Path planning method, system and equipment based on mixed A satellites and medium

By optimizing the hybrid A* algorithm using variable resolution raster maps and RS curve pre-computation, the problems of high computational cost and insufficient dynamic adaptability of the hybrid A* algorithm in complex scenarios are solved, and efficient and real-time path planning is achieved.

CN121761903APending Publication Date: 2026-03-31SINO TRUK JINAN POWER CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202512014290.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-29
Publication Date
2026-03-31

AI Technical Summary

Technical Problem

Existing hybrid A-Star path planning algorithms suffer from high computational cost and poor real-time performance in complex scenarios, and are not adaptable enough to dynamic environments, resulting in wasted computing resources and response delays.

Method used

By employing variable resolution raster map technology, the raster resolution is dynamically adjusted according to the distribution of obstacles. Combined with RS curve pre-calculation and parallel processing, invalid calculations are reduced and the path planning process is optimized.

Benefits of technology

It significantly improves the efficiency and real-time performance of path planning, making it suitable for high real-time scenarios such as autonomous driving mines and industrial parks, reducing computational load and enhancing responsiveness to dynamic environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121761903A_ABST
    Figure CN121761903A_ABST
Patent Text Reader

Abstract

The invention provides a path planning method, system and equipment based on a mixed satellite A and a medium, and belongs to the technical field of automatic driving path planning. The method comprises the following steps: constructing a variable-resolution grid map according to environment obstacle distribution, adopting a high resolution in an obstacle area, and adopting a low resolution in an open area; performing pose validity verification on the starting point and the target point; dynamically updating map obstacle information based on real-time sensor data; pre-calculating the RS curve connectivity from all discretized vehicle kinematics reachable points to a target point by adopting a parallel calculation mode, and establishing a directly connectable node set; in the process of executing the mixed A star algorithm, when the expansion nodes are located in the directly connectable node set, the pre-calculated RS curve is directly called to complete path planning, and search is terminated. According to the method, through the variable-resolution map and the pre-calculation mechanism, the calculation complexity is remarkably reduced, and the real-time performance and efficiency of path planning are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving path planning technology, and more specifically relates to a path planning method, system, device and medium based on hybrid A-star. Background Technology

[0002] Path planning is a core module of autonomous driving systems, and its performance directly affects the intelligence, safety, and real-time performance of vehicle driving. Among numerous global path planning algorithms, the Hybrid A* algorithm is widely used because it effectively combines the graph search completeness of the A* algorithm with the property of Reeds-Shepp (RS) curves adapting to vehicle kinematic constraints, generating a smooth path that is both geometrically and kinematically feasible. This algorithm discretizes the vehicle's continuous state space, thus addressing the problem of traditional A* algorithms neglecting vehicle motion characteristics and resulting in unexecutable planned paths while maintaining computational complexity.

[0003] However, existing hybrid AlphaGo algorithms still have significant limitations in practical applications. First, they typically rely on fixed-resolution raster maps for obstacle collision detection. To ensure the safety of the planned path in complex scenes, especially the accuracy when encountering nearby obstacles, high-resolution maps must be used, but this directly leads to a sharp increase in computational load. In open or semi-open areas, a large number of raster cells without obstacles are also included in the search and evaluation process, resulting in huge waste of computational resources and severely limiting the performance of the algorithm in real-time systems.

[0004] Secondly, during execution, to accelerate convergence and optimize the path, the algorithm needs to repeatedly verify whether a feasible RS curve exists between the current extended node and the final target point. Since calculating the RS curve involves complex geometric and kinematic solutions, this frequent and repetitive process incurs a heavy computational burden, becoming another key bottleneck restricting the overall efficiency of the algorithm. Furthermore, the traditional hybrid AlphaGo algorithm is insufficiently adaptable to dynamically changing driving environments. When obstacle information in the map is updated, the algorithm often needs to perform global map reconstruction and path replanning, and its long response latency makes it difficult to meet the stringent requirements for real-time obstacle avoidance in highly dynamic scenarios.

[0005] In summary, the inherent shortcomings of existing hybrid AlphaGo algorithms in terms of computational efficiency, redundancy elimination, and dynamic adaptability limit their further application in higher-order and more complex autonomous driving scenarios. Summary of the Invention

[0006] To address the above problems, the present invention aims to provide a path planning method, system, device, and medium based on hybrid AlphaGo. By reducing invalid calculations through variable resolution maps, pre-calculating nodes and building an RS curve library to avoid repeated verification, and optimizing the algorithm speed through hybrid AlphaGo execution, the invention solves the core problem of existing hybrid AlphaGo algorithms—the execution time problem—and is suitable for scenarios with high real-time requirements, such as autonomous driving mines and industrial parks.

