Vehicle path planning method and device, storage medium and vehicle system

Abstract

The invention provides a vehicle path planning method and device, a storage medium and a vehicle system, and the method comprises the steps: obtaining the point cloud data of a vehicle environment, and constructing a skeleton topological graph of the vehicle environment according to the point cloud data; obtaining the current position of the vehicle and a plurality of preset target positions, and determining a plurality of driving routes in the skeleton topological graph according to the current position and the plurality of preset target positions; determining the score of each driving route according to a preset function; and determining a target driving route according to the score, and controlling the vehicle to drive according to the target driving route. The problem that in the prior art, an automatic driving vehicle cannot flexibly select a passing path according to an environment structure in a complex scene is solved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and more specifically, to a vehicle path planning method, a vehicle path planning device, a computer-readable storage medium, and a vehicle system. Background Technology

[0002] Existing autonomous vehicles often rely on global grid search for path planning in multi-objective scenarios, which involves a large amount of computation and is difficult to combine with the actual traffic structure for planning. Summary of the Invention

[0003] The main objective of this application is to provide a vehicle path planning method, a vehicle path planning device, a computer-readable storage medium, and a vehicle system, so as to at least solve the problem in the prior art that autonomous vehicles cannot flexibly select travel paths according to the environmental structure in complex scenarios.

[0004] To achieve the above objectives, according to one aspect of this application, a vehicle path planning method is provided, comprising: acquiring point cloud data of a vehicle environment, and constructing a skeleton topology map of the vehicle environment based on the point cloud data, the skeleton topology map representing the spatial structure of the vehicle's passable areas; acquiring the vehicle's current position and multiple preset target positions, and determining multiple driving routes in the skeleton topology map based on the current position and the multiple preset target positions, wherein each driving route includes the current position and each preset target position; determining a score for each driving route based on a preset function, the preset function being a function at least related to the number of curves in the driving route; determining a target driving route based on the score, and controlling the vehicle's driving based on the target driving route.

[0005] Optionally, determining the score for each driving route according to a preset function includes: determining the number of curves in each driving route, and determining a complexity value for each driving route based on the number of curves, the complexity value representing the operational complexity of the vehicle driving the driving route; determining the route length of each driving route; determining the number of obstacles in each driving route, and determining an obstacle risk value based at least on the number of obstacles, the obstacle risk value representing the degree of impact of the obstacles on the vehicle; determining the length of the reversing route in each driving route, and determining a reversible maneuverability value based at least on the length of the reversing route, the reversible maneuverability... The value represents the proportion of the reversing route length to the driving route; the first product is determined to be the product of the first preset weight coefficient and the complexity value; the second product is determined to be the product of the second preset weight coefficient and the route length; the third product is determined to be the product of the third preset weight coefficient and the obstacle risk value; the fourth product is determined to be the product of the fourth preset weight coefficient and the reversible maneuver value; the first sum is determined to be the sum of the first product and the second product; the second sum is determined to be the sum of the first sum and the third product; the third sum is determined to be the sum of the second sum and the fourth product; the score of the driving route is determined to be the third sum.

[0006] Optionally, determining the target driving route based on the score includes: determining the driving route corresponding to the minimum score as the target driving route.

[0007] Optionally, acquiring point cloud data of the vehicle environment and constructing a skeleton topology map of the vehicle environment based on the point cloud data includes: acquiring preliminary point cloud data of the vehicle environment; preprocessing the preliminary point cloud data to obtain the point cloud data, wherein the preprocessing includes removing noise points from the preliminary point cloud data; determining passable areas in the point cloud data, and determining the skeleton topology map of the vehicle environment based on the passable areas.

[0008] Optionally, determining the passable areas in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable areas includes: determining multiple access layers of the vehicle environment based on the point cloud data and determining the passable areas of each access layer; extracting the centerline of each passable area and sampling the centerline according to a preset step size to obtain multiple sampled centerlines; determining the endpoints of each sampled centerline as skeleton nodes of the skeleton topology and determining the skeleton topology of the vehicle environment based on the skeleton nodes.

[0009] Optionally, the method further includes: if the gradient of the vehicle's current driving route is greater than a preset gradient, controlling the vehicle to perform an avoidance operation, the avoidance operation including a reversing operation.

[0010] Optionally, the method further includes: determining the occupancy rate of an obstacle, the occupancy rate representing the degree to which the obstacle occupies the passage area of ​​the vehicle; and controlling the vehicle to perform an avoidance operation when the occupancy rate is greater than a preset occupancy rate.

[0011] According to another aspect of this application, a vehicle path planning device is provided, comprising: a first acquisition unit, configured to acquire point cloud data of a vehicle environment and construct a skeleton topology map of the vehicle environment based on the point cloud data, the skeleton topology map representing the spatial structure of the vehicle's passable area; a second acquisition unit, configured to acquire the vehicle's current position and multiple preset target positions, and determine multiple driving routes in the skeleton topology map based on the current position and the multiple preset target positions, wherein each driving route includes the current position and each preset target position; a first determination unit, configured to determine a score for each driving route based on a preset function, the preset function being a function at least related to the number of curves in the driving route; and a second determination unit, configured to determine a target driving route based on the score and control the vehicle's driving based on the target driving route.

[0012] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform any of the methods described.

[0013] According to another aspect of this application, a vehicle system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any one of the methods described.

[0014] By applying the technical solution of this application, point cloud data of the vehicle environment is acquired, and a skeleton topology map representing the spatial structure of the vehicle's passable area is constructed. Based on this, and combining the vehicle's current position with multiple preset target positions, multiple candidate driving routes are determined based on the skeleton topology map. Preset functions related to traffic characteristics such as the number of curves are used to comprehensively score each driving route. By comparing the scores of different candidate routes, the optimal target driving route is determined, and the vehicle is then controlled to travel along the target driving route. This enables autonomous vehicles to flexibly select travel paths according to the spatial structure of the environment, achieving intelligent and adaptive path planning, and improving the efficiency and safety of vehicle traffic in complex or irregular environments. This solves the problem in existing technologies where autonomous vehicles cannot flexibly select travel paths according to the environmental structure in complex scenarios. Attached Figure Description

[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1 A flowchart illustrating a vehicle routing method according to an embodiment of this application is shown.

[0017] Figure 2 A structural block diagram of a vehicle routing device provided according to an embodiment of this application is shown. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] As described in the background section, in the prior art, autonomous vehicles cannot flexibly select travel paths according to the environmental structure in complex scenarios. To solve the above problems, embodiments of this application provide a vehicle path planning method, a vehicle path planning device, a computer-readable storage medium, and a vehicle system.

