton guidance layered parking path planning method and device based on lane line constraints, and the method comprises the steps: obtaining lane line information and parking space information corresponding to a target real parking lot, so as to determine the lane constraints, and constructing a corresponding target grid map based on the lane constraints; parking information in the target grid map and the minimum turning radius of the to-be-parked vehicle are acquired to determine the optimal parking point of the to-be-parked vehicle; based on a preset improved skeletonization algorithm and lane constraints, skeletonization processing is carried out on the target grid map to obtain a skeleton map corresponding to the target grid map, a feasible path of the to-be-parked vehicle during parking is determined according to the optimal parking point, the skeleton map and the minimum turning radius, and the feasible path is tested and optimized to obtain the target grid map. Therefore, complex parking scenes such as a broken road under the constraint of a narrow lane can be effectively handled, and the real-time performance, the optimality and the safety of path planning are realized.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a skeleton-guided hierarchical parking path planning method and device based on lane line constraints. Background Art

[0002] With the rapid development of autonomous driving technology, autonomous valet parking (AVP) systems have become an important breakthrough in the field of smart travel.

[0003] Path planning technology plays a core role in the AVP system, with its performance directly determining the safety, reliability, and user experience of the parking process. Path planning in complex parking environments presents multiple challenges, requiring not only traditional obstacle avoidance and vehicle dynamics constraints but also adherence to strict lane restrictions and traffic regulations. Especially in structured parking environments, lane markings, which serve as a guide and safety barrier for vehicle movement, are crucial for the reliable operation of the AVP system.

[0004] After years of development, path planning algorithms have evolved into three broad branches: planning methods based on geometric curves (such as B-spline curves and Reeds-Shepp curves), planning methods based on random sampling (such as RRT and the RRT algorithm), and planning methods based on graph search (such as A* and the Hybrid A* algorithm). Each method has its own advantages. For example, geometric curve methods ensure smooth trajectories, sampling methods are adaptable to unstructured environments, and graph search algorithms excel at achieving global optimality.

[0005] However, in real-world parking environments, especially those with complex parking layouts like dead-ends, a single technology struggles to simultaneously meet the multiple requirements of path planning. Traditional methods often focus solely on obstacle avoidance or trajectory smoothing, overlooking the core role of lane constraints in real-world parking situations. This makes deployment difficult in real-world applications. Existing technologies struggle to generate feasible parking trajectories for dead-end parking spaces, particularly those at the end of lanes requiring multiple reverse maneuvers to access.

[0006] In automated parking scenarios, existing path planning algorithms still face three core challenges: first, ensuring accurate obstacle avoidance during parking; second, optimizing the global efficiency of path planning while strictly adhering to lane constraints; and third, addressing the problem of generating feasible paths in complex scenarios such as dead-end parking spaces.

[0007] In terms of safe path generation, existing technologies primarily achieve collision-free trajectory generation through multi-constraint modeling. For example, existing technologies can use circular curve-based parking path planning frameworks to ensure safety by integrating obstacle collision constraints with vehicle dynamics constraints. However, this path occupies a large space. Existing technologies can also use cubic B-spline curves combined with gradient descent algorithms to generate continuous curvature paths in narrow parking spaces, improving path smoothness. These safety-focused studies primarily focus on static obstacle avoidance, and lack the ability to utilize lane markings as driving regulatory constraints.

[0008] In terms of efficiency optimization, existing technologies focus on algorithm architecture innovation and computing resource compression. For example, by integrating the A* algorithm, Bezier curves and model predictive control (MPC), the number of search nodes and the real-time computing burden can be reduced. Existing technologies can also design a hierarchical planning framework that combines nonlinear optimization and improved RRT* algorithms with the Reeds-Shepp curve (RS curve), taking into account both global path planning and local trajectory optimization. In addition, existing technologies can also use a method that combines Voronoi diagrams with Hybrid A* to quickly generate safe paths using spatial segmentation characteristics. For the above-mentioned optimization schemes, the computational stability and path smoothness in complex environments still need to be verified.

[0009] To address lane constraints and adaptability to narrow and complex scenarios (such as dead-end parking spaces), some research focuses on optimizing specific scenarios. For example, existing technologies improve parking efficiency by reducing the number of perpendicular parking maneuvers, or they can use hierarchical motion planning methods to achieve avoidance during cruising and straight-ahead phases in crowded parking lots. Furthermore, existing technologies can also incorporate geometric curve manipulation to achieve straight-line parking in narrow spaces. While these studies have improved performance in specific operating conditions, none have effectively addressed extreme scenarios such as dead-end parking spaces, which combine spatial constraints and topological anomalies. Furthermore, dynamic tracking capabilities for lane constraints remain limited.

[0010] In summary, while existing technologies have achieved some improvements in obstacle avoidance and algorithm efficiency, they neglect the core role of lane lines as a carrier of driving regulations. For example, in real parking lot scenarios, failure to fully consider lane lines can easily lead to conflicts when vehicles are driving, triggering multiple path replannings and wasting time. Furthermore, there is a lack of consideration for the integration of complex "dead-end" parking spaces with lane planning. Breakthrough progress is still needed in balancing vehicle safety, path planning efficiency, scenario adaptability, and the integration of lane line constraints. Summary of the Invention

[0011] The present application provides a skeleton-guided hierarchical parking path planning method and device based on lane line constraints to address the problems of the existing technology ignoring the core role of lane lines as the carrier of driving regulations, and lacking consideration for the problem of combining complex "dead-end roads" parking spaces with lane planning, making it difficult to solve problems such as vehicle safety, path planning efficiency, scene adaptability and lane line constraints.

[0012] A first embodiment of the present application provides a skeleton-guided hierarchical parking path planning method based on lane line constraints, comprising the following steps: obtaining lane line information and parking space information corresponding to a target real parking lot, and determining corresponding lane constraints based on the lane line information and the parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints; obtaining parking information and a minimum turning radius of a vehicle to be parked in the target grid map, and determining an optimal parking entry point for the vehicle to be parked based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking starting point, a parking end point, parking space information, and a parking space type; skeletonizing the target grid map based on a preset improved skeletonization algorithm and the lane constraints to obtain a skeleton map corresponding to the target grid map, determining a feasible path for parking the vehicle to be parked based on the optimal parking entry point, the skeleton map, and the minimum turning radius, and testing and optimizing the feasible path to obtain an optimal parking path corresponding to the vehicle to be parked.