[0007] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, embodiments of this application provide a path planning method based on a hybrid A-star system, including: Based on environmental data from actual operating scenarios, and according to the distribution of obstacles in the environment, the raster resolution of the electronic map is dynamically adjusted to construct a variable resolution raster map. Determine the starting and ending poses of the vehicle in a variable resolution raster map, and verify the validity of the starting and ending poses. Based on real-time environmental data collected by sensors, the locations of obstacles in the variable resolution raster map are dynamically updated and marked. Traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, store the RS curves as a mapping table including start and end state pairs and curve parameters, and the corresponding original point sequence; perform collision detection through the variable resolution raster map, identify nodes that can be directly connected to the endpoint, and store them in a set of directly connectable nodes. By executing the hybrid A* algorithm for path planning from the starting point to the ending point, during the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the ending pose, terminating subsequent node expansion.

[0008] In one optional implementation, the nodes in the variable resolution raster map include the vehicle's position (x, y) and heading angle θ; the nodes are discretized with a fixed step size to determine the vehicle's travel distance; wherein the fixed angle step size is 15°; the RS curves for the start and end points of all discrete states are pre-calculated, and the storage of each RS curve includes a mapping table of start and end state pairs and curve parameters, as well as the corresponding original point sequence.

[0009] In an optional implementation, the step of dynamically adjusting the raster resolution of the electronic map based on environmental data from the actual operating scenario and the distribution of obstacles in the environment to construct a variable-resolution raster map includes: LiDAR is used to collect point cloud data of the actual driving area of ​​the vehicle. Clustering algorithm is used to identify obstacles in the actual driving area of ​​the vehicle, mark the obstacles, and determine the distribution information of the obstacles. Based on the electronic map of the actual driving area of ​​the vehicle, and according to the distribution information of obstacles, the area without obstacles in the electronic map is divided into multiple unit areas, and the straight-line distance between each unit area and the obstacle is calculated. In all unit areas, determine whether the straight-line distance between each unit area and any obstacle is less than a preset distance threshold; If yes, the unit area is considered the area surrounding the obstacle; otherwise, the unit area is considered an open area. For the area surrounding the obstacle, set its grid size to A; for the open area, set its grid size to B; A < B; After setting the raster size for all unit areas, a variable resolution raster map is generated, and the map information is recorded.

[0010] In an optional implementation, determining the vehicle's starting and ending poses in the variable-resolution raster map, and validating the starting and ending poses, includes: The vehicle's starting point pose information and ending point pose information are identified in the variable resolution raster map, and the pose information includes coordinates and heading angle; The starting point pose and the ending point pose are respectively verified for coordinate validity, obstacle verification, and kinematic constraint verification. If all validations pass, the validity of the starting point pose and the ending point pose is validated; otherwise, path planning fails.

[0011] In an optional implementation, the coordinate validity verification includes: Determine whether the coordinates of the starting point and the ending point are both within the boundary of the variable resolution raster map; if so, the coordinate validity verification passes; otherwise, the coordinate validity verification fails. The obstacle verification includes: Determine whether the grid cells containing the starting and ending points are marked as obstacles; if so, obstacle verification fails; otherwise, obstacle verification succeeds. The kinematic constraint verification includes: Determine whether the heading angles of the starting point and the ending point both meet the minimum turning radius requirement of the vehicle; if so, the kinematic constraint verification passes; otherwise, the kinematic constraint verification fails.

[0012] In an optional implementation, the step of dynamically updating and marking the obstacle positions in the variable-resolution raster map based on real-time environmental data collected by sensors includes: The system collects point cloud data of the actual driving area of ​​the vehicle in real time using LiDAR and updates the variable resolution raster map according to the preset update frequency. The system determines in real time whether the obstacle has moved based on the point cloud data. If no obstacles are moving, set the update frequency of the variable resolution raster map to 1Hz; If any obstacle is detected to have moved, the obstacle position information in the variable resolution raster map is dynamically updated, the moving obstacle is marked, and the update frequency of the variable resolution raster map is set to 10Hz.

[0013] In an optional implementation, the process involves traversing all discretized vehicle kinematic reachable points in the variable-resolution raster map, calculating the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, storing the RS curves as a mapping table including start-end state pairs and curve parameters, and the corresponding original point sequence; performing collision detection through the variable-resolution raster map to identify nodes directly connectable to the endpoint, and storing these nodes in a set of directly connectable nodes, including: Using the CUDA parallel framework, all discretized vehicle kinematic reachable points in the variable resolution raster map are traversed. The Reeds-Shepp curve model is used to calculate the shortest path between nodes and the endpoint, generating RS curves. The storage of RS curves includes a mapping table of initial and final state pairs and curve parameters, as well as the corresponding original point sequence. For each RS curve, determine whether the RS curve is completely located in an open area; if so, the RS curve is a connected curve. If not, the area around the obstacle that the RS curve passes through is identified, and the obstacle's location information is used to determine whether the RS curve collides with the obstacle; if yes, the RS curve is an infeasible path; if no, the RS curve is a connectable curve. For each connectable curve, based on the corresponding mapping table and the corresponding original point sequence, extract the node information of the start and end points of the connectable curve, record them as directly connectable nodes, and store them in the set of directly connectable nodes.