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0023] Figure 1 This is a flowchart of a vehicle routing method according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0024] Step S101: Obtain point cloud data of the vehicle environment, and construct a skeleton topology map of the vehicle environment based on the point cloud data. The skeleton topology map represents the spatial structure of the passable area of ​​the vehicle.

[0025] Step S102: Obtain the current position of the vehicle and multiple preset target positions, and determine multiple driving routes in the skeleton topology map based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position.

[0026] Specifically, each of the above-mentioned driving routes starts from the current location and passes through all the above-mentioned target locations in sequence, although the order of visits may differ.

[0027] Step S103: Determine the score of each of the above driving routes according to a preset function, wherein the preset function is a function that is at least related to the number of curves of the above driving routes.

[0028] Step S104: Determine the target driving route based on the above score, and control the vehicle to drive according to the above target driving route.

[0029] Through the above embodiments, point cloud data of the vehicle environment is acquired, and a skeleton topology map representing the spatial structure of the vehicle's passable area is constructed. Based on this, and combined with the vehicle's current position and multiple preset target positions, multiple candidate driving routes are determined based on the skeleton topology map. Preset functions related to traffic characteristics such as the number of curves are used to comprehensively score each driving route. By comparing the scores of different candidate routes, the optimal target driving route is determined, and the vehicle is controlled to travel along the target driving route. This enables autonomous vehicles to flexibly select travel paths according to the spatial structure of the environment, achieving intelligent and adaptive path planning, and improving the efficiency and safety of vehicle traffic in complex or irregular environments. This solves the problem in existing technologies where autonomous vehicles cannot flexibly select travel paths according to the environmental structure in complex scenarios.

[0030] In one optional scheme, determining the score for each of the aforementioned driving routes according to a preset function includes: determining the number of curves in each of the aforementioned driving routes, and determining a complexity value for each of the aforementioned driving routes based on the number of curves, wherein the complexity value characterizes the operational complexity of the vehicle driving the aforementioned driving route; determining the route length of each of the aforementioned driving routes; determining the number of obstacles in each of the aforementioned driving routes, and determining an obstacle risk value based at least on the number of obstacles, wherein the obstacle risk value characterizes the degree of impact of the obstacles on the vehicle; determining the length of the reversing route in each of the aforementioned driving routes, and determining a reversible maneuverability value based at least on the length of the reversing route, wherein the reversible... The maneuverability value represents the proportion of the reversing route length to the driving route; the first product is determined to be the product of the first preset weight coefficient and the complexity value; the second product is determined to be the product of the second preset weight coefficient and the route length; the third product is determined to be the product of the third preset weight coefficient and the obstacle risk value; the fourth product is determined to be the product of the fourth preset weight coefficient and the reversible maneuverability value; the first sum is determined to be the sum of the first product and the second product; the second sum is determined to be the sum of the first sum and the third product; the third sum is determined to be the sum of the second sum and the fourth product; the score of the driving route is determined to be the third sum.

[0031] In the above embodiments, by determining the score of each driving route, multiple candidate driving routes can be quantitatively compared and comprehensively evaluated. The function simultaneously considers factors such as route length, number of curves, obstacle distribution, and the proportion of reversing sections, comprehensively reflecting the driving cost, safety, and operational difficulty of each route. Thus, the vehicle no longer uses the shortest route as the sole criterion, but can select the overall optimal driving route under multi-dimensional indicators. This ensures vehicle driving efficiency while avoiding routes with sharp curves, dense obstacles, or excessive reversing ratios, thereby improving the rationality of path planning and traffic safety. Furthermore, by setting different weight coefficients, the planning emphasis can be flexibly adjusted according to actual scenarios (such as narrow passages, parking lots, and multi-target access tasks), achieving adaptive path planning strategies and improving the accuracy, stability, and safety of autonomous vehicles in complex environments.

[0032] Specifically, according to J=λ L ·L+λ M ·M+λ O ·O+λ R ·R, determine the score for each of the above driving routes, where J is the score and λ is the score. L The second preset weight is given above, where L is the route length, M is the route complexity value, and λ is the weight of the route. M The first preset weight is given above, O is the obstacle risk value of the above driving route, and λ is the weight given above. OThe third preset weight is given above, R is the reversible maneuver value of the above driving route, and λ is the weight given above. R This is the fourth preset weighting coefficient. Each of the aforementioned preset weights is pre-set based on the vehicle operation scenario, path planning objectives, and safety constraints.

[0033] λ L The route length weight is set based on the vehicle's required driving efficiency in the current task; when priority needs to be given to shortening driving time or energy consumption, λ L The value is relatively large; λ M λ is used as the weight for maneuver complexity, set according to the vehicle's handling performance, steering sensitivity, and driving stability requirements; when it is desired to reduce sharp turns or complex maneuvers, λ M The value is relatively large; λ O The obstacle risk weight is set based on the detection accuracy of the vehicle's perception system and safety strategy; when the environment is densely packed with obstacles or has high safety requirements, λ O The value is relatively large; λ R The reversible maneuver weight is set based on whether reversing or U-turns are permitted or required in the scenario; in scenarios where reversing is costly, such as narrow passages or parking lots, λ R The value is relatively large. Preferably, λ L , λ M , λ O , λ R All are real numbers greater than zero.

[0034] In determining the obstacle risk value for each of the aforementioned driving routes, environmental point cloud data is first acquired based on the vehicle's perception system (such as LiDAR, millimeter-wave radar, or cameras), and the number of obstacles is counted within a preset bandwidth range (e.g., 1.5 times the vehicle width) for each candidate driving route. The closest distance d between an obstacle and the centerline of the driving route is then used to determine the obstacle risk. i Less than the preset distance threshold D th When the distance is 2m, the obstacle is identified as a valid obstacle affecting passage, and its quantity is recorded as N. obs To more accurately reflect the impact of obstacle distance on passage risk, this embodiment applies distance-weighted processing to the number of obstacles, calculating the obstacle risk value O. The calculation formula is as follows: , where d iLet di be the closest distance between the i-th obstacle and the centerline of the driving route, and d0 be the distance attenuation coefficient (e.g., 2m), used to control the attenuation rate of the obstacle's distance influence. Therefore, when obstacles are close to the route or numerous, the obstacle risk value O increases, indicating a higher safety risk for the driving route; conversely, when obstacles are sparse or far apart, the value of O is smaller. In this embodiment, the method for determining the reversible maneuver value R is as follows: First, identify the reversing and forward segments in the driving route based on the path planning results. When the angle between the trajectory tangential direction and the vehicle's forward direction is greater than a preset threshold (e.g., 90°), or when the longitudinal speed is determined to be negative based on vehicle speed information, the corresponding trajectory segment is marked as a reversing segment; the lengths of all reversing segments are accumulated to obtain the reversing route length L. rev And calculate the total length L of the entire route. In the basic calculation, the reversible maneuver value R is taken as the proportion of the reverse route length to the total route length, i.e., R = L. rev / L.

