Method, device and electronic equipment for escape path planning

Abstract

This application provides a method, apparatus, and electronic device for planning an escape route for an automated valet parking vehicle. The method includes: acquiring reference lines, multiple anchor points, and scene type information of the scene where the vehicle is located; calculating the reference line cost of a search point based on the reference lines; calculating the anchor point cost of the search point based on the multiple anchor points; adjusting the weights of the reference line cost and anchor point cost according to the scene type information; and executing a path planning algorithm based on the adjusted reference line cost and anchor point cost to generate an escape route for the automated valet parking vehicle. This method aims to improve the search efficiency of the path planning algorithm.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to an escape path planning method, device and electronic device. Background Technology

[0002] Autonomous Valet Parking (AVP) is a crucial application scenario for autonomous driving, requiring vehicles to autonomously navigate from the parking lot entrance to the target parking space without human intervention. In actual parking situations, vehicles often encounter complex conditions such as U-turns, narrow passages, and obstructions, making it impossible to continue following the preset trajectory and leading to a stuck state. In such cases, path planning algorithms are needed to quickly output a passable escape route, enabling the vehicle to safely and efficiently escape the stuck situation and resume the autonomous parking process. In existing technologies, path planning algorithms often use heuristic functions with fixed weights to search for an escape route. However, existing technologies suffer from low search efficiency. Summary of the Invention

[0003] This application provides an escape path planning method, apparatus, and electronic device to improve the search efficiency of path planning algorithms.

[0004] In a first aspect, embodiments of this application provide an escape path planning method, including:

[0005] Obtain reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located.

[0006] The reference line cost of the search point is calculated based on the reference line, and the anchor point cost of the search point is calculated based on multiple anchor points.

[0007] Adjust the weights of reference line cost and anchor point cost based on scene type information.

[0008] A path planning algorithm is executed based on the adjusted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle.

[0009] In one possible implementation, in conjunction with the first aspect, calculating the reference line cost of the search point based on the reference line includes:

[0010] Determine the matching reference point on the reference line for the search point.

[0011] Calculate the distance cost between the search point and the matching reference point.

[0012] Calculate the heading angle cost between the heading of the search point and the heading of the reference line of the matching reference point.

[0013] Calculate the progress incentive cost of the search point along the reference line toward the target direction.

[0014] The reference line cost is generated based on the distance cost, heading angle cost, and progress incentive cost.

[0015] In one possible implementation, prior to obtaining multiple anchor points of the scene where the automated valet parking vehicle is located, in conjunction with the first aspect, the following is included:

[0016] Extract the reference line segment from the vehicle's current position to the target point.

[0017] Calculate the average curvature, maximum curvature, heading change, and path length of the reference line segment.

[0018] A path complexity score is generated based on the average curvature, maximum curvature, heading change, and path length.

[0019] The number of anchor points, anchor point spacing, and aiming distance are determined based on the path complexity score.

[0020] Generate primary and secondary anchor points on the reference line segment according to the determined number of anchor points, anchor point spacing, and pre-aiming distance; multiple anchor points include primary and secondary anchor points.

[0021] In one possible implementation, in conjunction with the first aspect, calculating the anchor cost of the search point based on multiple anchor points includes:

[0022] The anchor distance cost and anchor angle cost of the search point are calculated based on the main anchor point and auxiliary anchor points.

[0023] Anchor point costs are generated based on anchor point distance costs and anchor point angle costs.

[0024] In one possible implementation, before acquiring the reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, in conjunction with the first aspect, the method further includes:

[0025] Acquire vehicle status information and environmental perception data.

[0026] The duration of stationary position and trajectory planning status are determined based on vehicle status information.

[0027] The distribution of obstacles and road boundary information are determined based on environmental perception data.

[0028] An escape trigger flag is generated based on the static duration, trajectory planning status, obstacle distribution, and road boundary information.

[0029] In one possible implementation, before obtaining the reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, in conjunction with the first aspect, the method further includes:

[0030] Identify vehicle meeting scene markers and yielding scene markers based on environmental perception data.

[0031] Generate a trouble-relief flag based on the vehicle meeting scenario flag and the yielding scenario flag.

[0032] The timing of execution is controlled based on the timing of obtaining reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, according to the traction trigger flag and traction inhibition flag.

[0033] In one possible implementation, in conjunction with the first aspect, before calculating the reference line cost of the search point based on the reference line, the method further includes:

[0034] A KD-tree index is constructed based on multiple reference points on the reference line.

[0035] Query the KD-tree index to find a matching reference point based on the vehicle's current location.

[0036] Search for candidate target points along the reference line, starting from the matching reference point.

[0037] Use the direction of the reference line where the candidate target point is located as the target direction.

[0038] In one possible implementation, in conjunction with the first aspect, a path planning algorithm is executed based on the weighted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle, including:

[0039] Initialize the set of search nodes and kinematic extension parameters in the path planning algorithm.

[0040] The comprehensive cost of candidate search nodes is calculated based on obstacle distance cost, weighted reference line cost, and anchor point cost.

[0041] Expand candidate search nodes according to the overall cost.

[0042] When the distance between the candidate search node and the target point meets the preset connection conditions, a candidate parsing path from the candidate search node to the target point is generated by a preset vehicle geometry path generator.

[0043] Generate an escape path based on candidate parsing paths.

[0044] Secondly, embodiments of this application provide an escape path planning device, comprising:

[0045] The acquisition module is used to acquire reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located.

[0046] The calculation module is used to calculate the reference line cost of the search point based on the reference line, and to calculate the anchor point cost of the search point based on multiple anchor points.