[0013] Optionally, in one embodiment of the present application, obtaining parking information and a minimum turning radius of the vehicle to be parked in the target grid map, and determining an optimal parking point for the vehicle to be parked based on the parking information and the minimum turning radius, includes: obtaining AVP scenario information corresponding to the lane constraint, and obtaining parking endpoint coordinates of parking endpoints for each parking space in the target grid map based on the AVP scenario information; using the parking endpoint coordinates as an initial position and the posture of the vehicle to be parked as an initial direction, drawing a straight line segment of a preset length starting from the initial position and along the initial direction, and using the straight line segment as an initial path; using the minimum turning radius of the vehicle to be parked as a target radius, calculating a center point of a parking arc segment based on the target radius, the parking side of the vehicle parking method, and the parking endpoint, and calculating the optimal parking point for the vehicle to be parked based on the center point, the initial path, and a preset RS curve algorithm.

[0014] Optionally, in one embodiment of the present application, the target grid map is skeletonized based on the preset improved skeletonization algorithm and the lane constraint to obtain a skeleton map corresponding to the target grid map, and a feasible path for the vehicle to be parked is determined based on the optimal parking point, the skeleton map and the minimum turning radius, including: based on the optimal parking point, the vehicle to be parked is planned to be in the nearest forward lane, and according to the nearest forward lane and the improved skeletonization algorithm corresponding to the target grid map, the parking lot obstacle layer corresponding to the target grid map is optimized to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles; lane line obstacles are added to the target grid map, and based on the parking space Obstacles and lane line obstacles are extracted, skeleton intersections are constructed to construct the skeleton map, and multiple lane-level guidance points are obtained through the skeleton map, and a global guidance path is generated based on the multiple lane-level guidance points and a preset A* algorithm; multiple minimum turning radius constraint circles are constructed with each key node in the global guidance path as the center, and the intersection of each minimum turning radius constraint circle and the global guidance path is determined, and guidance path nodes are determined through the intersection points to construct an optimized guidance path that meets preset vehicle steering characteristics based on the guidance path nodes; each node in the optimized guidance path is used as a target point, and the target point is expanded to generate a feasible path that meets preset vehicle kinematic requirements based on the expanded target point and in combination with the RS curve algorithm.

[0015] Optionally, in one embodiment of the present application, the testing and optimization of the feasible path to obtain the optimal parking path corresponding to the vehicle to be parked includes: based on a preset global collision strategy, performing collision test processing on the feasible nodes in the feasible path to obtain corresponding detection results, and judging whether the current feasible path can avoid the parking space obstacle and the parking line obstacle according to the detection results; if the current feasible path can avoid the parking space obstacle and the parking line obstacle, determining the current feasible path as the optimal parking path for the vehicle to be parked; if the current feasible path cannot avoid the parking space obstacle and the parking line obstacle, performing topological structure optimization on the path segment where the collision occurs to obtain a corresponding reconstructed path topology, and using a preset Hybrid A* algorithm to replan the reconstructed path topology to obtain a collision-free optimal parking path.

[0016] According to a second aspect of the present application, an embodiment provides a skeleton-guided hierarchical parking path planning device based on lane line constraints, comprising: a map construction module for obtaining lane line information and parking space information corresponding to a target real parking lot, and determining corresponding lane constraints based on the lane line information and the parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints; a calculation module for obtaining parking information and a minimum turning radius of a parked vehicle in the target grid map, and determining an optimal parking entry point for the parked vehicle based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking starting point, a parking end point, parking space information, and a parking space type; and a path planning module for skeletonizing the target grid map based on a preset improved skeletonization algorithm and the lane constraints to obtain a skeleton map corresponding to the target grid map, determining a feasible path for the parked vehicle based on the optimal parking entry point, the skeleton map, and the minimum turning radius, and testing and optimizing the feasible path to obtain an optimal parking path for the parked vehicle.

[0017] Optionally, in one embodiment of the present application, the calculation module includes: an acquisition unit, configured to acquire AVP scenario information corresponding to the lane constraint, and acquire the parking end point coordinates of the parking end point of each parking space in the target grid map based on the AVP scenario information; a setting unit, configured to use the parking end point coordinates as an initial position and the posture of the vehicle to be parked as an initial direction, to draw a straight line segment of a preset length starting from the initial position along the initial direction, and to use the straight line segment as an initial path; and a calculation unit, configured to use the minimum turning radius of the vehicle to be parked as a target radius, and to calculate the center point of a parking arc segment based on the target radius, the parking side of the vehicle parking method, and the parking end point, so as to calculate the optimal parking point of the vehicle to be parked based on the center point, the initial path, and a preset RS curve algorithm.

[0018] Optionally, in one embodiment of the present application, the path planning module includes: an optimization unit for planning the vehicle to be parked to the nearest forward lane based on the optimal parking point, and optimizing the parking lot obstacle layer corresponding to the target grid map according to the improved skeletonization algorithm corresponding to the nearest forward lane and the target grid map, so as to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles; an extraction unit for adding lane line obstacles to the target grid map, and extracting skeleton intersections based on the parking space obstacles and the lane line obstacles to construct the skeleton map, and obtaining multiple lane-level guidance paths through the skeleton map. points, and generates a global guidance path based on the multiple lane-level guidance points and a preset A* algorithm; a construction unit, used to construct multiple minimum turning radius constraint circles with each key node in the global guidance path as the center, and determine the intersection of each minimum turning radius constraint circle and the global guidance path, and determine the guidance path nodes through the intersection, so as to construct an optimized guidance path that meets the preset vehicle steering characteristics according to the guidance path nodes; a generation unit, used to use each node in the optimized guidance path as a target point, and expand the target point, so as to generate a feasible path that meets the preset vehicle kinematic requirements based on the expanded target point and in combination with the RS curve algorithm.

[0019] Optionally, in one embodiment of the present application, the path planning module further includes: a judgment unit, configured to perform collision test processing on the feasible nodes in the feasible path based on a preset global collision strategy to obtain corresponding detection results, and to judge whether the current feasible path can avoid the parking space obstacle and the parking line obstacle based on the detection results; a determination unit, configured to determine the current feasible path as the optimal parking path for the vehicle to be parked if the current feasible path can avoid the parking space obstacle and the parking line obstacle; and a replanning unit, configured to perform topological structure optimization on the path segment where the collision occurs to obtain a corresponding reconstructed path topology if the current feasible path cannot avoid the parking space obstacle and the parking line obstacle, and to perform path replanning on the reconstructed path topology using a preset Hybrid A* algorithm to obtain a collision-free optimal parking path.

[0020] An embodiment of the third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the skeleton-guided hierarchical parking path planning method based on lane line constraints as described in the above embodiment.

[0021] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned skeleton-guided hierarchical parking path planning method based on lane line constraints.

[0022] The fifth aspect of the present application provides a computer program product, including a computer program, which is executed to implement the above-mentioned skeleton-guided hierarchical parking path planning method based on lane line constraints.