[0014] In one optional implementation, nodes in the variable-resolution raster map include the vehicle's position (x, y) and heading angle θ; the nodes are discretized with a fixed angular step size of θ = 15°; the vehicle's step size travel distance is determined. RS curves for the start and target endpoints of all discrete states are pre-calculated, and the storage of each RS curve includes a mapping table of start and end state pairs and curve parameters, as well as the corresponding original point sequence. In open areas of the variable-resolution map, one vehicle's step size travel distance 'a' is used, and in obstacle areas, another vehicle's step size travel distance 'b' is used, where a > b.

[0015] In an optional implementation, the step of directly calling the RS curve as a traversable path from the next pose point to the final pose point when the generated next pose point belongs to the directly connectable node set, and terminating subsequent node expansion, includes: The generated next pose point is matched with the directly connectable nodes in the node set. The straight-line distance between the next pose point and each directly connectable node is calculated, and it is determined whether the straight-line distance is less than 0.25 meters. If so, the match is successful, and the connectable curve where the directly connectable node is located is extracted. When a match is successful, the path segment generated by the currently executed hybrid A* algorithm is spliced ​​with the extracted connectable curve to form a complete drivable path.

[0016] Secondly, embodiments of this application also provide a path planning system based on a hybrid A-star, comprising: The map construction module is used to dynamically adjust the raster resolution of the electronic map based on environmental data from the actual operating scenario and the distribution of obstacles in the environment, thus constructing a variable resolution raster map. The pose processing module is used to determine the starting point pose and ending point pose of the vehicle in a variable resolution raster map, and to verify the validity of the starting point pose and ending point pose. The obstacle dynamic marking module is used to dynamically update and mark the positions of obstacles in the variable resolution raster map based on real-time environmental data collected by sensors. The pre-computation module is used to traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, and store the RS curves as a mapping table including start-end state pairs and curve parameters, as well as the corresponding original point sequence; and perform collision detection through the variable resolution raster map to identify nodes that can be directly connected to the endpoint and store them in a set of directly connectable nodes. The path planning module is used to plan the path from the starting point to the destination by executing the hybrid A* algorithm. During the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the destination pose, and the subsequent node expansion is terminated.

[0017] Thirdly, embodiments of this application also provide an electronic device, including 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 steps of the path planning method based on Hybrid A-Star as described in any of the above.

[0018] Fourthly, embodiments of this application also provide a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the path planning method based on Hybrid A-Star as described in any of the above claims.

[0019] As can be seen from the above technical solutions, the present invention has the following advantages: To address the issues of high computational cost and poor real-time performance in complex scenarios caused by fixed-resolution maps in existing hybrid A-star algorithms, this application proposes a technical solution combining variable-resolution raster map construction, pre-calculated RS curves for connectable nodes, and hybrid A-star execution optimization. By reducing map resolution in open areas and increasing resolution around obstacles, unnecessary computations are reduced. The connectivity of RS curves between map nodes and target points is verified and stored in advance to avoid redundant calculations during hybrid A-star execution. When the next pose point generated by hybrid A-star coincides with a pre-stored node, the pre-calculated result is directly invoked to plan the path. This invention effectively improves the execution speed of the hybrid A-star algorithm and significantly enhances path planning efficiency, making it suitable for scenarios with high real-time requirements, such as autonomous driving mining trucks and industrial parks.

[0020] This application introduces a variable-resolution raster map, employing high resolution near obstacles to ensure safety and low resolution in open areas to significantly reduce the number of search nodes, thereby drastically reducing the computational load of the hybrid A* algorithm. Simultaneously, it innovatively pre-computes and stores a set of feasible RS curves for all nodes and endpoints that satisfy the discretization of vehicle kinematic constraints, transforming the repetitive curve calculations and collision detection during algorithm runtime into one-time preprocessing and efficient set queries, effectively avoiding the bottleneck of repetitive RS curve calculations. Furthermore, combined with a perception update mechanism based on obstacle dynamics, it achieves adaptive adjustment of the map update frequency, enhancing real-time response capabilities to dynamic environments. Ultimately, this method comprehensively improves the planning efficiency and real-time performance of the hybrid A* algorithm while ensuring path safety and kinematic feasibility, making it more suitable for complex and dynamic autonomous driving scenarios.

[0021] This application achieves precise allocation and efficient utilization of computing resources by constructing a variable-resolution raster map. The method dynamically adjusts based on environmental characteristics. In areas with dense or adjacent obstacles, high resolution is used to ensure the accuracy of path planning and the safety of collision detection; while in open areas, low resolution is decisively adopted, thereby significantly reducing the number of node expansions in these areas. This "site-specific" strategy fundamentally solves the computational redundancy problem caused by fixed resolution, significantly reducing the computational load of the algorithm, freeing up system resources, and laying a solid foundation for improving the real-time performance of path planning.