[0047] The weight adjustment module is used to adjust the weights of the reference line cost and anchor point cost based on the scene type information.

[0048] The path planning module is used to execute a path planning algorithm based on the adjusted reference line cost and anchor point cost to generate an escape path for automated valet parking vehicles.

[0049] In one possible implementation, in conjunction with the second aspect, the computing module is specifically used for:

[0050] Determine the matching reference point on the reference line for the search point.

[0051] Calculate the distance cost between the search point and the matching reference point.

[0052] Calculate the heading angle cost between the heading of the search point and the heading of the reference line of the matching reference point.

[0053] Calculate the progress incentive cost of the search point along the reference line toward the target direction.

[0054] The reference line cost is generated based on the distance cost, heading angle cost, and progress incentive cost.

[0055] In one possible implementation, in conjunction with the second aspect, the acquisition module is specifically used for:

[0056] Extract the reference line segment from the vehicle's current position to the target point.

[0057] Calculate the average curvature, maximum curvature, heading change, and path length of the reference line segment.

[0058] A path complexity score is generated based on the average curvature, maximum curvature, heading change, and path length.

[0059] The number of anchor points, anchor point spacing, and aiming distance are determined based on the path complexity score.

[0060] Generate primary and secondary anchor points on the reference line segment according to the determined number of anchor points, anchor point spacing, and pre-aiming distance; multiple anchor points include primary and secondary anchor points.

[0061] In one possible implementation, in conjunction with the second aspect, the computing module is specifically used for:

[0062] The anchor distance cost and anchor angle cost of the search point are calculated based on the main anchor point and auxiliary anchor points.

[0063] Anchor point costs are generated based on anchor point distance costs and anchor point angle costs.

[0064] In one possible implementation, in conjunction with the second aspect, the acquisition module is further specifically used for:

[0065] Acquire vehicle status information and environmental perception data.

[0066] The duration of stationary position and trajectory planning status are determined based on vehicle status information.

[0067] The distribution of obstacles and road boundary information are determined based on environmental perception data.

[0068] An escape trigger flag is generated based on the static duration, trajectory planning status, obstacle distribution, and road boundary information.

[0069] In one possible implementation, in conjunction with the second aspect, the acquisition module is further specifically used for:

[0070] Identify vehicle meeting scene markers and yielding scene markers based on environmental perception data.

[0071] Generate a trouble-relief flag based on the vehicle meeting scenario flag and the yielding scenario flag.

[0072] The timing of execution is controlled based on the timing of obtaining reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, according to the traction trigger flag and traction inhibition flag.

[0073] In one possible implementation, in conjunction with the second aspect, the calculation module is further specifically used for:

[0074] A KD-tree index is constructed based on multiple reference points on the reference line.

[0075] Query the KD-tree index to find a matching reference point based on the vehicle's current location.

[0076] Search for candidate target points along the reference line, starting from the matching reference point.

[0077] Use the direction of the reference line where the candidate target point is located as the target direction.

[0078] In one possible implementation, in conjunction with the second aspect, the path planning module is specifically used for:

[0079] Initialize the set of search nodes and kinematic extension parameters in the path planning algorithm.

[0080] The comprehensive cost of candidate search nodes is calculated based on obstacle distance cost, weighted reference line cost, and anchor point cost.

[0081] Expand candidate search nodes according to the overall cost.

[0082] When the distance between the candidate search node and the target point meets the preset connection conditions, a candidate parsing path from the candidate search node to the target point is generated by a preset vehicle geometry path generator.

[0083] Generate an escape path based on candidate parsing paths.

[0084] Thirdly, embodiments of this application provide an escape path planning device, including: a memory, and a processor communicatively connected to the memory;

[0085] The memory stores the instructions that the computer executes;

[0086] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0087] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0088] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0089] This application provides an escape path planning method, device, and electronic device. By acquiring reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, the method calculates the reference line cost of the search point based on the reference lines and the anchor point cost based on the multiple anchor points. The weights of the reference line cost and anchor point cost are adjusted according to the scene type information. A path planning algorithm is then executed based on the adjusted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle. This technical approach ensures the path follows the reference line direction through reference line cost constraints, avoiding deviation from the road structure; it constrains the path direction through multi-anchor point cost guidance, reducing the randomness of the search path, especially avoiding sharp turns or jagged paths in U-shaped bends or narrow passage scenarios; and it dynamically adjusts the cost weights according to the scene type information, further optimizing path smoothness and feasibility. Ultimately, this technical approach improves the search efficiency of the path planning algorithm in complex and constrained environments. Attached Figure Description

[0090] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0091] Figure 1 A schematic diagram of a scenario for an escape path planning method provided in this application;

[0092] Figure 2 A flowchart illustrating an escape path planning method provided in this application. Figure 1 ;

[0093] Figure 3 A flowchart illustrating an escape path planning method provided in this application. Figure 2 ;

[0094] Figure 4 A specific example diagram of an escape path planning method provided in this application;

[0095] Figure 5 A schematic diagram of the structure of an escape path planning device provided in this application;

[0096] Figure 6 This is a structural schematic diagram of an escape path planning device provided in this application.

[0097] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0098] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0099] The application background of the embodiments of this application will be explained below:

[0100] Automated valet parking is a crucial application scenario for autonomous driving, requiring vehicles to autonomously navigate from the parking lot entrance to the target parking space without human intervention. During actual parking, vehicles often encounter complex conditions such as U-turns, narrow passages, and obstructions, making it impossible to continue following the preset trajectory and resulting in a stuck state. In such situations, path planning algorithms are needed to quickly output a passable escape route, enabling the vehicle to safely and efficiently escape the pre-defined path and resume the autonomous parking process. Current solutions often employ heuristic functions with fixed weights to search for an escape route. However, existing technologies suffer from low search efficiency.