[0023] Therefore, the embodiments of the present application have the following beneficial effects:

[0024] The embodiments of the present application can construct a target grid map corresponding to the target real parking lot by acquiring lane line information and parking space information corresponding to the target real parking lot and determining corresponding lane constraints based on the lane line information and parking space information. The embodiments also acquire parking information and the minimum turning radius of the vehicle to be parked from the target grid map and determine the optimal parking entry point for the vehicle to be parked based on the parking information and the minimum turning radius, wherein the parking information includes the vehicle parking method, parking starting point, parking end point, parking space information, and parking space type. The target grid map is skeletonized based on a preset improved skeletonization algorithm and lane constraints to obtain a skeleton map corresponding to the target grid map. A feasible path for the vehicle to be parked is determined based on the optimal parking entry point, the skeleton map, and the minimum turning radius. The feasible path is then tested and optimized to obtain the optimal parking path for the vehicle to be parked. The present application can effectively address complex parking scenarios such as dead-end roads under narrow lane constraints, thereby achieving real-time, optimal, and safe path planning. This solves the problem that the existing technology ignores the core role of lane lines as the carrier of driving regulations, lacks consideration for the problem of combining parking spaces with lane planning in complex "dead-end roads", and is difficult to effectively integrate multi-objective collaborative optimization and lane line constraints.

[0025] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0027] Figure 1 A flowchart of a lane-constrained skeleton-guided hierarchical parking path planning method provided in accordance with an embodiment of the present application;

[0028] Figure 2 A schematic diagram of a grid map corresponding to a parking lot scene based on lane constraints provided in one embodiment of the present application;

[0029] Figure 3 A schematic diagram of the execution logic of a lane-constrained skeleton-guided hierarchical parking path planning method provided by one embodiment of the present application;

[0030] Figure 4 A schematic diagram of a lane-level entry and exit path planning effect provided in one embodiment of the present application;

[0031] Figure 4 (a) is a schematic diagram of a path planning effect corresponding to a corner parking space provided by an embodiment of the present application;

[0032] Figure 4 (b) is a schematic diagram of a path planning effect corresponding to a parking space near an intersection provided by an embodiment of the present application;

[0033] Figure 4 (c) is a schematic diagram of a path planning effect corresponding to a conventional parking space provided by an embodiment of the present application;

[0034] Figure 4 (d) is a schematic diagram of a path planning effect corresponding to a parking space near an exit provided by an embodiment of the present application;

[0035] Figure 5 A schematic diagram of a global path effect provided for an embodiment of the present application;

[0036] Figure 6 A schematic diagram of lane-constrained entry and exit guidance points corresponding to a grid map provided in one embodiment of the present application;

[0037] Figure 7 A schematic diagram of an improved curve skeleton provided in one embodiment of the present application;

[0038] Figure 8 A schematic diagram of a guiding path effect provided by an embodiment of the present application;

[0039] Figure 9 A schematic diagram of a conventional parking space entry and exit effect based on lane constraints provided in one embodiment of the present application;

[0040] Figure 9 (a) is a schematic diagram of a conventional parking space entry effect based on lane constraints provided by an embodiment of the present application;

[0041] Figure 9 (b) is a schematic diagram of a conventional parking space exit effect based on lane constraints provided by an embodiment of the present application;

[0042] Figure 10A schematic diagram of the effect of entering and exiting a parking space at a "dead-end road" based on lane constraints provided in one embodiment of the present application;

[0043] Figure 10 (a) is a schematic diagram of parking space entry into a "dead-end road" based on lane constraints, provided in one embodiment of the present application;

[0044] Figure 10 (b) is a schematic diagram of a parking exit effect of a "dead-end road" parking space based on lane constraints provided by an embodiment of the present application;

[0045] Figure 11 This is an example diagram of a skeleton-guided hierarchical parking path planning device based on lane line constraints according to an embodiment of the present application;

[0046] Figure 12 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0047] Among them, 10-skeleton-guided hierarchical parking path planning device based on lane line constraints; 100-map construction module, 200-computing module, 300-path planning module; 1201-memory, 1202-processor, 1203-communication interface. DETAILED DESCRIPTION

[0048] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0049] The following describes a skeleton-guided hierarchical parking path planning method and apparatus based on lane line constraints according to an embodiment of the present application with reference to the accompanying drawings. To address the problems mentioned in the above background technology, the present application provides a skeleton-guided hierarchical parking path planning method based on lane line constraints. In this method, lane line information and parking space information corresponding to a target real parking lot are obtained, and corresponding lane constraints are determined based on the lane line information and parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints. Parking information and a minimum turning radius of a vehicle to be parked in the target grid map are obtained, and an optimal parking entry point for the vehicle to be parked is determined based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking start point, a parking end point, parking space information, and a parking space type. Based on a preset improved skeletonization algorithm and lane constraints, the target grid map is skeletonized to obtain a skeleton map corresponding to the target grid map. A feasible path for the vehicle to be parked is determined based on the optimal parking entry point, the skeleton map, and the minimum turning radius, and the feasible path is tested and optimized to obtain an optimal parking path corresponding to the vehicle to be parked. This application can effectively address complex parking scenarios such as dead-end roads with narrow lane constraints, thereby achieving real-time, optimal, and safe path planning. This addresses the existing problems of neglecting the core role of lane lines as a vehicle for driving regulations, lacking consideration for complex "dead-end road" parking spaces combined with lane planning, and failing to effectively integrate multi-objective collaborative optimization and lane line constraints.

[0050] Specifically, Figure 1 This is a flowchart of a skeleton-guided hierarchical parking path planning method based on lane line constraints provided in an embodiment of the present application.

[0051] like Figure 1 As shown, the lane line constraint-based skeleton-guided hierarchical parking path planning method includes the following steps:

[0052] In step S101, lane line information and parking space information corresponding to the target real parking lot are obtained, and corresponding lane constraints are determined according to the lane line information and parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints.

[0053] The embodiments of the present application may first convert the real parking environment into a digital grid form through a grid map, and simultaneously use a high-resolution grid map to represent the environment, so as to establish a parking environment map containing lane lines and parking space information.

[0054] Specifically, the embodiment of the present application takes into account the lane guidance characteristics of a real parking lot. When constructing the map, the actual environment is converted into map information suitable for the planning algorithm. The embodiment of the present application can establish a parking lot model based on the vertical parking space scene and set a parking lot with a size of 102m*102m at a resolution of 0.1m. Figure 2 As shown, the parking lot entrance and exit are in the upper right corner and lower right corner of the map respectively. There are 236 parking spaces in total. The size of each parking space is set to 3*6m. There are five "dead-end" parking spaces as shown in the red circle in the figure. The lane lines are as follows Figure 2 The yellow dashed line in the middle represents the lane direction. The black arrow in the figure indicates the lane's direction of travel. When a vehicle is traveling on a road section, if the planned route passes through the opposite lane of the current road section, it is considered to have entered a dangerous area. The detailed vehicle simulation parameter data used in the experiment is shown in Table 1.