[0022] This application successfully overcomes the performance bottleneck of repetitive RS curve calculation in traditional hybrid AlphaGo systems by employing pre-computation and parallelization techniques. Utilizing parallel computing frameworks such as CUDA, it performs a one-time, batch traversal of RS curve connectivity between all discretized nodes satisfying vehicle kinematic constraints and the endpoint, and performs collision detection using a variable-resolution map, caching the results to a set of directly connectable nodes. This innovation transforms the previously repetitive and computationally expensive RS curve generation and verification work during the search process into a one-time preprocessing step. When the main algorithm runs, a simple set query is sufficient to determine whether the search can be terminated early and directly call the optimal path. This significantly reduces the real-time computational pressure on the algorithm during pathfinding, thereby improving search efficiency.

[0023] This application significantly enhances the responsiveness to dynamic scenes by introducing an adaptive update mechanism based on perception feedback. When the environment is static, a lower update frequency is used to save computational resources; once the sensor detects movement of an obstacle, the update frequency of the variable-resolution map is immediately increased, and the new location of the obstacle is quickly marked. This dynamic adjustment mechanism ensures the timeliness of environmental representation, enabling path replanning to be initiated rapidly based on the latest obstacle information, effectively shortening the delay from perception to decision-making. This allows for faster generation of safe new paths in the face of unexpected situations, improving the overall robustness of the system.

[0024] In the hybrid A* search process, once the next pose point of the extended algorithm is determined to belong to the set of directly connectable nodes, it indicates that there exists a known, kinematically feasible, collision-free RS path from this point to the endpoint. The algorithm can then immediately terminate the subsequent tedious node expansion and directly concatenate the currently searched path segment with this pre-stored, usually shortest or optimal RS curve, to form a complete path. This mechanism not only avoids unnecessary searches near the endpoint of the path, directly accelerating the convergence speed of the algorithm, but more importantly, it ensures that the last segment of the planned path strictly follows the RS curve model, thus mathematically achieving optimal path length or smoothness, ultimately outputting a high-quality path that combines efficient search and kinematic optimization. Attached Figure Description

[0025] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating the path planning method based on Hybrid A-Star provided in this application.

[0027] Figure 2 This is a schematic diagram illustrating the operation flow of the hybrid A-Star algorithm provided in this application.

[0028] Figure 3 A schematic diagram of the path planning system based on Hybrid A-Star provided in this application.

[0029] Figure 4 A schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0030] The various embodiments of this disclosure will be described more fully in the detailed steps of the hybrid A-star-based path planning method described below. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0031] In the following, the terms “comprising” or “may include”, which may be used in various embodiments of this disclosure, indicate the presence of the disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. Furthermore, as used in various embodiments of this disclosure, the terms “comprising,” “having,” and their cognates are intended only to indicate a particular feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more combinations of the foregoing.

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Please see Figure 1 The diagram shows a flowchart of a path planning method based on a hybrid A-satellite in a specific embodiment. The method includes: S1: Based on environmental data from actual operating scenarios, dynamically adjust the raster resolution of the electronic map according to the distribution of obstacles in the environment to construct a variable resolution raster map.

[0034] In a specific implementation, based on environmental data from the actual operating scenario, low-resolution grids are used to divide open areas and areas far from obstacles, while high-resolution grids are used to divide the areas around obstacles.

[0035] First, point cloud data of the actual driving area of ​​the vehicle is collected using LiDAR. Then, obstacles in the actual driving area are identified by clustering algorithms, and the obstacles are marked and their distribution information is determined.

[0036] Then, based on the electronic map of the actual driving area of ​​the vehicle, according to the distribution information of obstacles, the area without obstacles in the electronic map is divided into multiple unit areas, and the straight-line distance between each unit area and the obstacle is calculated; in all unit areas, it is determined whether the straight-line distance between each unit area and any obstacle is less than a preset distance threshold; if so, the unit area is regarded as the area around the obstacle; if not, the unit area is regarded as an open area. For the area surrounding obstacles, the grid size is set to 1 / 4 to 1 / 6 of the vehicle width; for open areas, the grid size is set to 1 / 2 to 1 / 4 of the vehicle width. That is, the area surrounding obstacles is a high-resolution area, and the open area is a low-resolution area. The boundary of the high-resolution area must extend beyond the obstacle boundary by at least one low-resolution grid cell.

[0037] Once the raster size for all unit areas is set, a variable resolution raster map is generated, and the map information is recorded.

[0038] Taking urban road scenarios as an example, point cloud data collected by LiDAR is used to identify obstacles (such as pedestrians, vehicles, and curbs) through clustering algorithms. A high-resolution 0.5m x 0.5m grid is used for the area within 5 meters of an obstacle; a low-resolution 2m x 2m grid is used for areas far from obstacles (such as the center of the road). Map data is stored as a two-dimensional array, where each element records information such as the grid's resolution and whether it represents an obstacle.

[0039] S2: Determine the starting point pose and ending point pose of the vehicle in the variable resolution raster map, and verify the validity of the starting point pose and ending point pose.