[0101] To address the aforementioned problems, the inventors propose an escape path planning method. This method involves acquiring reference lines, multiple anchor points, and scene type information of the automated valet parking vehicle's location. Based on the reference lines, the reference line cost of the search point is calculated, and based on the multiple anchor points, the anchor point cost is calculated. The weights of the reference line cost and anchor point cost are adjusted according to the scene type information. Based on the adjusted reference line cost and anchor point cost, a path planning algorithm is executed to generate an escape path for the automated valet parking vehicle. This technique ensures the path follows the reference line direction through reference line cost constraints, avoiding deviation from the road structure. It also guides the path direction through multiple anchor point costs, reducing the randomness of the search path, especially avoiding sharp turns or jagged paths in U-shaped bends or narrow passage scenarios. Furthermore, it dynamically adjusts the cost weights according to the scene type information, further optimizing path smoothness and feasibility. Ultimately, this technique improves the search efficiency of the path planning algorithm in complex and constrained environments.

[0102] Taking the scenario of a vehicle encountering a narrow passage in a parking lot as an example, combined with Figure 1 This illustrates the specific application scenario of the obstacle avoidance path planning method provided in this application. For example... Figure 1 As shown, the specific application scenarios of this application include a perception system 101, a positioning system 102, a navigation system 103, and a controller 104 capable of performing path planning, all equipped on a vehicle. The controller 104 collects vehicle status information and environmental perception data through the perception system 101, obtains a reference line through the navigation system 103, and acquires the vehicle's current position through the positioning system 102. After determining whether to enter the traction escape mode based on the vehicle status information and environmental perception data, it generates an traction escape path based on the vehicle status information, environmental perception data, reference line, and the vehicle's current position.

[0103] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0104] Figure 2 A flowchart illustrating an escape path planning method provided in this application. Figure 1 ,like Figure 2 As shown, the method includes:

[0105] S201. Obtain the reference line, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located.

[0106] In this step, multiple anchor points include primary anchor points and secondary anchor points, which can be determined based on the complexity of the route from the vehicle's current position to the target point. Scene type information can be determined by analyzing reference lines and environmental information, including special scenes such as U-turns (heading changes exceeding 120 degrees), narrow passages (passage width less than 2.5 meters), and multi-level passages.

[0107] S202. Calculate the reference line cost of the search point based on the reference line, and calculate the anchor point cost of the search point based on multiple anchor points.

[0108] In this step, the matching reference point on the reference line is first determined. Then, the distance cost between the search point and the matching reference point is calculated. Next, the heading angle cost between the search point's heading and the matching reference point's heading along the reference line is calculated. Then, the progress incentive cost of the search point along the reference line towards the target direction is calculated. Finally, the reference line cost is generated based on the distance cost, heading angle cost, and progress incentive cost. Simultaneously, the anchor point distance cost and anchor point angle cost of the search point are calculated based on the primary and secondary anchor points among the multiple anchor points. Then, the anchor point cost is generated based on the anchor point distance cost and anchor point angle cost.

[0109] S203. Adjust the weights of reference line cost and anchor point cost based on scene type information.

[0110] In this step, the curvature of the matching reference point, the density of obstacles around the matching reference point, and the rate of change of the heading of the reference line segment are first obtained. Then, the dynamic distance factor corresponding to the distance cost is calculated based on the curvature and obstacle density, and the dynamic angle factor corresponding to the anchor point angle cost is calculated based on the rate of change of heading. Subsequently, the weights corresponding to the distance cost are adjusted based on the dynamic distance factor, and the weights corresponding to the anchor point angle cost are adjusted based on the dynamic angle factor.

[0111] S204. Based on the adjusted reference line cost and anchor point cost, execute the path planning algorithm to generate an escape path for the automated valet parking vehicle.

[0112] In this step, the search node set and kinematic extension parameters in the path planning algorithm are first initialized. Then, the comprehensive cost of the candidate search nodes is calculated based on the obstacle distance cost, the weighted reference line cost, and the anchor point cost. Subsequently, the candidate search nodes are expanded according to the comprehensive cost. When the distance between the candidate search node and the target point meets the preset connection conditions, a candidate analytical path from the candidate search node to the target point is generated through a preset vehicle geometry path generator. Finally, an escape path is generated based on the candidate analytical path. In some embodiments, the preset vehicle geometry path generator is a Reeds-Shepp planner or a Clothoid spiral planner.

[0113] In one possible implementation, after generating the escape path, a smoothing process is performed on the escape path, the escape path is updated according to the smoothed curvature sequence, the gear sequence is pruned at the end of the escape path, and the pruned escape path is output.

[0114] In one possible implementation, the path planning algorithm is a hybrid A* search. The specific process of generating an escape path for an automated valet parking vehicle based on the weighted reference line cost and anchor point cost is as follows:

[0115] The first step is initialization. Initializing the search node set and kinematic extension parameters in the path planning algorithm includes: setting the coordinate resolution (default 0.2 meters) and heading resolution (default 5 degrees) to construct a 2D grid map; setting the steering wheel angle discrete number (default 3 directions: left turn, straight, right turn) and calculating the steering step size for each direction; initializing the collision detector and dynamically adjusting the detection level according to the search progress (fast detection in the early stage and fine detection in the later stage); and initializing the node expander and configuring the extension parameters for both forward and backward driving directions.