[0035] To more accurately reflect the costs of different types of reversible operations, this embodiment further incorporates the number of turning operations and the reversing cost coefficient for correction. When a U-turn or multiple forward / reverse switching operations are detected in the route, the number of turning operations N is introduced. turn and reversing cost coefficient k r According to: R=L rev / L+k r ×N turn , where k r The value used to characterize the additional cost of reversing or making a U-turn is preferably between 1.2 and 1.5. Through the above calculations, the reversible maneuverability value R comprehensively reflects the degree and cost of a vehicle performing reversible operations along a driving route. When the proportion of reversing or U-turns on the route is high, the R value increases accordingly, indicating that the route has higher costs in terms of maneuverability and feasibility.

[0036] In another alternative approach, determining the target driving route based on the above scores includes: determining the driving route corresponding to the minimum score among the above scores as the target driving route.

[0037] In the above embodiments, the target driving route is determined based on the score, and the route with the lowest score is selected as the target driving route. This allows the vehicle to automatically choose the path with the lowest overall cost and optimal overall trafficability from multiple candidate routes. Since the score comprehensively considers factors such as route length, curve complexity, obstacle risk, and reversible maneuverability, selecting the route with the lowest score is equivalent to selecting the route with the lowest driving cost, highest safety, and simplest operation. This avoids the risks associated with using the shortest distance or shortest time as the sole criterion, making the path planning results more consistent with the vehicle's actual trafficability and driving stability, and improving the accuracy of path selection, driving safety, and overall traffic efficiency in complex environments.

[0038] In some exemplary embodiments, acquiring point cloud data of a vehicle environment and constructing a skeleton topology map of the vehicle environment based on the point cloud data includes: acquiring preliminary point cloud data of the vehicle environment; preprocessing the preliminary point cloud data to obtain the point cloud data, wherein the preprocessing includes removing noise points from the preliminary point cloud data; determining passable areas in the point cloud data, and determining the skeleton topology map of the vehicle environment based on the passable areas.

[0039] In the above embodiments, by acquiring and preprocessing preliminary point cloud data of the vehicle environment, noise points and abnormal data can be effectively removed, improving the accuracy and stability of the point cloud information. Furthermore, by identifying passable areas in the point cloud data and generating a skeleton topology map of the vehicle environment, the structural relationships of the vehicle's passable space can be intuitively described. This skeleton topology map significantly reduces data complexity while preserving the main geometric features of the environment, thus providing a clear, simplified, and efficient spatial representation for subsequent path planning and navigation decisions, improving the accuracy and computational efficiency of path search.

[0040] In other exemplary embodiments, determining the passable areas in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable areas includes: determining multiple access levels of the vehicle environment based on the point cloud data and determining the passable areas of each access level; extracting the center lines of each passable area and sampling the center lines according to a preset step size to obtain multiple sampled center lines; determining the endpoints of each sampled center line as skeleton nodes of the skeleton topology and determining the skeleton topology of the vehicle environment based on the skeleton nodes.

[0041] In the above embodiments, by determining multiple access levels based on point cloud data and identifying the passable areas of each level, driving spaces at different heights or levels can be spatially distinguished, and the access boundaries and connectivity of each level can be clearly defined. Based on this, the centerlines of each passable area are extracted and sampled according to a preset step size, allowing the skeleton elements to express the main travel path with a unified spatial granularity, facilitating control of node density and subsequent computational overhead. Furthermore, the endpoints of each sampled centerline are used as skeleton nodes, and a skeleton topology is determined accordingly. This transforms the vehicle's passable space from a continuous region into a structured representation composed of nodes and connectivity relationships, thereby significantly reducing the complexity of the environment representation while ensuring traffic continuity, providing a clear and efficient topological foundation for subsequent path search and route comparison.

[0042] For example, firstly, raw point cloud data and pose information of the environment are collected in real time using the vehicle-mounted LiDAR and inertial measurement unit to obtain preliminary point cloud data of the vehicle environment. This point cloud data covers the space in front of, to the sides of, and around the vehicle, describing the spatial distribution characteristics of the ground, obstacles, and elevation changes in the environment. Then, the preliminary point cloud data is processed using a Simultaneous Localization and Mapping (SLAM) algorithm to achieve point cloud registration and map construction. Specifically, the SLAM module performs time synchronization and spatial alignment on continuously acquired point cloud frames, estimates the vehicle pose through feature point matching and loop closure detection mechanisms, and fuses multiple point cloud frames to form a dense environmental point cloud map in a unified coordinate system. After obtaining the structured point cloud map, the point cloud data undergoes preprocessing, including: noise removal: statistical filtering and radius filtering methods are used to remove isolated points and anomalous reflection points; voxel grid algorithm is used to reduce the amount of data; ground segmentation: a plane fitting algorithm based on RANSAC (RANdom Sampling Consensus) is used to distinguish ground points from obstacle points; non-passable points with heights exceeding a preset threshold (e.g., 2m) are deleted. Next, the point cloud is layered according to the height value Z of each point to distinguish different passable levels. When the height difference between adjacent points is greater than a first preset threshold (e.g., 0.1m), they are determined to be on different levels. Within each passable level, the 3D point cloud is projected onto a 2D plane to generate an occupancy grid, identifying low-elevation-difference connected regions as passable areas. For the passable areas of each passable level, morphological thinning algorithms or medial axis transformation methods are used to extract the centerline of the passable area, obtaining the skeleton line representing the main passable path. The skeleton lines are then discretized with a fixed step size to obtain multiple skeleton nodes, and connections are established between adjacent nodes to form an intra-layer skeleton structure. When the height difference and slope between adjacent layers meet preset conditions (e.g., Δz≤0.1m, slope≤15°), inter-layer connections are established between the corresponding nodes to represent the vertical connectivity of ramps or passages. Finally, a skeleton topology map containing multiple layers of nodes, connections, and inter-layer connections is generated. This skeleton topology map fully represents the spatial structure and hierarchical relationships of passable areas in the vehicle environment and can serve as input data for the path planning module to achieve subsequent multi-objective path search and reversible maneuver strategy planning.

[0043] In some exemplary embodiments of this application, the method further includes: when the slope of the current driving route of the vehicle is greater than a preset slope, controlling the vehicle to perform an avoidance operation, the avoidance operation including a reversing operation.