[0040] In a specific implementation, the starting point (x1, y1, ...) of the vehicle is determined in the variable resolution raster map. Pose and endpoint (x2, y2, The pose is used to verify whether the starting point pose and the planned ending point pose are located in a non-obstacle area and meet the vehicle kinematic constraints.

[0041] Specifically, the starting point pose and the ending point pose are verified for coordinate validity, obstacle verification, and kinematic constraint verification, respectively. If all verifications pass, the validity verification of the starting point pose and the ending point pose is successful; otherwise, the path planning fails.

[0042] The validity verification includes: Coordinate validity: Whether the starting and ending point coordinates are within the map boundaries; Obstacle verification: Whether the grid cells containing the starting point and the ending point are marked as obstacles; Kinematic constraints: Whether the heading angles of the starting and ending points meet the minimum turning radius requirements of the vehicle.

[0043] S3: Based on real-time environmental data collected by sensors, dynamically update and mark the locations of obstacles in the variable resolution raster map.

[0044] In a specific implementation, point cloud data of the actual driving area of ​​the vehicle is collected in real time by LiDAR, and the variable resolution raster map is updated according to a preset update frequency.

[0045] The update frequency of the variable resolution raster map is linked to the dynamic change frequency of obstacles: when the obstacles are stationary, the update frequency is 1Hz; when the obstacles are moving, the update frequency is 10Hz.

[0046] For example, the system can determine in real time whether an obstacle has moved based on the point cloud data. If no obstacles are moving, set the update frequency of the variable resolution raster map to 1Hz; If any obstacle is detected to have moved, the obstacle position information in the variable resolution raster map is dynamically updated, the moving obstacle is marked, and the update frequency of the variable resolution raster map is set to 10Hz.

[0047] S4: Traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, store the RS curves as a mapping table including initial and final state pairs and curve parameters, and the corresponding original point sequence; perform collision detection through the variable resolution raster map, identify nodes that can be directly connected to the endpoint, and store them in the set of directly connectable nodes.

[0048] The endpoint is a specific point located within the endpoint region.

[0049] In a specific implementation, the CUDA parallel framework is first used to traverse all discretized vehicle kinematic reachable points in the variable resolution raster map. The Reeds-Shepp curve model is used to calculate the shortest path between the node and the destination, generating the RS curve. The RS curve is stored in a mapping table that includes state pairs and curve parameters, corresponding to the original point sequence. The shortest path satisfies the vehicle's kinematic constraints, such as minimum turning radius and maximum steering angle.

[0050] Then, collision detection is performed, which adopts a layered detection strategy of variable resolution raster map: first, obviously infeasible paths are quickly eliminated through low resolution raster, and then path details are verified through high resolution raster.

[0051] Specifically, for each RS curve, determine whether the RS curve is completely located in an open area; if so, the RS curve is a connected curve.

[0052] If not, the area around the obstacle that the RS curve passes through is identified, and the location information of the obstacle is used to determine whether the RS curve collides with the obstacle; if yes, the RS curve is an infeasible path; if no, the RS curve is a connectable curve.

[0053] Finally, for each connectable curve, based on the corresponding mapping table and the corresponding original point sequence, the node information of the start and end points of the connectable curve is extracted, recorded as directly connectable nodes, and stored in the set of directly connectable nodes.

[0054] It should be noted that the nodes in the variable resolution raster map include the vehicle's position (x, y) and heading angle θ. These nodes are discretized with a fixed step size to determine the vehicle's travel distance. The fixed angle step size is 15°. The RS curves for the start and end points of all discrete states are pre-calculated. The storage of each RS curve includes a mapping table of start and end state pairs and curve parameters, as well as the corresponding original point sequence. A single vehicle travel distance is used in open areas of the variable resolution map, while a shorter vehicle travel distance is used in obstacle areas.

[0055] S5: Path planning from the starting point to the ending point is performed by executing the hybrid A* algorithm. During the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the ending pose, and subsequent node expansion is terminated.

[0056] In a specific implementation, this step includes two processes: node matching and path concatenation.

[0057] Node matching: The next pose point (including coordinates and heading angle) generated by the hybrid A-star algorithm is matched with the directly connectable nodes in the node set. The straight-line distance between the next pose point and each directly connectable node is calculated, and it is determined whether the straight-line distance is less than 0.25 meters. If so, the match is successful, and the connectable curve where the directly connectable node is located is extracted.

[0058] Path splicing: When a match is successful, the path segment generated by the current execution of the hybrid A* algorithm is spliced ​​with the extracted connectable curve to form a complete drivable path.

[0059] Taking an autonomous vehicle traveling from start point A to destination B as an example, the Hybrid A-Star algorithm initially expands nodes starting from start point A. When expanding to node C, by matching coordinates and heading angles (matching threshold of 0.25 meters), it is found that node C belongs to the pre-stored set of directly connectable nodes. At this point, the pre-calculated RS curve (node ​​C to destination B) is directly called, and the path generated by Hybrid A-Star (A→C) is spliced ​​to form the complete path (A→C→B), terminating subsequent node expansion. Experiments show that this optimization reduces the number of nodes expanded by Hybrid A-Star by 60%, and shortens the path planning time from less than 0.15 seconds to less than 0.1 seconds.