[0116] The second step is the search. The starting node is added to the open list (priority queue), and its total cost (heuristic cost + traveled cost) is calculated. Here, the heuristic cost includes obstacle distance cost, weighted reference line cost, and anchor point cost. The node with the lowest cost is selected from the open list as the current node. The current node is expanded to generate multiple successor nodes, each corresponding to an expansion action (e.g., forward left turn, forward straight, backward right turn, etc.). Collision detection is performed on each successor node; if successful, its heuristic cost and traveled cost are calculated, and it is added to the open list. The process of selecting the node with the lowest cost from the open list as the current node and subsequent steps is repeated until the target point is reached or the search times out. During node expansion, multiple expansion strategies are considered, including: an arc expansion strategy that generates arc path segments according to a preset turning angle and step size; a straight expansion strategy that generates straight path segments when the turning angle is 0; and a forward / backward expansion strategy that expands in both forward and backward directions, with the expansion step size dynamically adjusted according to the scene (default 1 meter). The expanded node positions and headings are calculated based on the vehicle kinematics model to ensure that the path meets the vehicle motion constraints.

[0117] The third step involves using an analytical planner for connection. When a search node approaches the target point (distance less than 3 meters), an analytical planner, such as the Reeds-Shepp planner or the Clothoid planner, is attempted to directly connect it to the target point. The analytical planner generates optimal paths that satisfy vehicle kinematic constraints, improving planning efficiency and path quality. If the analytical planning succeeds, the path is returned directly; otherwise, a hybrid A* search continues.

[0118] The fourth step is path smoothing and post-processing. After a successful search, the generated path is smoothed. This can be done using a Quadratic Programming Smoother (QP) smoother to perform secondary planning and smoothing on each path segment individually, optimizing the curvature and rate of curvature change of path points; alternatively, an iterative anchor point smoother can be used to iteratively smooth the entire path, using anchor point constraints to keep the path within a safe distance from obstacles. Then, the final reverse gear segment is trimmed to ensure the end of the path is in the forward direction, facilitating a switch back to cruise mode later.

[0119] Step 5: Path Evaluation and Candidate Selection. During the search process, each feasible path found is added to the candidate path list, and its quality is evaluated based on the following criteria: prioritizing shorter paths; reducing the number of forward / backward switching operations; and selecting paths with less curvature to improve feasibility. When the number of candidate paths reaches a threshold (default 5) or the search time expires, the optimal path is selected from the candidate paths and returned to obtain the escape route for the automated valet parking vehicle.

[0120] This application provides an escape path planning method that obtains reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located. It calculates the reference line cost of the search point based on the reference lines and the anchor point cost based on the multiple anchor points. The weights of the reference line cost and anchor point cost are adjusted according to the scene type information. A path planning algorithm is then executed based on the adjusted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle. This technical approach ensures that the path follows the reference line direction through reference line cost constraints, avoiding deviation from the road structure. It also constrains the path direction through multi-anchor point cost guidance, reducing the randomness of the search path, especially avoiding sharp turns or jagged paths in U-shaped bends or narrow passage scenarios. Furthermore, it dynamically adjusts the cost weights according to the scene type information, further optimizing the path smoothness and feasibility. Ultimately, this technical approach improves the search efficiency of the path planning algorithm in complex and constrained environments.

[0121] Figure 3 A flowchart illustrating an escape path planning method provided in this application. Figure 2 ,like Figure 3 As shown, in this embodiment... Figure 2 Based on the embodiments, a method for planning an escape route is described in detail, which includes:

[0122] S301. Obtain vehicle status information and environmental perception data.

[0123] In this step, the vehicle's real-time motion parameters can be collected by the perception system on the vehicle as described in the previous embodiments to obtain vehicle status information, and the current parking scenario can be detected to obtain environmental perception data such as obstacles and passage boundaries.

[0124] S302. Based on vehicle status information and environmental perception data, determine whether to enter the escape mode.

[0125] In this step, the duration of stillness and trajectory planning status can be determined based on vehicle status information. Then, obstacle distribution and road boundary information can be determined based on environmental perception data. An escape trigger flag is generated based on the stillness duration, trajectory planning status, obstacle distribution, and road boundary information. Simultaneously, oncoming traffic scenario markers and yielding scenario markers are identified based on environmental perception data. An escape suppression flag is generated based on these markers. Finally, the timing of obtaining reference lines, multiple anchor points, and scenario type information for the automated valet parking vehicle's location is controlled based on the escape trigger flag and the escape suppression flag. Specifically, when an escape trigger flag is generated but an escape suppression flag is not generated, the escape mode is triggered.

[0126] In one possible implementation, the escape mode is triggered when any of the following conditions are met: the vehicle remains stationary for more than 8 seconds and trajectory planning fails; the remaining trajectory distance is less than 550mm and the stationary time exceeds 5 seconds; or a collision occurs and the stationary time exceeds 3 seconds.

[0127] In one possible implementation, environmental perception data is used to determine whether the vehicle is in a scenario involving oncoming traffic, yielding, or pedestrians crossing. If such a scenario exists, a traction suppression flag is generated to prevent accidental triggering of the traction suppression mode during temporary blockages. For example, the traction suppression function is automatically disabled after the suppression lasts for more than 8 seconds until the vehicle begins to move.

[0128] S303. Obtain the reference line and scene type information of the scene where the automated valet parking vehicle is located.

[0129] S304. Construct a KD-tree index based on the reference line to determine the target point and target direction.

[0130] KD-tree is a data structure used for efficient querying of multidimensional spatial data. It recursively divides the space into hyperplanes and supports fast nearest neighbor search.