[0044] In the above embodiments, by controlling the vehicle to perform avoidance operations, including reversing, when the slope of the vehicle's current driving route exceeds a preset slope, the system can actively adjust the vehicle's driving posture and path direction when the slope is too steep or traffic safety is reduced, avoiding the risks of insufficient power, slippage, or instability caused by excessively steep slopes. This step enables the system to maintain the vehicle's safe and controllable state in complex slopes or multi-level traffic environments. When the slope exceeds the limit, it can promptly trigger reversing or replan the route, thereby improving the vehicle's environmental adaptability and operational safety in uneven terrain, and ensuring the feasibility and reliability of the path planning results in actual road environments.

[0045] In some other exemplary embodiments of this application, the method further includes: determining the occupancy rate of an obstacle, the occupancy rate representing the degree to which the obstacle occupies the passage area of ​​the vehicle; and controlling the vehicle to perform an avoidance operation when the occupancy rate is greater than a preset occupancy rate.

[0046] In the above embodiments, by determining the obstacle encroachment rate and controlling the vehicle to perform avoidance operations when the encroachment rate exceeds a preset threshold, the system can take proactive maneuvering measures when obstacles significantly occupy passage space or pose a passage risk, thereby preventing the vehicle from approaching or colliding with the obstacle. The encroachment rate reflects the real-time impact of the obstacle on the vehicle's passable area. By dynamically calculating the encroachment rate and comparing it with a preset safety threshold, the system can achieve quantitative assessment and automatic response to potential passage risks. When the encroachment rate exceeds the safe range, the system triggers corresponding avoidance operations (such as deceleration, detour, or reversing), effectively improving the vehicle's safety and autonomous decision-making capabilities in dynamic and complex environments.

[0047] Specifically, the aforementioned avoidance maneuvers also include turning maneuvers, lateral detour maneuvers, and temporary stopping and rerouting maneuvers. Turning maneuvers: When the vehicle detects a closed area ahead or an obstructed path, it performs a turning maneuver using a combination of forward, reverse, and steering actions to change the vehicle's direction of travel and re-enter a passable path. This maneuver can be based on hybrid A... The algorithm (Hybrid A, a heuristic search algorithm for finding the shortest path from start to finish in a grid map) plans a sequence of reversible paths, ensuring minimum radius U-turns and direction adjustments in confined spaces. Lateral detour operation: When obstacles exist near the vehicle's path but local passable space remains, a local offset path is generated in the current skeleton topology map, allowing the vehicle to bypass the obstacle laterally and return to the main route. This operation can quickly avoid local obstacles without affecting the overall planned path, reducing the frequency of global replanning and improving traffic efficiency. Temporary stop and replanning operation: When a steep slope, dense obstacles, or sensor anomalies are detected, causing the current route to no longer be safe to pass, the system controls the vehicle to perform a safe stop operation and re-invokes the SLAM mapping and path planning modules to update environmental information and generate a new drivable route.

[0048] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the vehicle routing method of this application will be described in detail below with reference to specific embodiments.

[0049] This embodiment relates to a specific vehicle path planning method. First, environmental point cloud and attitude information around the vehicle are collected using a LiDAR, camera, and inertial measurement unit installed on the vehicle to obtain preliminary point cloud data. Based on the SLAM algorithm, the continuously acquired point cloud frames are time-synchronized and spatially registered to obtain a dense environmental point cloud map in a unified coordinate system. Subsequently, noise filtering, ground segmentation, and obstacle removal are performed on the point cloud, and layering is performed based on height differences to identify passable areas in each layer. Morphological thinning or median transformation algorithms are used to extract the centerline of the passable area, and the centerline is discretized into multiple skeleton nodes with a preset step size. Connections between adjacent nodes are established to construct a skeleton topology map representing the passable spatial structure of the vehicle. The operating state of the autonomous driving system is managed through a state machine mechanism, which includes at least Exploring (exploration and mapping state), Routing (path planning state), Parking (parking state), and BackToLane (return to lane state). The system switches states based on environmental changes, task execution status, and path feasibility: in Exploring state, it performs environmental exploration and SLAM mapping; in Routing state, it performs global path planning; in Parking state, it performs parking tasks; and in Back To Lane state, it plans a return lane path. State transitions are based on dynamic obstacle detection results and path traversability analysis, thus achieving coordinated management of global task logic and local maneuver planning. In Routing or Back To Lane states, reversible maneuver trigger conditions are continuously monitored. A reversible maneuver strategy is triggered when any of the following conditions are met: the current destination distance is less than twice the vehicle length; the forward traffic bandwidth is less than 1.2 times the vehicle width; and the predicted dynamic obstacle encroachment rate is greater than 30%. The traffic bandwidth is determined by the distance between the current node and adjacent nodes in the SLAM skeleton graph, and the dynamic obstacle encroachment rate reflects the real-time proportion of obstacles occupying the vehicle's passage area. When the trigger conditions are met, the system performs avoidance operations such as reversing, turning, or detouring, achieving adaptive adjustment of the local path. A unified cost function is used to comprehensively score different candidate paths; when a reversible maneuver strategy is triggered, a hybrid A / B algorithm is employed. The algorithm performs path planning. It simultaneously expands the search tree from both the starting point (forward direction) and the ending point (reverse direction), generating a smooth and continuous drivable trajectory based on the vehicle's kinematics model. A cost coefficient of 1.5 is introduced for the reversing path to reflect the additional cost of the reversing action. Through bidirectional search and cost constraints, the system can generate optimal paths containing both forward and reverse segments in narrow spaces or complex scenarios. In the Back To Lane state, the system records the curvature sequence of the most recent N paths and calculates hash values ​​to form a path feature table. When the hash matching degree between the newly planned path and the historical paths exceeds 80%, the system automatically triggers backup path planning to prevent repeated entry into the same dead-end areas, improving the system's robustness. In the Routing and Parking states, the system dynamically updates the weights of affected skeleton edges based on the speed and distance information of obstacles. The weight update function is: f(n) = g(n) + h(n) + λd(n), where g(n) is the current cumulative cost, h(n) is the heuristic cost, d(n) is the distance between the node and the obstacle, and λ is the dynamic obstacle avoidance weight coefficient. Where g(n) represents the cumulative path cost from the starting point to the current node n, which is the actual cost term that the algorithm gradually accumulates during the search tree expansion process. In this embodiment, the cumulative cost consists of the motion cost between each adjacent skeleton node on the path, including factors such as travel distance, steering angle change, and reversing operation. The calculation formula is as follows: Where g(p) is the cumulative cost of the previous skeleton node; ΔL is the Euclidean distance between the current skeleton node and the previous skeleton node; Δθ is the change in steering angle, used to reflect the turning complexity; ΔR is the reversing cost coefficient (e.g., 1.5 times the distance cost) if the current segment is a reversing path, and ΔR=0 if it is not a reversing segment (i.e., the vehicle is in a forward state); wL, wM, and wR are the weight coefficients of length, maneuver, and reversible cost, respectively. By updating the edge weights in real time, the path cost can be automatically increased when an obstacle approaches, guiding the vehicle to actively avoid obstacles. After the path planning result is generated, the control module executes steering, acceleration / deceleration, and reversible maneuver actions according to the target path. During driving, the SLAM module updates the vehicle pose in real time. When environmental changes, obstacle approach, or path infeasibility are detected, the system dynamically recalculates the cost and triggers local replanning, forming a continuous closed-loop control. Through the coordination of state machine management and reversible maneuver strategy, the organic integration of global path planning and local obstacle avoidance maneuver is achieved. Hybrid A The bidirectional search and dynamic edge weight adjustment mechanism enables vehicles to make stable and autonomous passage decisions even in narrow, congested, or high-precision map-free scenarios. By combining state memory mechanism to prevent repeated advances and retreats, path deduplication planning, and reversible cost control, the system's adaptability and safety in complex dynamic environments are effectively improved.