[0055] Table 1

[0056]

[0057] It should be noted that the parking information in the embodiment of the present application includes but is not limited to the vehicle parking method, parking starting point, parking end point, and parking space information (parking space size, coordinates, direction and type (whether it is a "dead-end" parking space)); among them, the parking starting point indicates the position corresponding to the parking lot entrance, and the parking end point indicates the coordinates of the parking space where the vehicle to be parked is finally parked (one parking space corresponds to one parking end point). The vehicle parking methods include perpendicular parking space parking, parallel parking space parking and diagonal parking space parking. In the embodiment of the present application, since the grid map is constructed based on a perpendicular parking lot with lane constraints, the subsequent vehicle parking methods are all perpendicular parking space parking.

[0058] In step S102, parking information and a minimum turning radius of the vehicle to be parked are obtained from the target grid map, and an optimal parking point for the vehicle to be parked is determined based on the parking information and the minimum turning radius. The parking information includes the vehicle parking method, the parking starting point, the parking end point, parking space information, and the parking space type.

[0059] Furthermore, if Figure 3 As shown, the embodiment of the present application also needs to determine the optimal parking point for the vehicle to be parked based on the parking information of the parking lot grid map with lane constraints determined above (mainly including the vehicle parking method, parking starting point, parking end point and parking space information, etc.), vehicle entry and exit decisions (the following will take garage entry as an example. The principle of vehicle exit is the same as that of garage entry and will not be repeated here), lane constraints and the minimum turning radius of the vehicle to be parked.

[0060] During the actual implementation process, the embodiments of the present application can determine the target posture of the vehicle to be parked according to the parking method and the parking end point, and use the RS curve algorithm to plan the path from the parking start point to the target end point according to the target posture, the parking start point, the lane constraints, and the minimum turning radius of the vehicle to be parked and the minimum turning radius, so as to plan from the current position to the nearest point in the forward lane after obtaining the optimal parking point of the vehicle, and use it as the starting point of the global path planning.

[0061] Optionally, in one embodiment of the present application, parking information and a minimum turning radius of the vehicle to be parked in a target grid map are obtained, and an optimal parking point for the vehicle to be parked is determined based on the parking information and the minimum turning radius, including: obtaining AVP scenario information corresponding to the lane constraint, and obtaining the parking end point coordinates of the parking end points of each parking space in the target grid map based on the AVP scenario information; using the parking end point coordinates as an initial position and the posture of the vehicle to be parked as an initial direction, drawing a straight line segment of a preset length starting from the initial position and along the initial direction, and using the straight line segment as an initial path; using the minimum turning radius of the vehicle to be parked as a target radius, and calculating the center point of a parking arc segment based on the target radius and the parking side and parking end point of the vehicle parking method, so as to calculate the optimal parking point for the vehicle to be parked based on the center point, the initial path, and a preset RS curve algorithm.

[0062] It should be noted that the embodiments of the present application can obtain the parking end coordinates of each parking space based on the AVP scenario information of lane constraints, and then use the coordinates of the parking end as the initial position and the posture of the vehicle to be parked as the initial direction, and use the RS curve algorithm to plan the path from the parking starting point to the target end point.

[0063] As an achievable method, the embodiment of the present application can start from the coordinates of the parking end point and draw a straight line segment of a preset length along the initial direction as the initial path; using the minimum turning radius of the vehicle to be parked as the radius, the center of the parking arc segment is calculated based on the parking side and the parking end point of the vehicle parking method; at the same time, the RS curve is used to calculate the optimal parking point, and collision detection is performed. If a collision occurs, the distance of the straight line out of the garage is increased until no collision occurs.

[0064] Therefore, the embodiments of the present application can use the RS curve algorithm to calculate the optimal parking point based on the parking starting point coordinates and the minimum turning radius based on the AVP scenario information of the lane constraints.

[0065] In step S103, the target grid map is skeletonized based on the preset improved skeletonization algorithm and lane constraints to obtain a skeleton map corresponding to the target grid map. A feasible parking path for the vehicle to be parked is determined based on the optimal parking point, the skeleton map, and the minimum turning radius. The feasible path is tested and optimized to obtain the optimal parking path for the vehicle to be parked.

[0066] Furthermore, embodiments of the present application can determine feasible parking paths for a vehicle based on the vehicle's optimal parking point, a skeleton map corresponding to the parking grid map with lane constraints, and the vehicle's minimum turning radius. These feasible paths are then tested and optimized to obtain the optimal parking path for the vehicle. The skeleton map is obtained by skeletonizing the parking grid map using a preset improved skeletonization algorithm, while also optimizing the skeletonization algorithm by adding lane line constraints.

[0067] It can be understood that the embodiments of the present application, through the hierarchical parking path planning algorithm that takes lane line constraints into consideration, significantly reduces the calculation time while maintaining the optimality of the path compared to traditional methods that do not consider lane line constraints, thereby achieving the goal of quickly and accurately providing the optimal path in complex parking scenarios based on lane constraints. In particular, in dead-end road scenarios based on lane constraints, the embodiments of the present application demonstrate excellent performance, can effectively cope with the challenges of spatial constraints and kinematic constraints, and provide an effective solution for the practical application of autonomous parking systems; thus, the embodiments of the present application not only improve the efficiency of path planning, but also ensure that high-quality path planning results can be provided in various parking scenarios based on lane constraints, including scenarios that pose a challenge to traditional methods that do not consider lane constraints.

[0068] Optionally, in one embodiment of the present application, based on a preset improved skeletonization algorithm and lane constraints, the target grid map is skeletonized to obtain a skeleton map corresponding to the target grid map, and a feasible path for the vehicle to be parked is determined based on the optimal parking point, the skeleton map and the minimum turning radius, including: based on the optimal parking point, the vehicle to be parked is planned to be placed in the nearest forward lane, and according to the nearest forward lane and the improved skeletonization algorithm corresponding to the target grid map, the parking lot obstacle layer corresponding to the target grid map is optimized to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles; lane line obstacles are added to the target grid map, and based on the parking space obstacles, a parking space obstacle layer ... and lane line obstacles, extract skeleton intersections to construct a skeleton map, and obtain multiple lane-level guidance points through the skeleton map, and generate a global guidance path based on the multiple lane-level guidance points and the preset A* algorithm; construct multiple minimum turning radius constraint circles with each key node in the global guidance path as the center, and determine the intersection of each minimum turning radius constraint circle and the global guidance path, and determine the guidance path nodes through the intersection points, so as to construct an optimized guidance path that meets the preset vehicle steering characteristics according to the guidance path nodes; use each node in the optimized guidance path as a target point, and expand the target point to generate a feasible path based on the expanded target point and in combination with the RS curve algorithm and the A* algorithm.