[0131] In this step, a KD-tree index is constructed based on multiple reference points on the reference line. Then, a matching reference point is queried from the KD-tree index based on the vehicle's current position. Subsequently, candidate target points are searched along the reference line starting from the matching reference point, with the direction of the reference line where the candidate target point is located taken as the target direction. The reference points contain information such as position coordinates, heading angle, curvature, and cumulative distance. Using a KD-tree structure accelerates nearest neighbor queries and supports quickly finding the matching position of the vehicle's current position or a candidate target point on the reference line.

[0132] In one possible implementation, when searching for candidate target points along a reference line starting from a matching reference point, the search can be performed by accumulating distances along the reference line from the vehicle's current position based on a preset target distance (e.g., 15 meters). After a candidate target point is found, the distribution of obstacles in a fan-shaped area in front of the candidate target point is detected. The passability status of the candidate target point is determined based on the obstacle distribution. If the passability status is not passable, the target point position is updated along the navigation reference line, and the reference point corresponding to the updated target point position is used as the target point. For example, if the detected obstacle distance is less than a threshold, the target point position is adjusted to ensure that the target point is reachable and safe. In some embodiments, when detecting the distribution of obstacles in a fan-shaped area in front of a candidate target point, historical trajectory data of the obstacles can be obtained first. The future position of the obstacles can be predicted based on the historical trajectory data. A safety boundary corresponding to the candidate target point can be generated based on the future position, and the target point position can be determined based on the safety boundary.

[0133] S305. Calculate the path complexity score based on the characteristics of the reference line between the vehicle's current position and the target point, and generate the main anchor point and auxiliary anchor point based on the path complexity score.

[0134] In this step, a reference line segment is extracted from the vehicle's current position to the target point. Then, the average curvature, maximum curvature, heading change, and path length of the reference line segment are calculated. Subsequently, a path complexity score is generated based on the average curvature, maximum curvature, heading change, and path length. Then, the number of anchor points, anchor point spacing, and pre-aiming distance are determined according to the path complexity score. Finally, main anchor points and auxiliary anchor points are generated on the reference line segment according to the determined number of anchor points, anchor point spacing, and pre-aiming distance.

[0135] In one possible implementation, the formula for calculating the path complexity score is as follows:

[0136]

[0137] in, Curvature complexity; For heading complexity; The length complexity is given by [reference to a specific function]. The curvature complexity, heading complexity, and length complexity are obtained by normalizing the maximum curvature, heading change, and path length, respectively.

[0138] The number of anchor points is determined based on the path complexity score. Generally, there are 3 anchor points for simple scenarios and a maximum of 5 anchor points for complex scenarios. The specific formula is as follows:

[0139]

[0140] The anchor point spacing is determined based on the path complexity score. Generally, the basic spacing is 3 meters, and the spacing can be increased appropriately for complex scenarios. The specific formula is as follows:

[0141]

[0142] The aiming distance is determined based on the path complexity score. Generally, the base aiming distance is 5 meters, and the specific formula is as follows:

[0143]

[0144] In one possible implementation, the process of generating anchor points is as follows: starting from the first anchor point spacing, accumulate distance along the reference line to find anchor point positions; search for the reference point with the smallest curvature near each candidate position, and prioritize the position with smaller curvature as the anchor point to improve path smoothness; repeat the above process until the upper limit of the number of anchor points is reached or the accumulated distance exceeds the target point distance; use the last anchor point as the main anchor point and the remaining anchor points as auxiliary anchor points.

[0145] In one possible implementation, the main guide position is determined on the reference line segment based on the pre-aiming distance, and the reference point corresponding to the main guide position is set as the main anchor point; multiple auxiliary anchor points are generated between the main anchor point and the current position of the vehicle according to the anchor point spacing, and multiple auxiliary anchor points are generated between the main anchor point and the target point according to the anchor point spacing.

[0146] In one possible implementation, the distribution of obstacles in the fan-shaped area in front of each main anchor point and each auxiliary anchor point is detected, the occupancy status of the corresponding anchor point is determined according to the obstacle distribution, the anchor point corresponding to the occupancy status is subjected to lateral offset processing, and the corresponding main anchor point or auxiliary anchor point is updated based on the offset anchor point.

[0147] S306. Calculate the anchor cost of the search point based on the main anchor point and the auxiliary anchor point.

[0148] In this step, the anchor distance cost and anchor angle cost of the search point are calculated based on the main anchor point and auxiliary anchor points in the multiple anchor points. Then, the anchor cost is generated based on the anchor distance cost and anchor angle cost, and the weight of the anchor cost is adjusted according to the scene type information.

[0149] In one possible implementation, anchor point cost Including the cost of the primary anchor point and the cost of the secondary anchor point, the formula is as follows:

[0150]

[0151] in, The distance from the search point to the main anchor point. For heading difference, For distance factor, Angle factor Cost of auxiliary anchor points.

[0152] The cost of auxiliary anchor points is calculated using a weighted summation, with the weight of each auxiliary anchor point decreasing according to its location, as shown in the following formula:

[0153]

[0154] in, To assist in the number of anchor points, For the first The weights of each auxiliary anchor point (decaying exponentially with a decay factor of 0.7). and They are respectively up to the number The distance and heading difference between the auxiliary anchor points.

[0155] In one possible implementation, the factors in the anchor point cost calculation process are adjusted based on the scene type information. If the scene type information includes a U-shaped bend scene identifier, the weights of the anchor point angle factor and distance factor are increased based on the U-shaped bend scene identifier; for example, the anchor point angle factor is increased by 1.0 and the distance factor by 0.2 for a U-shaped bend scene. If the scene type information includes a narrow channel scene identifier, the distance factor is adjusted based on the narrow channel scene identifier; for example, an additional coefficient for the anchor point distance factor is added for a narrow channel scene.