[0050] In a preferred embodiment of the present invention, to achieve autonomous exploration and multi-task collaborative control of vehicles in environments without high-precision maps, a state machine-based automated valet parking control architecture is provided. This architecture, by constructing clear task states and transition logic, enables the orderly switching of the system across different driving stages such as exploration, path planning, overtaking, and parking, thereby ensuring the continuity and stability of the entire parking process. The following detailed description of the structure and operation of the aforementioned state machine-based automated valet parking (AVP) control method, in conjunction with specific embodiments, illustrates this: This embodiment also relates to a state machine-based automated valet parking (AVP) control method. The automated valet parking system provided in this embodiment is suitable for Level 2 autonomous vehicles in environments without high-precision maps. The system generates an environmental map through autonomous exploration and performs tasks such as path planning, overtaking, U-turns, and parking based on this map. The entire system adopts a state machine-driven architecture, achieving autonomous parking of the vehicle in unknown environments through multi-task collaboration. The system mainly includes the following modules: AVP state machine module; Exploring module; Routing module; Overtaking module; Parking module; Back To Lane module; U-Turn module; NavToParkingSpace module; and Route optimization and task decision module. These modules are linked through state transition logic to achieve a continuous and stable automated valet parking process. The system uses a finite state machine (FSM) for task scheduling. Based on the vehicle's current task state and perception results, the system can dynamically switch between different states. The state machine includes the following states: AVPNotReady (Not Ready): The system is initialized or no task instruction has been received; Exploring: The vehicle autonomously explores unknown areas and generates a local environment map; Routing: The vehicle travels along the centerline or planned path to the next exploration or parking target; Overtaking: When the current driving path is blocked by an obstacle, the vehicle performs an overtaking action; Parking: The vehicle decelerates, detects and enters a parking space; BackToLane: When the vehicle explores to a dead end or fails to park, it performs a U-turn and returns to the lane; AVPPished (Task Completed): The automatic parking task is completed.During mission execution, the vehicle switches states based on environmental conditions, with the specific transition relationships as follows: When the vehicle detects an obstacle ahead during exploration, it switches from Exploring to Overtaking; when the vehicle finds an available parking space at a suitable distance, it switches from Exploring to Parking; when there is no centerline ahead and a path target exists, it switches from Exploring to BackToLane; when the vehicle approaches the target lane after a U-turn and its previous state was Routing, it switches from BackToLane to Routing; when the vehicle reaches its destination, it switches from Routing to Exploring; when an obstacle is detected on the path, it switches from Routing to Overtaking; when a suitable parking space is detected during the journey, it switches from Routing to Parking; when the vehicle can re-enter the path after overtaking... When returning to the centerline and the previous state was Exploring, the system transitions from Overtaking to Exploring; when the previous state was Routing and the vehicle can return to the centerline, the system transitions from Overtaking to Routing; when a suitable parking space is detected during overtaking, the system transitions from Overtaking to Parking; when parking fails or the overlap between the vehicle and the parking space is small, the system transitions from Parking to BackToLane; when the vehicle makes a U-turn and approaches the target lane and the previous state was Exploring, the system transitions from BackToLane to Exploring; when the previous state was Routing and the vehicle approaches the target lane, the system transitions from BackToLane to Routing; when the parking task is completed and the overlap between the vehicle and the parking space is large, the system transitions from Parking to AVPPished. Through the above state transition logic, the system can achieve automatic connection between tasks and self-recovery from anomalies. AVP Exploration Algorithm (Exploring Module): In the exploration phase, the system mainly uses the MCTS (Monte Carlo Tree Search) algorithm. This algorithm uses the vehicle's current position as the root node to search and evaluate possible future paths, thereby selecting the optimal exploration direction. To improve the continuity and feasibility of paths, the system introduces the following mechanisms: Path Optimizer: Corrects the MCTS output path according to road rules (such as driving on the right); Refresh Search: Periodically re-evaluates local paths to prevent search bias; Refresh Evaluation: Updates node scores to improve decision accuracy; Blocked Replan: Automatically triggers path recalculation when obstacles are encountered. When the optimized path differs significantly from the search path, the system will re-execute MCTS based on the optimized path to avoid erroneous exploration decisions.Meanwhile, the system supports cross-frame state extension, ensuring temporal consistency in the search process. The AVP routing module can serve as a routing tool for remote mobility assistance (RMA) and is also suitable for any task with the following input information: a map constructed by SLAM; a custom start and end point; and several key nodes that must be visited (e.g., parking areas). The system completes path calculation through the following steps: extracting the road skeleton structure from the SLAM map; establishing connecting edges between adjacent nodes; establishing specific connections between key nodes (such as parking spaces); calculating the optimal access order of key nodes using the TSP (Traveling Salesman Problem) algorithm; and searching for the shortest path on the skeleton based on the access order. In the final generated path, green lines and arrows indicate the access order of key nodes, and yellow paths indicate the shortest driving trajectory. The path output by this module can be directly executed by the vehicle or used for MPP (Memorized Parking Pilot), i.e., the vehicle drives according to the memorized trajectory. The AVP U-turn module automatically triggers U-turn logic when the vehicle reaches a dead end during exploration. This module is based on Hybrid A. Search algorithm (Hybrid A) The system establishes a U-turn task and calculates feasible reverse paths. Two U-turn strategies are available: if the target point is close, reverse directly to the target; if the target point is far, complete the U-turn first, then drive forward to the target point. After the U-turn, the system executes a new routing task to the unexplored area. The module identifies potential gaps around the trajectory and uses them as the next exploration destination. Then, the Exploring module is reused to reach the target area via the optimal path, achieving continuous exploration. The AVP Navigation to Parking Space module (NavToParkingSpace) introduces a new state, "Navigate to Parking Space," to improve parking success rate. After detecting an empty parking space, the system first evaluates the relative position and angle between the vehicle and the parking space; if the posture does not yet meet parking conditions, an intermediate navigation path is generated to guide the vehicle closer to the parking space; when the vehicle position and angle are suitable, it automatically switches to parking state; if the parking space is occupied during navigation or the posture cannot be adjusted to a suitable position, the system returns to the exploration state to continue the task. This module helps the system avoid missing parking opportunities due to mismatched position and attitude. The AVP parking module (Parking): In parking mode, the system determines the availability of parking spaces based on real-time perception data and executes path following control. After parking, the system calculates the overlap between the vehicle and the parking space: if the overlap is large, the task ends and the system enters the AVPFinished state; if the overlap is small or an anomaly occurs, the system transitions to the BackToLane state for replanning. The entire automated valet parking process can be summarized as follows: After system startup, it enters the AVPotReady state; upon receiving the parking task, it switches to the Exploring state to explore the environment and generate a map; during exploration, the MCTS algorithm is called to expand the path, and based on the results, it enters the corresponding sub-task: if an obstacle is found, it enters the Overtaking state; if an empty parking space is detected, it enters the NavToParkingSpace state; if a dead end is reached, it enters the BackToLane or UTurn state; when the target parking space is reached and the attitude is appropriate, the system executes the Parking state to complete parking; after parking, it enters the AVPFinished state, and the task ends. The entire process achieves complete automation logic from exploration, navigation, obstacle avoidance, U-turn to parking, exhibiting high intelligence and robustness. Through the above embodiments, this invention can achieve autonomous vehicle exploration and automatic parking in environments without high-precision maps. The system has advantages such as adaptive exploration and path planning, dynamic task scheduling, intelligent parking guidance, and robust anomaly recovery.This embodiment achieves autonomous map construction and path planning based on real-time perception data by combining the Monte Carlo Tree Search (MCTS) algorithm with a path optimizer. A state machine-based task switching mechanism enables seamless integration between multiple tasks such as exploration, path following, overtaking, and parking. The introduction of a "NavToParkingSpace" state effectively improves the accuracy and success rate of parking guidance. Furthermore, "BackToLane" and "UTurn" mechanisms enable task self-recovery and stable execution in complex environments. The AVP state machine and task control system provided in this embodiment have a reasonable structure and complete logic, enabling highly reliable automated valet parking functionality on a low-cost hardware platform and without high-precision maps.