[0069] In the specific implementation process, the embodiment of the present application can determine the planning from the optimal parking point to the corresponding lane based on the vehicle's optimal parking point and the starting point of the global path planning, obtain the position of the vehicle's parking entry / exit point through the parking space information, and plan the vehicle to the nearest forward lane in combination with the parking lot map and the vehicle kinematic model.

[0070] Specifically, after obtaining the optimal parking point for the vehicle, the embodiment of the present application can plan from the current position to the nearest point in the oncoming lane as the starting point of the global path planning. The vehicle is planned from the current position to the nearest point in the oncoming lane, and this node is used as the starting point / end point of the global path planning, so that the vehicle can take lane factors into consideration in advance when planning in the next stage. The partial results of the lane-level parking entry / exit path planning are as follows: Figure 4 As shown, the corner parking spaces, parking spaces near intersections, regular parking spaces and parking spaces near exits are respectively as follows Figure 4 (a) Figure 4 (b) Figure 4 (c) and Figure 4 The pseudo code of the parking path planning stage is shown in the following table:

[0071] Table 2

[0072]

[0073]

[0074] Secondly, the embodiment of the present application can optimize the parking lot obstacle layer based on the improved curve skeleton corresponding to the grid map with lane constraints and the vehicle is planned to be in the nearest forward lane, so as to treat all parking spaces as obstacles and add lane line obstacles, further reducing the generation of redundant skeleton nodes, and obtaining a series of lane-level guidance points by extracting skeleton intersections, and using the A* algorithm to generate a global guidance path, such as Figure 5 shown.

[0075] Specifically, the embodiment of the present application can establish a topology guidance node that meets the vehicle kinematic characteristics by incorporating lane markings and parking areas into obstacle constraints based on the skeleton improvement algorithm, further reducing the generation of redundant skeleton nodes, and obtaining the following by extracting the skeleton intersection points: Figure 6 A series of lane-level guidance points are shown, wherein the skeleton intersection point is a skeleton node between the parking entry point and the parking start point and located on the curve skeleton based on the lane constraint. The target skeleton node may include a turning node, an intersection node, and a terminal point that maintains path connectivity. In embodiments of the present application, the criteria for extracting the target skeleton node are based on path connectivity, spatial decomposition efficiency, safety and obstacle avoidance, lane constraints, and vehicle kinematic constraints.

[0076] Therefore, the embodiment of the present application uses the above-mentioned preprocessing mechanism to make the skeleton-generated guidance nodes meet the lane-level path planning requirements, reduce the number of invalid guidance points, and significantly reduce the search space of the A* algorithm (the high-precision grid map environment in the embodiment of the present application reduces the search nodes by 103 times compared with the traditional method), and the generated guidance path strictly follows the lane constraints to ensure the compliance of the vehicle parking / exit process trajectory, improving the skeleton map such as Figure 7 shown.

[0077] Afterwards, the embodiments of the present application can obtain a global guidance path based on the use of the A* algorithm, and construct a minimum turning radius constraint circle with each intermediate node of the global guidance path as the center, and determine the guidance path nodes by finding the intersection of these circles and the global guidance path, thereby constructing an optimized guidance path that conforms to the vehicle's steering characteristics.

[0078] It should be noted that, in the embodiment of the present application, the starting point of the vehicle's optimal turning path is often at the minimum turning radius from the lane inflection point (i.e., the vehicle's minimum turning radius). Therefore, when performing kinematic path planning, a circle is drawn with each node of the global guidance path as the center and the vehicle's minimum turning radius, such as Figure 8 The red dots shown in the figure represent the intersection points. A guidance path is constructed by finding the points that intersect with the global guidance path.

[0079] Specifically, in the kinematic path planning stage, the embodiment of the present application constructs a minimum turning radius constraint circle with each intermediate node of the global path as the center, determines the guidance path node by calculating the intersection of these circles and the global path, and then generates an optimized guidance path that meets the vehicle steering characteristics. The guidance path result is as follows: Figure 8 The pseudo code of the A* algorithm is shown in Table 3:

[0080] Table 3

[0081]

[0082] Then, the embodiment of the present application can expand each node of the obtained optimized guidance path that meets the vehicle steering characteristics as the target point, and use the RS curve to guide the A* algorithm to generate a feasible path that meets the kinematics.

[0083] It is understood that in the embodiment of the present application, the target node can be expanded using the RS curve algorithm to ensure that the path not only meets the vehicle's kinematic constraints but also can safely and effectively navigate in a complex parking environment with lane constraints. The specific expansion process is as follows:

[0084] In an embodiment of the present application, starting from the first target node of the guidance path, an RS curve path segment is inserted between two adjacent target nodes in sequence (i.e., the order in which the vehicle to be parked passes the target nodes); wherein each RS curve path segment is obtained by connecting the coordinates and angles of two adjacent target nodes, and each RS curve path segment is obtained and added to the final feasible path in sequence until all target nodes of the guidance path are processed, and finally a smooth feasible path that complies with the vehicle motion constraints is obtained.

[0085] Therefore, in response to the common problems existing in path planning in current parking environments, the present invention proposes a multi-stage parking path planning algorithm framework that combines lane-level skeletonization with an A* algorithm and RS curve kinematic path generation. By integrating improved skeleton extraction technology with the A* algorithm, it effectively solves the problems of high computational resource consumption, high time complexity, and poor path smoothness faced by existing technologies in automatic valet parking systems. This integration strategy first uses the skeletonization algorithm to incorporate lane markings and parking space areas into obstacle constraints, which not only simplifies the complex parking environment and reduces its complexity, but also preserves key topological features, reducing the search space and computational resource consumption. Secondly, the present invention can also optimize the parking entrance posture based on lane constraints and the bidirectional motion capability of the Reeds-Shepp curve, and use the A* algorithm to generate a global guidance path. This four-stage path planning method significantly reduces time complexity and improves the speed and accuracy of path planning. In addition, the present invention ensures path smoothness and kinematic feasibility through kinematic trajectory refinement, further improving the real-time performance of path planning.