[0156] S307. Calculate the distance cost, heading angle cost, and progress incentive cost of the search point based on the reference line, and generate the reference line cost based on the above costs.

[0157] In one possible implementation, the reference line cost Taking into account the three dimensions of distance, angle, and progress, the formula is as follows:

[0158]

[0159] in, For the cost of distance, For the sake of perspective, Incentives for progress come at a cost.

[0160] Distance Cost The sigmoid function is used to smooth the transition, as shown in the following formula:

[0161]

[0162] in, To find the shortest distance from the search point to the reference line, This is the distance factor (default 0.1). For curvature factor, It is a distance factor (affected by distance).

[0163] The curvature factor is dynamically adjusted based on the curvature of the current reference point, as shown in the following formula:

[0164]

[0165] in, This is the minimum curvature factor (default 0.3). Curvature sensitivity (default 2.0). The curvature is the reference point. The greater the curvature, the smaller the factor, allowing the search point to deviate appropriately from the reference line at curves.

[0166] The angle cost is the difference between the heading of the search point and the heading of the nearest reference point. .in, Angle factor (default 0.1). This is due to the difference in heading.

[0167] The progress incentive cost guides the search toward the goal point, as shown in the following formula:

[0168]

[0169] in, The progress factor (default 0.05). r This represents the progress percentage (from 0 to 1) of the current point on the reference line. The closer the progress is to the finish line, the smaller the penalty, encouraging progress towards the goal.

[0170] S308. Adjust the weights of reference line cost and anchor point cost based on scene type information.

[0171] S309. Based on the adjusted reference line cost and anchor point cost, execute the path planning algorithm to generate an escape path for the automated valet parking vehicle.

[0172] This application provides an escape path planning method. It acquires vehicle status information and environmental perception data, and determines whether to enter an escape mode based on these data. If an escape mode is entered, it acquires the reference line and scene type information of the scene where the automated valet parking vehicle is located, and constructs a KD-tree index based on the reference line to determine the target point and target direction. Then, it calculates a path complexity score based on the characteristics of the reference line between the vehicle's current position and the target point, and generates primary and secondary anchor points based on the path complexity score. Subsequently, it calculates the anchor point cost of the search point based on the primary and secondary anchor points, and calculates the distance cost, heading angle cost, and progress incentive cost of the search point based on the reference line to generate the reference line cost. Next, it adjusts the weights of the reference line cost and anchor point cost based on the scene type information. Finally, it executes a path planning algorithm based on the adjusted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle. The above technical methods ensure rapid target point location in complex scenarios through KD-tree indexing; generate primary and secondary anchor points through path complexity scoring, which quantifies path difficulty through features such as curvature and heading changes, providing a dynamic basis for anchor point generation; ultimately, it improves the search efficiency of path planning algorithms and the quality of generated paths in complex and constrained environments.

[0173] Based on any of the above embodiments, the following, in conjunction with Figure 4 This paper provides a detailed explanation of a path planning method for escaping difficulties through specific examples.

[0174] In this embodiment, an autonomous vehicle performs a valet parking task at a U-shaped bend in a parking lot. During the turn, the vehicle's path is obstructed by a temporarily parked obstacle (such as another vehicle that is not turned off). The system detects a trajectory planning failure and triggers an escape mode. At this point, the vehicle needs to use a path planning algorithm to bypass the obstacle and resume driving along a reference line, while avoiding planning failure due to excessive curve curvature or high path complexity. This embodiment uses... Figure 1 The controller shown is the execution entity, and its specific process for implementing escape path planning is as follows:

[0175] S401, Escape Trigger Judgment and Scene Detection.

[0176] Specifically, in this step, the controller continuously monitors the vehicle's status (such as stationary time and the number of trajectory planning failures) and environmental information (such as obstacle distribution). When the vehicle's stationary time exceeds 8 seconds and trajectory planning fails consecutively, the escape mode is triggered. Simultaneously, the controller analyzes reference line characteristics (such as heading changes exceeding 120 degrees) to identify the current scenario as a U-turn and detects the presence of obstacles behind. It also performs a scenario suppression judgment to determine that there are no oncoming vehicles or yielding situations, thus allowing the escape function to execute normally.

[0177] S402, Dynamic target point generation.

[0178] Specifically, in this step, the controller constructs a KD-tree index based on the navigation reference line to accelerate the query of matching points on the reference line where the vehicle's current position is located. It searches for candidate target points within a 15-meter range along the reference line and detects whether there are obstacles in the fan-shaped area (e.g., a 30-degree field of view) in front of them. If an obstacle is detected at a distance less than a safety threshold (e.g., 0.5 meters), the target point's position is adjusted to the next feasible point on the reference line to ensure the target point is reachable and safe. For example, if the original candidate target point is unreachable due to an obstacle behind it, the controller adjusts it to a reference point on the outside of a curve to avoid the obstacle's influence.

[0179] S403, Multi-anchor point generation based on path complexity.

[0180] Specifically, in this step, the controller analyzes the characteristics of the reference line from the vehicle's current position to the target point (such as average curvature, heading change, and path length) and calculates the path complexity score. For U-shaped bend scenarios, the controller adaptively increases the number of anchor points (e.g., from the default 3 to 5) and shortens the anchor point spacing (e.g., from 3 meters to 2 meters) to more finely constrain the path direction. Subsequently, main anchor points (located at the bend exit) and auxiliary anchor points (located at the bend entrance and middle) are generated on the reference line according to the adaptive spacing, and obstacles in front of each anchor point are detected. If an auxiliary anchor point is unavailable due to obstruction by an obstacle, a lateral offset (e.g., 0.3 meters) is applied to avoid the obstacle.