[0060] In this embodiment, by introducing a variable-resolution raster map, precise allocation of computing resources is achieved, significantly reducing the computational load of the algorithm in open areas. Utilizing a parallel pre-computation mechanism, all discretized feasible Reeds-Shepp curves from vehicle kinematic reachable points to the destination are generated and cached in one go, completely avoiding the overhead of repetitive calculations in traditional methods. Combined with an adaptive update strategy based on environmental dynamics, the system's real-time response capability to dynamic obstacles is effectively improved. Finally, the search process is prematurely terminated through "direct connection of node sets," which not only accelerates algorithm convergence but also ensures the kinematic optimality of the final path. These innovative synergistic effects comprehensively improve the planning efficiency, real-time performance, and dynamic environment adaptability of the hybrid A* algorithm while ensuring path safety and feasibility.

[0061] In one embodiment of the present invention, based on step S5, a possible embodiment will be given below, and its specific implementation will be described in a non-limiting manner.

[0062] refer to Figure 2 As shown in the figure, this embodiment discloses the operation flow of a hybrid A* algorithm, which specifically includes the following steps: 1. Initialize the hybrid A-star algorithm, set the starting point, target point, and variable resolution map of the algorithm, and generate the next pose point by executing the algorithm.

[0063] 2. Determine whether the next pose point is in the pre-stored node library (a set of nodes that can be directly connected).

[0064] If so, the pre-stored RS curve is called to connect to the target point, that is, the path segment generated by the currently executing hybrid A* algorithm is spliced ​​with the extracted connectable curve to form a complete drivable path, and the algorithm ends.

[0065] If not, continue with the normal algorithm operations, including calculating the cost of nodes (such as heuristic cost, movement cost), updating the Openlist priority queue, and proceeding to the next step.

[0066] 3. Determine whether the next pose point generated has reached the target point or whether there is no path available.

[0067] If so, output the current planned path directly.

[0068] If not, update the next pose point by executing the algorithm and return to step 2.

[0069] like Figure 3 As shown, the following are embodiments of the path planning system based on Hybrid A-satellite provided in this disclosure. This system and the path planning method based on Hybrid A-satellite in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the path planning system based on Hybrid A-satellite, please refer to the embodiments of the path planning method based on Hybrid A-satellite.

[0070] A path planning system based on Hybrid A-Star includes: The map construction module is used to dynamically adjust the raster resolution of the electronic map based on environmental data from the actual operating scenario and the distribution of obstacles in the environment, thereby constructing a variable resolution raster map.

[0071] The pose processing module is used to determine the starting point pose and ending point pose of the vehicle in a variable resolution raster map, and to verify the validity of the starting point pose and ending point pose.

[0072] The obstacle dynamic marking module is used to dynamically update and mark the positions of obstacles in the variable resolution raster map based on real-time environmental data collected by sensors.

[0073] The pre-computation module is used to traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, and store the RS curves as a mapping table including start-end state pairs and curve parameters, as well as the corresponding original point sequence; and perform collision detection through the variable resolution raster map to identify nodes that can be directly connected to the endpoint and store them in the set of directly connectable nodes.

[0074] The path planning module is used to plan the path from the starting point to the destination by executing the hybrid A* algorithm. During the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the destination pose, and the subsequent node expansion is terminated.

[0075] The path planning system based on Hybrid A-Star provided in this embodiment achieves intelligent allocation of computing resources by introducing a variable-resolution raster map. High resolution is used in areas near obstacles to ensure safety, while low resolution is used in open areas to significantly reduce computing nodes and improve search efficiency. At the same time, a parallel pre-computation method is adopted to complete the connectivity detection and caching of RS curves between all discretized vehicle kinematic reachable points and the destination in one go, transforming the repetitive calculations during algorithm runtime into efficient set queries, fundamentally solving the performance bottleneck of repeated RS curve calculations. Combined with an adaptive map update mechanism based on dynamic environmental changes, real-time collaboration between perception and planning is achieved, effectively improving the response speed in dynamic scenarios. Finally, by directly connecting the node set and using an early termination mechanism, the algorithm is ensured to converge quickly and generate a kinematically optimal complete path, thereby achieving a comprehensive improvement in four aspects: computing efficiency, real-time performance, dynamic adaptability, and path quality.

[0076] Figure 4 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present invention.

[0077] The path planning method based on Hybrid A-Star provided in this application can be applied to electronic devices. Those skilled in the art will understand that the electronic device structure involved in the embodiments of this invention does not constitute a limitation on the electronic device. An electronic device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the electronic device includes, but is not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0078] Electronic devices may include processors, external memory interfaces, internal memory, universal serial bus (USB) interfaces, charging management modules, power management modules, batteries, wireless communication modules, audio modules, speakers, microphones, sensor modules, buttons, cameras, displays, and SIM card interfaces, etc.