[0086] Optionally, in one embodiment of the present application, the feasible path is tested and optimized to obtain the optimal parking path corresponding to the vehicle to be parked, including: based on a preset global collision strategy, performing collision test processing on the feasible nodes in the feasible path to obtain corresponding detection results, and judging whether the current feasible path can avoid parking space obstacles and parking line obstacles based on the detection results; if the current feasible path can avoid parking space obstacles and parking line obstacles, then the current feasible path is determined as the optimal parking path for the vehicle to be parked; if the current feasible path cannot avoid parking space obstacles and parking line obstacles, then the topology structure of the path segment where the collision occurs is optimized to obtain a corresponding reconstructed path topology, and the preset Hybrid A* algorithm is used to replan the reconstructed path topology to obtain a collision-free optimal parking path.

[0087] It should be noted that the embodiments of the present application can perform collision testing on feasible nodes in a feasible path according to a preset collision strategy. If the detection result obtained by the collision test indicates that the current feasible path can avoid all obstacles, the current feasible path can be determined as the target parking path for the vehicle; otherwise, the subsequent feasible paths including the feasible nodes where the collision occurred are replanned using the Hybrid A* algorithm until a collision-free target parking path is found. The obtained conventional parking space entry and exit effect based on lane constraints is as follows: Figure 9 As shown, the storage results are as follows Figure 9 As shown in (a), the outbound results are as follows Figure 9 As shown in (b); the effect of parking in and out of the “dead-end road” parking space based on lane constraints is as follows Figure 10As shown, the storage results are as follows Figure 10 As shown in (a), the outbound results are as follows Figure 10 As shown in (b);

[0088] Specifically, the embodiment of the present application performs collision testing and optimization on feasible nodes in a feasible path to obtain a target parking path for the vehicle as follows:

[0089] 1. Global collision detection:

[0090] The embodiment of the present application uses a two-circle collision model to traverse the kinematic path nodes, and constructs a front / rear detection circle (radius R = 0.5W) with the front / rear axle center of the vehicle as the reference. d ), and verify the feasibility of the path through the geometric constraint equation:

[0091] d(O i ,C j )≥R+δ(1)

[0092] Among them, O i represents the coordinates of the obstacle; C j is the coordinate of the center of the detection circle; δ is the safety margin, which can be set to 0.2m.

[0093] It should be noted that in the embodiment of the present application, obstacle detection can be performed on each feasible node using the two-circle detection method. The specific process is as follows:

[0094] (1) Define the front and rear centers of the vehicle at each feasible node, which correspond to the front and rear axle centers of the vehicle, respectively, to simulate the position and shape of the vehicle in space;

[0095] (2) Determine the positions of the front circle and the rear circle in the global coordinate system. If the position of the front circle or the rear circle overlaps with an obstacle, the vehicle is deemed to have collided. Otherwise, the vehicle is deemed to be in a safe state. When a collision is detected, the system identifies the state of the current path and optimizes the path segment where the collision occurred, ultimately generating an optimal parking path that avoids the obstacle.

[0096] 2. “Dead-end road” scene recognition:

[0097] When a collision risk is detected, the terminal path characteristics are analyzed to determine whether they satisfy the following equation:

[0098] |θ start -θ end |≥170°∩Δs>3m (2)

[0099] Among them, θ start is the starting position of the end path, θ end is the end point pose of the terminal path, and Δs is the length of the terminal path.

[0100] If the equation (2) is satisfied, it is determined to be a “dead-end road” parking space (e.g. Figure 2 The red circle indicates the triggering of special coping strategies, as follows:

[0101] First, the embodiment of the present application can identify whether the node where the collision occurred is located at the last section of the path, and determine whether the first body posture angle at the node is opposite to the second body posture angle at the end of the path (or differs by 170 degrees). If these conditions are met, the last section of the path will be deleted, and the previous node of the collision node will be redirected to expand the path between it and the parking end point to generate a first subsequent target path. Therefore, the vehicle can smoothly reverse into the "dead-end" parking space through this new path, avoiding the risk of being unable to enter the parking space in a narrow scene; if the collision detection fails and the parking space is not a "dead-end road", the path between the collision node and the parking end point will be re-planned through the A* algorithm. In this process, the embodiment of the present application will still perform obstacle detection based on the two-circle detection method to generate a safe path away from obstacles. After expansion, the path forms the second subsequent target path, which is then merged with the path before the collision point to finally obtain a complete optimal parking path.

[0102] 3. Topology optimization:

[0103] For the "dead-end road" scenario, kinematic reverse deduction is used to reconstruct the path topology. That is, after deleting the conflicting path segments, new trajectory adjustment points are inserted to replan the terminal path segments.

[0104] 4. Adaptive replanning:

[0105] Establish a path backtracking mechanism to locate the first collision point P when conventional obstacles exist col , with P col The starting point of the path segment is used as the starting point and the Hybrid A* algorithm is used to generate a safe path.

[0106] Therefore, the embodiments of the present application can perform collision detection on the results of the path search to ensure that the planned path avoids all obstacles. If it is detected that the path collides with an obstacle, the path is replanned. In addition, the embodiments of the present application can also construct a parking scene topology map and a structured lane line constraint model based on the skeletonization algorithm, and based on the optimal parking point, improved skeletonization algorithm, A* algorithm, RS curve expansion, collision detection and replanning operations, a multi-stage parking path planning algorithm that takes into account lane line constraints is obtained, thereby efficiently planning a safe path to address the problem of failure in planning narrow lane-constrained "dead-end" parking spaces in AVP scenarios.

[0107] According to the lane line constraint-based skeleton-guided hierarchical parking path planning method proposed in an embodiment of the present application, a grid map corresponding to a real parking lot map with lane constraints is determined; an optimal parking entry point for the vehicle to be parked is determined based on parking information in the grid map and the minimum turning radius of the vehicle to be parked; a feasible path for the vehicle to be parked is determined based on the optimal parking entry point, a skeleton map corresponding to the lane line constraint parking lot grid map, and the minimum turning radius of the vehicle; and the feasible path is tested and optimized to obtain the optimal parking path for the vehicle to be parked, thereby ensuring the timeliness, accuracy, and safety of parking path planning.

[0108] Next, a skeleton-guided hierarchical parking path planning device based on lane line constraints proposed in an embodiment of the present application will be described with reference to the accompanying drawings.

[0109] Figure 11 4 is a block diagram of a lane line constraint-based skeleton-guided hierarchical parking path planning device according to an embodiment of the present application.

[0110] like Figure 11 As shown, the skeleton-guided hierarchical parking path planning device 10 based on lane line constraints includes: a map construction module 100, a calculation module 200 and a path planning module 300.

[0111] Among them, the map construction module 100 is used to obtain lane line information and parking space information corresponding to the target real parking lot, and determine the corresponding lane constraints based on the lane line information and parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints.