[0051] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0052] This application also provides a vehicle routing device. It should be noted that the vehicle routing device of this application can be used to execute the vehicle routing method provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0053] The following describes the vehicle routing device provided in the embodiments of this application.

[0054] Figure 2 This is a schematic diagram of a vehicle routing device according to an embodiment of this application. Figure 2 As shown, the device includes:

[0055] The first acquisition unit 10 is used to acquire point cloud data of the vehicle environment and construct a skeleton topology map of the vehicle environment based on the point cloud data. The skeleton topology map represents the spatial structure of the passable area of ​​the vehicle.

[0056] The second acquisition unit 20 is used to acquire the current position of the vehicle and multiple preset target positions, and to determine multiple driving routes in the skeleton topology map based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position.

[0057] The first determining unit 30 is used to determine the score of each of the above driving routes according to a preset function, wherein the preset function is a function that is at least related to the number of curves of the above driving routes.

[0058] The second determining unit 40 is used to determine the target driving route based on the above score, and to control the driving of the vehicle based on the above target driving route.

[0059] In one optional embodiment, the first determining unit comprises: a first determining module, configured to determine the number of curves in each of the aforementioned driving routes, and to determine a complexity value for each of the aforementioned driving routes based on the number of curves, wherein the complexity value characterizes the operational complexity of the vehicle driving the aforementioned driving routes; a second determining module, configured to determine the route length of each of the aforementioned driving routes; a third determining module, configured to determine the number of obstacles in each of the aforementioned driving routes, and to determine an obstacle risk value based at least on the number of obstacles, wherein the obstacle risk value characterizes the degree of impact of the obstacles on the vehicle; a fourth determining module, configured to determine the length of the reversing route in each of the aforementioned driving routes, and to determine a reversible maneuver value based at least on the length of the reversing route, wherein the reversible maneuver value characterizes the proportion of the reversing route length to the driving route; The fifth determining module is used to determine that the first product is the product of the first preset weight coefficient and the aforementioned complexity value; the sixth determining module is used to determine that the second product is the product of the second preset weight coefficient and the aforementioned route length; the seventh determining module is used to determine that the third product is the product of the third preset weight coefficient and the aforementioned obstacle risk value; the eighth determining module is used to determine that the fourth product is the product of the fourth preset weight coefficient and the aforementioned reversible maneuver value; the ninth determining module is used to determine that the first sum is the sum of the aforementioned first product and the aforementioned second product; the tenth determining module is used to determine that the second sum is the sum of the aforementioned first sum and the aforementioned third product; the eleventh determining module is used to determine that the third sum is the sum of the aforementioned second sum and the aforementioned fourth product; and the twelfth determining module is used to determine that the aforementioned score of the aforementioned driving route is the aforementioned third sum.

[0060] In another alternative, the second determining unit includes a thirteenth determining module, used to determine the driving route corresponding to the minimum value among the scores as the target driving route.

[0061] As an optional solution, the first acquisition unit includes: an acquisition module for acquiring preliminary point cloud data of the vehicle environment; a preprocessing module for preprocessing the preliminary point cloud data to obtain the point cloud data, wherein the preprocessing includes removing noise points from the preliminary point cloud data; and a fourteenth determination module for determining passable areas in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable areas.

[0062] In one optional scheme, the fourteenth determining module includes: a first determining submodule, used to determine multiple access layers of the vehicle environment based on the point cloud data, and determine the passable area of ​​each access layer; an extraction submodule, used to extract the centerline of each passable area, and sample the centerline according to a preset step size to obtain multiple sampled centerlines; and a second determining submodule, used to determine the endpoints of each sampled centerline as skeleton nodes of the skeleton topology, and determine the skeleton topology of the vehicle environment based on the skeleton nodes.