[0181] S404, Multi-dimensional Heuristic Cost Function Design.

[0182] Specifically, in this step, the controller uses preset reference line cost functions and anchor point cost functions to calculate the reference line cost and anchor point cost of the search point. The reference line cost function comprehensively considers distance cost (the vertical distance from the vehicle to the reference line), heading angle cost (the difference between the vehicle's heading and the reference line's heading), and progress incentive cost (guiding the path towards the target point). For U-shaped bend scenarios, the controller dynamically adjusts the cost weights: increasing the anchor point angle factor (e.g., from 0.1 to 1.0) to avoid excessive turning within the bend; simultaneously decreasing the distance factor (e.g., from 0.1 to 0.2) to allow the vehicle to deviate appropriately from the reference line within the bend to avoid obstacles. Furthermore, the controller combines obstacle distance cost and gear shifting cost to further optimize path feasibility.

[0183] S405, Hybrid A* Search and Path Generation.

[0184] Specifically, in this step, the controller employs a hybrid A* search algorithm for path planning. The controller first initializes the hybrid A* algorithm parameters (e.g., coordinate resolution of 0.2 meters, heading resolution of 5 degrees) and combines this with adaptive anchor point constraints on the search direction. During the search, the controller prioritizes expanding nodes along the reference line direction (e.g., forward left turn, forward straight), and guides the path towards the main anchor point (curve exit) through anchor point costs. When a search node approaches the target point (e.g., the distance is less than 3 meters), the controller calls the Reeds-Shepp planner to directly connect to the target point, generating the optimal path that satisfies the vehicle's kinematic constraints. Finally, the path curvature is optimized using a QP smoother, and the last segment of the reverse path is trimmed to ensure the final segment of the path is in the forward direction, facilitating subsequent resumption of cruise mode.

[0185] It should be noted that, in Figure 4 The processing steps S401-S405 shown in the embodiments do not constitute a specific limitation on an escape path planning method. In other embodiments of this application, an escape path planning method may include more than Figure 4 The embodiments may include more or fewer steps; for example, an escape path planning method may include... Figure 4 Some steps in the embodiments, or, Figure 4 Some steps in the embodiments can be replaced by steps with the same function, or, Figure 4 Some steps in the embodiments can be broken down into multiple steps, etc.

[0186] Figure 5 This application provides a schematic diagram of the structure of an escape path planning device, as shown below. Figure 5 As shown, the escape path planning device 50 provided in this embodiment includes:

[0187] The acquisition module 501 is used to acquire reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located.

[0188] The calculation module 502 is used to calculate the reference line cost of the search point based on the reference line, and to calculate the anchor point cost of the search point based on multiple anchor points.

[0189] The weight adjustment module 503 is used to adjust the weights of the reference line cost and the anchor point cost based on the scene type information.

[0190] The path planning module 504 is used to execute a path planning algorithm based on the adjusted reference line cost and anchor point cost to generate an escape path for the automated valet parking vehicle.

[0191] In one possible implementation, the computing module 502 is specifically used for:

[0192] Determine the matching reference point on the reference line for the search point.

[0193] Calculate the distance cost between the search point and the matching reference point.

[0194] Calculate the heading angle cost between the heading of the search point and the heading of the reference line of the matching reference point.

[0195] Calculate the progress incentive cost of the search point along the reference line toward the target direction.

[0196] The reference line cost is generated based on the distance cost, heading angle cost, and progress incentive cost.

[0197] In one possible implementation, the acquisition module 501 is specifically used for:

[0198] Extract the reference line segment from the vehicle's current position to the target point.

[0199] Calculate the average curvature, maximum curvature, heading change, and path length of the reference line segment.

[0200] A path complexity score is generated based on the average curvature, maximum curvature, heading change, and path length.

[0201] The number of anchor points, anchor point spacing, and aiming distance are determined based on the path complexity score.

[0202] Generate primary and secondary anchor points on the reference line segment according to the determined number of anchor points, anchor point spacing, and pre-aiming distance; multiple anchor points include primary and secondary anchor points.

[0203] In one possible implementation, the computing module 502 is specifically used for:

[0204] The anchor distance cost and anchor angle cost of the search point are calculated based on the main anchor point and auxiliary anchor points.

[0205] Anchor point costs are generated based on anchor point distance costs and anchor point angle costs.

[0206] In one possible implementation, the acquisition module 501 is further specifically used for:

[0207] Acquire vehicle status information and environmental perception data.

[0208] The duration of stationary position and trajectory planning status are determined based on vehicle status information.

[0209] The distribution of obstacles and road boundary information are determined based on environmental perception data.

[0210] An escape trigger flag is generated based on the static duration, trajectory planning status, obstacle distribution, and road boundary information.

[0211] In one possible implementation, the acquisition module 501 is further specifically used for:

[0212] Identify vehicle meeting scene markers and yielding scene markers based on environmental perception data.

[0213] Generate a trouble-relief flag based on the vehicle meeting scenario flag and the yielding scenario flag.

[0214] The timing of execution is controlled based on the timing of obtaining reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, according to the traction trigger flag and traction inhibition flag.

[0215] In one possible implementation, the computing module 502 is further specifically used for:

[0216] A KD-tree index is constructed based on multiple reference points on the reference line.

[0217] Query the KD-tree index to find a matching reference point based on the vehicle's current location.