[0079] A processor may include one or more processing units, such as: a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural network processing unit (NPU). Different processing units may be independent devices or integrated into one or more processors.

[0080] The processor can serve as the nerve center and command center of an electronic device. The controller can generate operation control signals based on the instruction opcode and timing signals to control the fetching and execution of instructions.

[0081] The processor may also include memory for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory can store instructions or data that the processor has just used or that are used repeatedly. If the processor needs to use the instruction or data again, it can retrieve it directly from this memory. This avoids repeated accesses, reduces processor latency, and thus improves system efficiency.

[0082] An external storage interface (ESI) can be used to connect external memory cards, such as microSD cards, to expand the storage capacity of electronic devices. The external memory card communicates with the processor through the ESI to perform data storage functions, such as saving music and video files on the external memory card.

[0083] Internal memory can be used to store computer executable program code, which includes instructions. The processor executes various functional applications and data processing of electronic devices by running the instructions stored in internal memory. Internal memory can include a program storage area and a data storage area. Internal memory can include high-speed random access memory, and can also include non-volatile memory, such as at least one disk storage device, flash memory device, universal flash storage (UFS), etc.

[0084] Wireless communication functionality in electronic devices can be achieved through antennas, wireless communication modules, modem processors, and baseband processors.

[0085] Wireless communication modules can provide solutions for wireless communication applications in electronic devices, including wireless local area networks (WLANs) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), and infrared (IR) technologies.

[0086] Electronic devices can implement audio functions through audio modules, speakers, receivers, microphones, headphone jacks, and application processors.

[0087] Electronic devices can achieve shooting functions through ISPs, cameras, video codecs, GPUs, displays, and application processors.

[0088] Electronic devices can achieve display functions through GPUs, displays, and application processors.

[0089] A GPU is a microprocessor for image processing, connected to the display screen and application processor. GPUs are used to perform mathematical and geometric calculations for graphics rendering. A processor may include one or more GPUs, which execute program instructions to generate or modify display information.

[0090] A display screen is used to display images, videos, etc. A display screen includes a display panel.

[0091] The aforementioned electronic device implements the path planning method based on the hybrid A-star proposed in this application by integrating a series of collaborative technologies, including variable resolution map construction, RS curve pre-computation caching, dynamic environment adaptive updates, and direct connection node termination search. This achieves the beneficial effects of significantly improving algorithm computational efficiency, enhancing real-time response capabilities in dynamic scenes, and optimizing the final path quality while strictly ensuring path safety and kinematic feasibility.

[0092] The storage medium provided in this application stores a program product capable of implementing a path planning method based on Hybrid A-Star.

[0093] In some possible implementations, the hybrid A-star-based path planning method of this disclosure can be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0094] The storage medium disclosed herein may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0095] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A path planning method based on hybrid A-star, characterized in that, include: Based on environmental data from actual operating scenarios, and according to the distribution of obstacles in the environment, the raster resolution of the electronic map is dynamically adjusted to construct a variable resolution raster map. Determine the starting and ending poses of the vehicle in a variable resolution raster map, and verify the validity of the starting and ending poses. Based on real-time environmental data collected by sensors, the locations of obstacles in the variable resolution raster map are dynamically updated and marked. Traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, store the RS curves as a mapping table including start and end state pairs and curve parameters, and the corresponding original point sequence; perform collision detection through the variable resolution raster map, identify nodes that can be directly connected to the endpoint, and store them in a set of directly connectable nodes. By executing the hybrid A* algorithm for path planning from the starting point to the ending point, during the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the ending pose, terminating subsequent node expansion.

2. The path planning method based on hybrid A-satellite as described in claim 1, characterized in that, The environmental data based on the actual operating scenario, according to the distribution of obstacles in the environment, dynamically adjusts the raster resolution of the electronic map to construct a variable resolution raster map, including: LiDAR is used to collect point cloud data of the actual driving area of ​​the vehicle. Clustering algorithm is used to identify obstacles in the actual driving area of ​​the vehicle, mark the obstacles, and determine the distribution information of the obstacles. Based on the electronic map of the actual driving area of ​​the vehicle, and according to the distribution information of obstacles, the area without obstacles in the electronic map is divided into multiple unit areas, and the straight-line distance between each unit area and the obstacle is calculated. In all unit areas, determine whether the straight-line distance between each unit area and any obstacle is less than a preset distance threshold; If yes, the unit area is considered the area surrounding the obstacle; otherwise, the unit area is considered an open area. For the area surrounding the obstacle, set its grid size to A; for the open area, set its grid size to B; A < B; After setting the raster size for all unit areas, a variable resolution raster map is generated, and the map information is recorded.