[0063] In another alternative embodiment, the device further includes a first control unit, configured to control the vehicle to perform an avoidance operation when the gradient of the vehicle's current driving route is greater than a preset gradient, the avoidance operation including a reversing operation.

[0064] In some exemplary embodiments, the above-described apparatus further includes: a third determining unit, configured to determine the occupancy rate of an obstacle, the occupancy rate representing the degree to which the obstacle occupies the passage area of ​​the vehicle; and a second control unit, configured to control the vehicle to perform an avoidance operation when the occupancy rate is greater than a preset occupancy rate.

[0065] The aforementioned vehicle routing device includes a processor and a memory. The first acquisition unit, the second acquisition unit, the first determination unit, and the second determination unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve their respective functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0066] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can address the problem in existing technologies where autonomous vehicles cannot flexibly choose routes based on environmental structures in complex scenarios.

[0067] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0068] This invention provides a computer-readable storage medium including a stored program, wherein the program, when running, controls the device containing the computer-readable storage medium to execute the vehicle routing method.

[0069] Specifically, vehicle routing methods include:

[0070] Step S101: Obtain point cloud data of the vehicle environment, and construct a skeleton topology map of the vehicle environment based on the point cloud data. The skeleton topology map represents the spatial structure of the passable area of ​​the vehicle.

[0071] Step S102: Obtain the current position of the vehicle and multiple preset target positions, and determine multiple driving routes in the skeleton topology map based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position.

[0072] Specifically, each of the above-mentioned driving routes starts from the current location and passes through all the above-mentioned target locations in sequence, although the order of visits may differ.

[0073] Step S103: Determine the score of each of the above driving routes according to a preset function, wherein the preset function is a function that is at least related to the number of curves of the above driving routes.

[0074] Step S104: Determine the target driving route based on the above score, and control the vehicle to drive according to the above target driving route.

[0075] In one embodiment of this application, determining the score of each driving route according to a preset function includes: determining the number of curves in each driving route, and determining a complexity value for each driving route based on the number of curves, wherein the complexity value characterizes the operational complexity of the vehicle driving the driving route; determining the route length of each driving route; determining the number of obstacles in each driving route, and determining an obstacle risk value based at least on the number of obstacles, wherein the obstacle risk value characterizes the degree of impact of the obstacles on the vehicle; determining the length of the reversing route in each driving route, and determining a reversible maneuverability value based at least on the length of the reversing route. The reversible maneuver value represents the proportion of the reversing route length to the driving route; the first product is determined to be the product of the first preset weight coefficient and the complexity value; the second product is determined to be the product of the second preset weight coefficient and the route length; the third product is determined to be the product of the third preset weight coefficient and the obstacle risk value; the fourth product is determined to be the product of the fourth preset weight coefficient and the reversible maneuver value; the first sum is determined to be the sum of the first product and the second product; the second sum is determined to be the sum of the first sum and the third product; the third sum is determined to be the sum of the second sum and the fourth product; the above-mentioned fraction of the driving route is determined to be the third sum.

[0076] In one embodiment of this application, determining the target driving route based on the above scores includes: determining the driving route corresponding to the minimum value among the above scores as the target driving route.

[0077] In one embodiment of this application, acquiring point cloud data of a vehicle environment and constructing a skeleton topology map of the vehicle environment based on the point cloud data includes: acquiring preliminary point cloud data of the vehicle environment; preprocessing the preliminary point cloud data to obtain the point cloud data, wherein the preprocessing includes removing noise points from the preliminary point cloud data; determining passable areas in the point cloud data, and determining the skeleton topology map of the vehicle environment based on the passable areas.

[0078] In one embodiment of this application, determining a passable area in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable area includes: determining multiple access layers of the vehicle environment based on the point cloud data and determining the passable area of ​​each access layer; extracting the centerline of each passable area and sampling the centerline according to a preset step size to obtain multiple sampled centerlines; determining the endpoints of each sampled centerline as skeleton nodes of the skeleton topology and determining the skeleton topology of the vehicle environment based on the skeleton nodes.

[0079] In one embodiment of this application, the method further includes: when the gradient of the current driving route of the vehicle is greater than a preset gradient, controlling the vehicle to perform an avoidance operation, the avoidance operation including a reversing operation.

[0080] In one embodiment of this application, the method further includes: determining the occupancy rate of an obstacle, wherein the occupancy rate characterizes the degree to which the obstacle occupies the passage area of ​​the vehicle; and controlling the vehicle to perform an avoidance operation when the occupancy rate is greater than a preset occupancy rate.

[0081] This invention provides a vehicle system, including one or more processors, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps: Step S101, acquiring point cloud data of the vehicle environment, and constructing a skeleton topology map of the vehicle environment based on the point cloud data, wherein the skeleton topology map represents the spatial structure of the vehicle's passable area.

[0082] Step S102: Obtain the current position of the vehicle and multiple preset target positions, and determine multiple driving routes in the skeleton topology map based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position.

[0083] Specifically, each of the above-mentioned driving routes starts from the current location and passes through all the above-mentioned target locations in sequence, although the order of visits may differ.

[0084] Step S103: Determine the score of each of the above driving routes according to a preset function, wherein the preset function is a function that is at least related to the number of curves of the above driving routes.

[0085] Step S104: Determine the target driving route based on the above score, and control the vehicle to drive according to the above target driving route.

[0086] In one embodiment of this application, determining the score of each driving route according to a preset function includes: determining the number of curves in each driving route, and determining a complexity value for each driving route based on the number of curves, wherein the complexity value characterizes the operational complexity of the vehicle driving the driving route; determining the route length of each driving route; determining the number of obstacles in each driving route, and determining an obstacle risk value based at least on the number of obstacles, wherein the obstacle risk value characterizes the degree of impact of the obstacles on the vehicle; determining the length of the reversing route in each driving route, and determining a reversible maneuverability value based at least on the length of the reversing route. The reversible maneuver value represents the proportion of the reversing route length to the driving route; the first product is determined to be the product of the first preset weight coefficient and the complexity value; the second product is determined to be the product of the second preset weight coefficient and the route length; the third product is determined to be the product of the third preset weight coefficient and the obstacle risk value; the fourth product is determined to be the product of the fourth preset weight coefficient and the reversible maneuver value; the first sum is determined to be the sum of the first product and the second product; the second sum is determined to be the sum of the first sum and the third product; the third sum is determined to be the sum of the second sum and the fourth product; the above-mentioned fraction of the driving route is determined to be the third sum.