[0112] The calculation module 200 is configured to obtain parking information and a minimum turning radius of the vehicle to be parked from the target grid map, and determine an optimal parking point for the vehicle to be parked based on the parking information and the minimum turning radius. The parking information includes the vehicle parking method, the parking starting point, the parking end point, parking space information, and the parking space type.

[0113] The path planning module 300 is used to skeletonize the target grid map based on a preset improved skeletonization algorithm and lane constraints to obtain a skeleton map corresponding to the target grid map. The module then determines a feasible parking path for the vehicle to be parked based on the optimal parking point, the skeleton map, and the minimum turning radius. The module then tests and optimizes the feasible path to obtain the optimal parking path for the vehicle to be parked.

[0114] Optionally, in one embodiment of the present application, the calculation module 100 includes: an acquisition unit, a setting unit and a calculation unit.

[0115] The acquiring unit is configured to acquire AVP scenario information corresponding to the lane constraint, and acquire the parking end point coordinates of each parking space in the target grid map according to the AVP scenario information.

[0116] The setting unit is used to use the coordinates of the parking end point as the initial position and the posture of the vehicle to be parked as the initial direction, so as to draw a straight line segment of a preset length along the initial direction starting from the initial position, and use the straight line segment as the initial path.

[0117] The computing unit is configured to calculate the center of a parking arc segment based on the minimum turning radius of the vehicle to be parked as a target radius and the parking side and end point of the vehicle parking method, thereby calculating the optimal parking point for the vehicle to be parked based on the center of the arc, the initial path, and a preset RS curve algorithm.

[0118] Optionally, in one embodiment of the present application, the path planning module 200 includes: an optimization unit, an extraction unit, a construction unit, and a generation unit.

[0119] Among them, the optimization unit is used to plan the parked vehicles to the nearest downstream lane based on the optimal parking point, and optimize the parking lot obstacle layer corresponding to the target grid map according to the improved skeletonization algorithm corresponding to the nearest downstream lane and the target grid map, so as to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles.

[0120] The extraction unit is used to add lane line obstacles to the target grid map and extract skeleton intersections based on parking space obstacles and lane line obstacles to construct a skeleton map. The skeleton map is used to obtain multiple lane-level guidance points, and a global guidance path is generated based on the multiple lane-level guidance points and a preset A* algorithm.

[0121] The construction unit is used to construct multiple minimum turning radius constraint circles with each key node in the global guidance path as the center, determine the intersection of each minimum turning radius constraint circle and the global guidance path, and determine the guidance path node through the intersection, so as to construct an optimized guidance path that meets the preset vehicle steering characteristics based on the guidance path nodes.

[0122] The generation unit is used to take each node in the optimized guidance path as a target point and expand the target point to generate a feasible path that meets the preset vehicle kinematic requirements based on the expanded target point and in combination with the RS curve algorithm.

[0123] Optionally, in one embodiment of the present application, the path planning module 300 further includes: a judgment unit, a determination unit, and a re-planning unit.

[0124] Among them, the judgment unit is used to perform collision test processing on feasible nodes in the feasible path based on a preset global collision strategy to obtain corresponding detection results, and judge whether the current feasible path can avoid parking space obstacles and parking line obstacles based on the detection results.

[0125] The determining unit is configured to determine the current feasible path as the optimal parking path for the vehicle to be parked if the current feasible path can avoid parking space obstacles and parking space line obstacles.

[0126] The replanning unit is used to optimize the topology of the path segment where the collision occurs to obtain the corresponding reconstructed path topology if the current feasible path cannot avoid parking space obstacles and parking line obstacles. The reconstructed path topology is then replanned using the preset Hybrid A* algorithm to obtain the optimal collision-free parking path.

[0127] It should be noted that the above explanation of the embodiment of the skeleton-guided hierarchical parking path planning method based on lane line constraints is also applicable to the skeleton-guided hierarchical parking path planning device based on lane line constraints in this embodiment, and will not be repeated here.

[0128] The lane-constrained skeleton-guided hierarchical parking path planning device proposed in an embodiment of the present application includes a map construction module 100 for obtaining lane information and parking space information corresponding to a target real parking lot, determining corresponding lane constraints based on the lane information and parking space information, and constructing a target grid map corresponding to the target real parking lot based on the lane constraints; a calculation module 200 for obtaining parking information and a minimum turning radius of a parked vehicle from the target grid map, and determining an optimal parking entry point for the parked vehicle based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking start point, a parking end point, parking space information, and a parking space type; and a path planning module 300 for skeletonizing the target grid map based on a preset improved skeletonization algorithm and lane constraints to obtain a skeleton map corresponding to the target grid map, determining a feasible path for the parked vehicle based on the optimal parking entry point, the skeleton map, and the minimum turning radius, and testing and optimizing the feasible path to obtain an optimal parking path for the parked vehicle, thereby ensuring the timeliness, accuracy, and safety of parking path planning.

[0129] Figure 12 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0130] A memory 1201 , a processor 1202 , and a computer program stored in the memory 1201 and executable on the processor 1202 .

[0131] When the processor 1202 executes the program, the lane line constraint-based skeleton-guided hierarchical parking path planning method provided in the above embodiment is implemented.

[0132] Furthermore, the electronic device further includes:

[0133] The communication interface 1203 is used for communication between the memory 1201 and the processor 1202 .

[0134] The memory 1201 is used to store computer programs that can be run on the processor 1202 .

[0135] The memory 1201 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0136] If the memory 1201, processor 1202, and communication interface 1203 are implemented independently, the communication interface 1203, memory 1201, and processor 1202 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 12 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0137] Optionally, in a specific implementation, if the memory 1201, the processor 1202 and the communication interface 1203 are integrated on a chip, the memory 1201, the processor 1202 and the communication interface 1203 can communicate with each other through an internal interface.

[0138] The processor 1202 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0139] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned skeleton-guided hierarchical parking path planning method based on lane line constraints.

[0140] An embodiment of the present application further provides a computer program product, including a computer program, which, when executed, is used to implement the above-mentioned skeleton-guided hierarchical parking path planning method based on lane line constraints.

[0141] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0142] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0143] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0144] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program can be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0145] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0146] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0147] In addition, the functional units in the various embodiments of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into a module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0148] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A skeleton-guided hierarchical parking path planning method based on lane line constraints, characterized by: The following steps are involved: Acquiring lane line information and parking space information corresponding to a target real parking lot, and determining corresponding lane constraints based on the lane line information and the parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints; Obtaining parking information and a minimum turning radius of the vehicle to be parked in the target grid map, and determining an optimal parking point for the vehicle to be parked based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking starting point, a parking end point, parking space information, and a parking space type; Based on a preset improved skeletonization algorithm and the lane constraints, the target grid map is skeletonized to obtain a skeleton map corresponding to the target grid map. A feasible parking path for the vehicle to be parked is determined based on the optimal parking entry point, the skeleton map, and the minimum turning radius. The feasible path is tested and optimized to obtain an optimal parking path for the vehicle to be parked.