3. The path planning method based on hybrid A-satellite as described in claim 2, characterized in that, The process of determining the vehicle's starting and ending poses in a variable-resolution raster map and validating the starting and ending poses includes: The vehicle's starting point pose information and ending point pose information are identified in the variable resolution raster map, and the pose information includes coordinates and heading angle; The starting point pose and the ending point pose are respectively verified for coordinate validity, obstacle verification, and kinematic constraint verification. If all validations pass, the validity of the starting point pose and the ending point pose is validated; otherwise, path planning fails.

4. The path planning method based on hybrid A-satellite as described in claim 3, characterized in that, The coordinate validity verification includes: Determine whether the coordinates of the starting point and the ending point are both within the boundary of the variable resolution raster map; if so, the coordinate validity verification passes; otherwise, the coordinate validity verification fails. The obstacle verification includes: Determine whether the grid cells containing the starting and ending points are marked as obstacles; if so, obstacle verification fails; otherwise, obstacle verification succeeds. The kinematic constraint verification includes: Determine whether the heading angles of the starting point and the ending point both meet the minimum turning radius requirement of the vehicle; if so, the kinematic constraint verification passes; otherwise, the kinematic constraint verification fails.

5. The path planning method based on hybrid A-satellite as described in claim 4, characterized in that, The method of dynamically updating and marking the obstacle positions in the variable resolution raster map based on real-time environmental data collected by sensors includes: The system collects point cloud data of the actual driving area of ​​the vehicle in real time using LiDAR and updates the variable resolution raster map according to the preset update frequency. The system determines in real time whether the obstacle has moved based on the point cloud data. If no obstacles are moving, set the update frequency of the variable resolution raster map to 1Hz; If any obstacle is detected to have moved, the obstacle position information in the variable resolution raster map is dynamically updated, the moving obstacle is marked, and the update frequency of the variable resolution raster map is set to 10Hz.

6. The path planning method based on hybrid A-satellite as described in claim 5, characterized in that, The process involves traversing all discretized vehicle kinematic reachable points in the variable resolution raster map, calculating the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, and storing the RS curves as a mapping table including start and end state pairs and curve parameters, along with the corresponding original point sequence. Collision detection is performed using the variable resolution raster map to identify nodes that can be directly connected to the endpoint, and these nodes are stored in a set of directly connectable nodes, including: Using the CUDA parallel framework, all discretized vehicle kinematic reachable points in the variable resolution raster map are traversed. The Reeds-Shepp curve model is used to calculate the shortest path between nodes and the endpoint, generating RS curves. The storage of RS curves includes a mapping table of initial and final state pairs and curve parameters, as well as the corresponding original point sequence. For each RS curve, determine whether the RS curve is completely located in an open area; if so, the RS curve is a connected curve. If not, the area around the obstacle that the RS curve passes through is identified, and the obstacle's location information is used to determine whether the RS curve collides with the obstacle; if yes, the RS curve is an infeasible path; if no, the RS curve is a connectable curve. For each connectable curve, based on the corresponding mapping table and the corresponding original point sequence, extract the node information of the start and end points of the connectable curve, record them as directly connectable nodes, and store them in the set of directly connectable nodes.

7. The path planning method based on hybrid A-satellite as described in claim 6, characterized in that, When the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly used as the traversable path from the next pose point to the final pose, terminating subsequent node expansion, including: The generated next pose point is matched with the directly connectable nodes in the node set. The straight-line distance between the next pose point and each directly connectable node is calculated, and it is determined whether the straight-line distance is less than 0.25 meters. If so, the match is successful, and the connectable curve where the directly connectable node is located is extracted. When a match is successful, the path segment generated by the currently executed hybrid A* algorithm is spliced ​​with the extracted connectable curve to form a complete drivable path.

8. A path planning system based on Hybrid A-Star, characterized in that, The system employs the path planning method based on hybrid A-star as described in any one of claims 1 to 7; The system includes: The map construction module is used to dynamically adjust the raster resolution of the electronic map based on environmental data from the actual operating scenario and the distribution of obstacles in the environment, thus constructing a variable resolution raster map. The pose processing module is used to determine the starting point pose and ending point pose of the vehicle in a variable resolution raster map, and to verify the validity of the starting point pose and ending point pose. The obstacle dynamic marking module is used to dynamically update and mark the positions of obstacles in the variable resolution raster map based on real-time environmental data collected by sensors. The pre-computation module is used to traverse all discretized vehicle kinematic reachable points in the variable resolution raster map, calculate the connectivity between each discretized vehicle kinematic reachable point and the endpoint pose using RS curves, and store the RS curves as a mapping table including start-end state pairs and curve parameters, as well as the corresponding original point sequence; and perform collision detection through the variable resolution raster map to identify nodes that can be directly connected to the endpoint and store them in a set of directly connectable nodes. The path planning module is used to plan the path from the starting point to the destination by executing the hybrid A* algorithm. During the algorithm execution, when the generated next pose point belongs to the set of directly connectable nodes, the RS curve is directly called as the drivable path from the next pose point to the destination pose, and the subsequent node expansion is terminated.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the path planning method based on Hybrid A-Star as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the path planning method based on Hybrid A-Star as described in any one of claims 1 to 7.