[0087] In one embodiment of this application, determining the target driving route based on the above scores includes: determining the driving route corresponding to the minimum value among the above scores as the target driving route.

[0088] In one embodiment of this application, acquiring point cloud data of a vehicle environment and constructing a skeleton topology map of the vehicle environment based on the point cloud data includes: acquiring preliminary point cloud data of the vehicle environment; preprocessing the preliminary point cloud data to obtain the point cloud data, wherein the preprocessing includes removing noise points from the preliminary point cloud data; determining passable areas in the point cloud data, and determining the skeleton topology map of the vehicle environment based on the passable areas.

[0089] In one embodiment of this application, determining a passable area in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable area includes: determining multiple access layers of the vehicle environment based on the point cloud data and determining the passable area of ​​each access layer; extracting the centerline of each passable area and sampling the centerline according to a preset step size to obtain multiple sampled centerlines; determining the endpoints of each sampled centerline as skeleton nodes of the skeleton topology and determining the skeleton topology of the vehicle environment based on the skeleton nodes.

[0090] In one embodiment of this application, the method further includes: when the gradient of the current driving route of the vehicle is greater than a preset gradient, controlling the vehicle to perform an avoidance operation, the avoidance operation including a reversing operation.

[0091] In one embodiment of this application, the method further includes: determining the occupancy rate of an obstacle, wherein the occupancy rate characterizes the degree to which the obstacle occupies the passage area of ​​the vehicle; and controlling the vehicle to perform an avoidance operation when the occupancy rate is greater than a preset occupancy rate.

[0092] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0093] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0094] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0096] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0098] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0099] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0100] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0103] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0104] The vehicle path planning method of this application acquires point cloud data of the vehicle environment and constructs a skeleton topology map representing the spatial structure of the vehicle's traversable area. Based on this, and combining the vehicle's current position with multiple preset target positions, it determines multiple candidate driving routes based on the skeleton topology map. It then uses preset functions related to traffic characteristics such as the number of curves to comprehensively score each driving route. By comparing the scores of different candidate routes, the optimal target driving route is determined, and the vehicle is controlled to travel along this target route. This enables autonomous vehicles to flexibly select travel paths according to the spatial structure of the environment, achieving intelligent and adaptive path planning, and improving the efficiency and safety of vehicle traffic in complex or irregular environments. It solves the problem in existing technologies where autonomous vehicles cannot flexibly select travel paths according to environmental structure in complex scenarios.

[0105] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A vehicle route planning method, characterized in that, include: The point cloud data of the vehicle environment is acquired, and a skeleton topology map of the vehicle environment is constructed based on the point cloud data. The skeleton topology map represents the spatial structure of the vehicle's passable area. The current position of the vehicle and multiple preset target positions are obtained, and multiple driving routes in the skeleton topology map are determined based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position; The score for each driving route is determined according to a preset function, wherein the preset function is a function that is at least related to the number of curves in the driving route; The target driving route is determined based on the score, and the vehicle is controlled to drive according to the target driving route.

2. The method according to claim 1, characterized in that, The score for each driving route is determined according to a preset function, including: The number of curves in each of the driving routes is determined, and the complexity value of each driving route is determined based on the number of curves. The complexity value characterizes the operational complexity of the vehicle driving the driving route. Determine the route length of each of the aforementioned driving routes; The number of obstacles in each of the driving routes is determined, and at least based on the number of obstacles, an obstacle risk value is determined, wherein the obstacle risk value characterizes the degree of impact of the obstacle on the vehicle; The length of the reversing route in each of the driving routes is determined, and at least based on the length of the reversing route, a reversible maneuver value is determined, wherein the reversible maneuver value represents the proportion of the length of the reversing route to the driving route; The first product is determined to be the product of the first preset weight coefficient and the complexity value; The second product is determined to be the product of the second preset weighting coefficient and the route length; The third product is determined to be the product of the third preset weighting coefficient and the obstacle risk value; The fourth product is determined to be the product of the fourth preset weighting coefficient and the reversible maneuver value; The first sum is determined to be the sum of the first product and the second product; The second sum is determined to be the sum of the product of the first sum and the third sum; The third sum is determined to be the sum of the product of the second sum and the fourth sum; The score of the driving route is determined to be the third sum value.

3. The method according to claim 1, characterized in that, Determining the target driving route based on the score includes: The driving route corresponding to the minimum score is determined as the target driving route.

4. The method according to claim 1, characterized in that, Acquire point cloud data of the vehicle environment, and construct a skeleton topology map of the vehicle environment based on the point cloud data, including: Obtain preliminary point cloud data of the vehicle environment; The pre-processed point cloud data is obtained by preprocessing the pre-point cloud data, and the preprocessing includes removing noise points from the pre-point cloud data. Determine the passable areas in the point cloud data, and determine the skeleton topology of the vehicle environment based on the passable areas.

5. The method according to claim 4, characterized in that, Determining passable areas in the point cloud data and determining the skeleton topology of the vehicle environment based on the passable areas includes: Based on the point cloud data, multiple access layers of the vehicle environment are determined, and the passable area of ​​each access layer is determined. Extract the center line of each passable area, and sample the center line according to a preset step size to obtain multiple sampled center lines; The endpoints of each of the sampling center lines are determined as skeleton nodes of the skeleton topology, and the skeleton topology of the vehicle environment is determined based on the skeleton nodes.

6. The method according to claim 1, characterized in that, The method further includes: If the gradient of the vehicle's current driving route is greater than a preset gradient, the vehicle is controlled to perform an avoidance operation, which includes reversing.

7. The method according to claim 1, characterized in that, The method further includes: Determine the occupancy rate of the obstacle, wherein the occupancy rate characterizes the degree to which the obstacle occupies the passage area of ​​the vehicle; If the encroachment rate is greater than a preset encroachment rate, the vehicle is controlled to perform an avoidance operation.

8. A vehicle route planning device, characterized in that, include: The first acquisition unit is used to acquire point cloud data of the vehicle environment and construct a skeleton topology map of the vehicle environment based on the point cloud data. The skeleton topology map represents the spatial structure of the vehicle's passable area. The second acquisition unit is used to acquire the current position of the vehicle and multiple preset target positions, and to determine multiple driving routes in the skeleton topology map based on the current position and multiple preset target positions, wherein each driving route includes the current position and each preset target position; The first determining unit is used to determine the score of each of the driving routes according to a preset function, wherein the preset function is a function that is at least related to the number of curves of the driving route; The second determining unit is used to determine the target driving route based on the score, and control the vehicle to drive according to the target driving route.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 7.

10. A vehicle system, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising methods for performing any one of claims 1 to 7.

Images