2. The method according to claim 1, characterized in that The acquiring of parking information and a minimum turning radius of the vehicle to be parked in the target grid map, and determining an optimal parking point for the vehicle to be parked based on the parking information and the minimum turning radius, includes: Obtaining AVP scenario information corresponding to the lane constraint, and obtaining the parking end point coordinates of each parking space in the target grid map according to the AVP scenario information; Using the coordinates of the parking end point as an initial position and the posture of the vehicle to be parked as an initial direction, drawing a straight line segment of a preset length along the initial direction starting from the initial position, and using the straight line segment as an initial path; The minimum turning radius of the vehicle to be parked is used as a target radius. The center of a parking arc segment is calculated based on the target radius, the parking side of the vehicle parking method, and the parking end point. The optimal parking point of the vehicle to be parked is calculated based on the center, the initial path, and a preset RS curve algorithm.

3. The method according to claim 2, characterized in that The target grid map is skeletonized based on a preset improved skeletonization algorithm and the lane constraint to obtain a skeleton map corresponding to the target grid map, and a feasible parking path for the vehicle to be parked is determined based on the optimal parking point, the skeleton map, and the minimum turning radius, including: Based on the optimal parking point, the vehicle to be parked is planned to be placed in the nearest oncoming lane, and according to the improved skeletonization algorithm corresponding to the nearest oncoming lane and the target grid map, the parking lot obstacle layer corresponding to the target grid map is optimized to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles; adding lane line obstacles to the target grid map, extracting skeleton intersection points based on the parking space obstacles and the lane line obstacles to construct the skeleton map, obtaining a plurality of lane-level guidance points from the skeleton map, and generating a global guidance path based on the plurality of lane-level guidance points and a preset A* algorithm; Constructing a plurality of minimum turning radius constraint circles with each key node in the global guidance path as a center, determining the intersection of each minimum turning radius constraint circle with the global guidance path, and determining a guidance path node based on the intersection, so as to construct an optimized guidance path that meets preset vehicle steering characteristics based on the guidance path nodes; Each node in the optimized guidance path is used as a target point, and the target point is expanded to generate a feasible path that meets the preset vehicle kinematic requirements based on the expanded target point and in combination with the RS curve algorithm.

4. The method according to claim 3, characterized in that The testing and optimizing of the feasible paths to obtain an optimal parking path corresponding to the vehicle to be parked includes: Based on a preset global collision strategy, a collision test is performed on the feasible nodes in the feasible path to obtain corresponding detection results, and according to the detection results, it is determined whether the current feasible path can avoid the parking space obstacle and the parking line obstacle; If the current feasible path can avoid the parking space obstacle and the parking space line obstacle, determining the current feasible path as the optimal parking path for the vehicle to be parked; If the current feasible path cannot avoid the parking space obstacle and the parking line obstacle, the topology structure of the path segment where the collision occurred is optimized to obtain a corresponding reconstructed path topology, and the preset Hybrid A* algorithm is used to replan the reconstructed path topology to obtain a collision-free optimal parking path.

5. A skeleton-guided hierarchical parking path planning device based on lane line constraints, characterized in that: include: a map construction module, configured to obtain lane line information and parking space information corresponding to a target real parking lot, and determine corresponding lane constraints based on the lane line information and the parking space information, so as to construct a target grid map corresponding to the target real parking lot based on the lane constraints; a calculation module, configured to obtain parking information and a minimum turning radius of a vehicle to be parked in the target grid map, and determine an optimal parking point for the vehicle to be parked based on the parking information and the minimum turning radius, wherein the parking information includes a vehicle parking method, a parking starting point, a parking end point, parking space information, and a parking space type; a path planning module, configured to skeletonize the target grid map based on a preset improved skeletonization algorithm and the lane constraints to obtain a skeleton map corresponding to the target grid map, determine a feasible path for the vehicle to be parked based on the optimal parking point, the skeleton map, and the minimum turning radius, and test and optimize the feasible path to obtain an optimal parking path corresponding to the vehicle to be parked.

6. The device according to claim 5, characterized in that The calculation module includes: an acquiring unit, configured to acquire AVP scenario information corresponding to the lane constraint, and acquire the parking end point coordinates of each parking space in the target grid map according to the AVP scenario information; a setting unit, configured to use the coordinates of the parking end point as an initial position, and the posture of the vehicle to be parked as an initial direction, to draw a straight line segment of a preset length starting from the initial position and along the initial direction, and to use the straight line segment as an initial path; a computing unit configured to calculate a center point of a parking arc segment based on a minimum turning radius of the vehicle to be parked and the target radius, the parking side of the vehicle parking method, and the parking endpoint, and to calculate an optimal parking point for the vehicle to be parked based on the center point, the initial path, and a preset RS curve algorithm.

7. The device according to claim 6, characterized in that The path planning module includes: an optimization unit, configured to plan the parked vehicle to a nearest oncoming lane based on the optimal parking point, and optimize the parking lot obstacle layer corresponding to the target grid map according to the nearest oncoming lane and an improved skeletonization algorithm corresponding to the target grid map, so as to perform obstacle processing on all parking spaces in the target grid map to generate corresponding parking space obstacles; an extraction unit, configured to add lane line obstacles to the target grid map, extract skeleton intersections based on the parking space obstacles and the lane line obstacles to construct the skeleton map, obtain a plurality of lane-level guidance points from the skeleton map, and generate a global guidance path based on the plurality of lane-level guidance points and a preset A* algorithm; a construction unit, configured to construct a plurality of minimum turning radius constraint circles with each key node in the global guidance path as a center, determine an intersection point between each minimum turning radius constraint circle and the global guidance path, and determine a guidance path node based on the intersection point, so as to construct an optimized guidance path that satisfies a preset vehicle steering characteristic based on the guidance path node; A generation unit is used to use each node in the optimized guidance path as a target point and expand the target point to generate a feasible path that meets the preset vehicle kinematic requirements based on the expanded target point and in combination with the RS curve algorithm.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the skeleton-guided hierarchical parking path planning method based on lane line constraints as described in any one of claims 1 to 4.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the skeleton-guided hierarchical parking path planning method based on lane line constraints as described in any one of claims 1 to 4.

10. A computer program product comprising a computer program, characterized in that The computer program is executed to implement the skeleton-guided hierarchical parking path planning method based on lane line constraints as described in any one of claims 1 to 4.

Images