[0218] Search for candidate target points along the reference line, starting from the matching reference point.

[0219] Use the direction of the reference line where the candidate target point is located as the target direction.

[0220] In one possible implementation, the path planning module 504 is specifically used for:

[0221] Initialize the set of search nodes and kinematic extension parameters in the path planning algorithm.

[0222] The comprehensive cost of candidate search nodes is calculated based on obstacle distance cost, weighted reference line cost, and anchor point cost.

[0223] Expand candidate search nodes according to the overall cost.

[0224] When the distance between the candidate search node and the target point meets the preset connection conditions, a candidate parsing path from the candidate search node to the target point is generated by a preset vehicle geometry path generator.

[0225] Generate an escape path based on candidate parsing paths.

[0226] The escape path planning device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0227] Figure 6 This is a structural schematic diagram of an escape path planning device provided in this application. Figure 6As shown, the escape route planning device 60 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the escape route planning device 60 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0228] In the specific implementation process, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to execute the above-mentioned escape path planning method.

[0229] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0230] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0231] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0232] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0233] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described obstacle avoidance path planning method.

[0234] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned obstacle avoidance path planning method.

[0235] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0236] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0237] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0238] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0239] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0240] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0241] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0242] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for planning an escape route, characterized in that, include: Obtain reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located; The reference line cost of the search point is calculated based on the reference line, and the anchor point cost of the search point is calculated based on the multiple anchor points. The weights of the reference line cost and the anchor point cost are adjusted according to the scenario type information. Based on the adjusted reference line cost and anchor point cost, a path planning algorithm is executed to generate an escape path for the automated valet parking vehicle.

2. The method according to claim 1, characterized in that, The calculation of the reference line cost for the search point based on the reference line includes: Determine the matching reference point of the search point on the reference line; Calculate the distance cost between the search point and the matching reference point; Calculate the heading angle cost between the heading of the search point and the reference line heading of the matching reference point; Calculate the progress incentive cost of the search point along the reference line toward the target direction; The reference line cost is generated based on the distance cost, the heading angle cost, and the progress incentive cost.

3. The method according to claim 2, characterized in that, Before obtaining the multiple anchor points of the scene where the automated valet parking vehicle is located, the method further includes: Extract the reference line segment from the vehicle's current position to the target point; Calculate the average curvature, maximum curvature, heading change, and path length of the reference line segment; A path complexity score is generated based on the average curvature, the maximum curvature, the heading change, and the path length. The number of anchor points, the spacing between anchor points, and the aiming distance are determined based on the path complexity score. The primary anchor point and the secondary anchor point are generated on the reference line segment according to the determined number of anchor points, the anchor point spacing, and the pre-aiming distance; the multiple anchor points include the primary anchor point and the secondary anchor point.

4. The method according to claim 3, characterized in that, The calculation of the anchor cost of the search point based on the multiple anchor points includes: Calculate the anchor point distance cost and anchor point angle cost of the search point based on the main anchor point and the auxiliary anchor point; The anchor point cost is generated based on the anchor point distance cost and the anchor point angle cost.

5. The method according to claim 1, characterized in that, Before obtaining the reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, the process also includes: Acquire vehicle status information and environmental perception data; The stationary duration and trajectory planning status are determined based on the vehicle status information. Based on the environmental perception data, determine the distribution of obstacles and road boundary information; An escape trigger flag is generated based on the static duration, the trajectory planning status, the obstacle distribution, and the road boundary information.

6. The method according to claim 5, characterized in that, Before obtaining the reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located, the method further includes: Based on the environmental perception data, identify vehicle meeting scenario markers and yielding scenario markers; Generate an escape inhibition flag based on the meeting scene identifier and the yielding scene identifier; The timing of obtaining the reference line, multiple anchor points, and scene type information of the scene where the automatic valet parking vehicle is located is controlled according to the escape trigger flag and the escape inhibition flag.

7. The method according to claim 2, characterized in that, Before calculating the reference line cost of the search point based on the reference line, the method further includes: A KD-tree index is constructed based on multiple reference points on the reference line; Based on the vehicle's current location, a matching reference point is queried in the KD-tree index; Search for candidate target points along the reference line, starting from the matching reference point; The direction of the reference line where the candidate target point is located is taken as the target direction.

8. The method according to claim 3, characterized in that, The path planning algorithm, based on the weighted reference line cost and the anchor point cost, generates an escape path for the automated valet parking vehicle, including: Initialize the search node set and kinematic extension parameters in the path planning algorithm; The comprehensive cost of the candidate search node is calculated based on the obstacle distance cost, the reference line cost after weight adjustment, and the anchor point cost. Expand candidate search nodes according to the comprehensive cost; When the distance between the candidate search node and the target point meets the preset connection conditions, a candidate parsing path from the candidate search node to the target point is generated by a preset vehicle geometry path generator. The escape path is generated based on the candidate parsing paths.

9. A path planning device for escaping difficulties, characterized in that, include: The acquisition module is used to acquire reference lines, multiple anchor points, and scene type information of the scene where the automated valet parking vehicle is located; The calculation module is used to calculate the reference line cost of the search point based on the reference line, and to calculate the anchor point cost of the search point based on the multiple anchor points; The weight adjustment module is used to adjust the weights of the reference line cost and the anchor point cost according to the scene type information. The path planning module is used to execute a path planning algorithm based on the reference line cost and the anchor point cost after adjustment of weights, and generate an escape path for the automated valet parking vehicle.

10. A path planning device for escaping difficulties, characterized in that, include: A memory, and a processor communicatively connected to the memory; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8.

Images