A valet parking trajectory planning method and device, electronic equipment and storage medium

Abstract

The application provides a method and device for planning a parking trajectory, electronic equipment and a storage medium. The method comprises obtaining driving information and external information of a vehicle to be parked, performing global path planning on an initial starting point information and a target parking position based on an electronic map, initial pose information and obstacle information to obtain a reference route, performing local path planning based on the reference route and current pose information to obtain an initial local route along the reference route, performing first coordinate conversion on the initial local route based on the current pose information to obtain a cost relationship, performing conditional constraint on the cost relationship to obtain an intermediate local route, the conditional constraint comprising calculating a discrete boundary of a drivable area in the initial local route based on the obstacle information, performing boundary constraint on the cost relationship based on the discrete boundary, and performing second coordinate conversion on the intermediate local route based on the reference route to obtain a target local route. The cost relationship and the conditional constraint increase the universality and safety of the valet parking.

Description

Technical Field

[0001] This application relates to the field of navigation technology, specifically to a valet parking trajectory planning method, device, electronic device, and storage medium. Background Technology

[0002] Valet parking refers to a vehicle automatically driving from the drop-off area into the parking lot according to parking instructions and finding an empty parking space to park; when a recall instruction is sent, the vehicle can automatically return to the passenger pick-up area according to the recall instruction, and the whole process is done without human intervention.

[0003] Currently, the most widely used systems are automatic parking assist systems, mostly based on ultrasonic sensors or cameras. These systems work by parking the vehicle near a parking space, activating the automatic parking assist function, and then using these sensors to detect the surrounding environment, plan a parking trajectory, and finally drive the vehicle into the parking space. Due to sensor limitations, these automatic parking assist systems typically require the driver to manually park the vehicle near the parking space and activate the automatic parking assist function; they cannot truly achieve autonomous valet parking.

[0004] To address the shortcomings of automatic parking assistance systems, autonomous valet parking systems add an autonomous navigation function from the parking lot entrance to the vicinity of the parking space, achieving global path planning. This autonomous navigation function typically uses sensors such as LiDAR and cameras on the vehicle to perceive environmental information and its own position. The system then uses a server to assign parking spaces, enabling the vehicle to autonomously move along a globally planned trajectory and park itself. However, LiDAR is expensive, cameras are highly dependent on ambient light, and the use of sensors and the assignment of parking spaces limit the types of parking lots that can be used, resulting in limited parking routes and an inability to provide localized paths based on the globally planned trajectory.

[0005] In related technologies, by providing local path planning for multiple different parking scenarios such as vehicle entry and exit from parking lots and intersection turning scenarios, the problems of single parking routes and inability to provide local planning paths are solved. However, this method requires maintaining multiple sets of local planning algorithms for different scenarios, which makes local path planning complex and requires trajectory splicing when switching parking scenarios, resulting in high maintenance costs and a high susceptibility to errors. Summary of the Invention

[0006] This application provides a valet parking trajectory planning method, device, electronic device, and storage medium to solve the technical problems of limited applicable parking scenario types and complex local path planning for valet parking.

[0007] This application provides a valet parking trajectory planning method, comprising: acquiring driving information and external information of a vehicle waiting to be parked, wherein the driving information includes at least initial pose information and current pose information, and the external information includes at least initial starting point information, target parking position, and obstacle information; performing global path planning on the path between the initial starting point information and the target parking position based on an electronic map, the initial pose information, and the obstacle information to obtain a reference route, and performing local path planning based on the reference route and the current pose information to obtain an initial local route along the reference route; performing a first coordinate transformation on the initial local route based on the current pose information to obtain a cost relationship, and applying conditional constraints to the cost relationship to obtain an intermediate local route, wherein the conditional constraints include calculating discrete boundaries of drivable areas in the initial local route based on the obstacle information, and applying boundary constraints to the cost relationship based on the discrete boundaries; and performing a second coordinate transformation on the intermediate local route based on the reference route to obtain a target local route, so as to perform valet parking according to the target local route.

[0008] In one embodiment of this application, calculating the discrete boundary of the drivable area in the initial local route based on the obstacle information, and constraining the cost relationship based on the discrete boundary includes: determining the guide point pose information of multiple path guide points according to the initial local route; establishing an initial obstacle boundary corresponding to each path guide point based on the pose information of each guide point, wherein the initial size of the initial obstacle boundary is the same as the rectangular outline size of the vehicle waiting to park; expanding the four sides of each initial obstacle boundary outward in a preset order and a preset fixed compensation interval until all four sides are target boundaries to obtain a drivable area, wherein the target boundary is determined as the target boundary if any outwardly expanded edge collides with an obstacle in the obstacle information, or if the expanded edge has been expanded to a preset maximum expansion distance; performing a third coordinate transformation on each target boundary of the drivable area based on the width of the vehicle waiting to park to obtain a discrete boundary; and constraining the cost relationship based on the discrete boundary.

[0009] In one embodiment of this application, performing a first coordinate transformation on the initial local route based on the current pose information to obtain a cost relationship includes: uniformly sampling the initial local route to obtain multiple reference path points; obtaining lateral offset data and channel lateral offset data based on the offset state of the current pose information relative to each of the reference path points; and determining a cost relationship based on the lateral offset data, the channel lateral offset data, and the previous frame trajectory point data in the initial local route.

[0010] In one embodiment of this application, determining the cost relationship based on the lateral offset data, the channel lateral offset data, and the previous frame trajectory point data in the initial local path includes: determining a trajectory smoothing term based on the square of each lateral offset, the square of each first-order lateral partial derivative, the square of each second-order lateral partial derivative, and the square of each third-order lateral partial derivative, wherein each lateral offset, the first-order lateral partial derivative, the second-order lateral partial derivative, and the third-order lateral partial derivative are obtained based on the lateral offset data; and determining a proximity to the centerline term based on the squared difference between each lateral offset and each channel lateral offset, wherein each channel lateral offset is used to characterize the narrow channel centerline relative to each of the reference path points. Lateral offset; similarity terms of the trajectory in the previous frame are determined based on the squared difference between each lateral offset and the lateral offset of the previous frame; end pose proximity terms are determined based on the squared difference between the pre-terminal lateral offset and the end-terminal lateral offset, the squared difference between the first-order pre-terminal lateral derivative and the first-order end-terminal lateral derivative, and the squared difference between the second-order pre-terminal lateral derivative and the second-order end-terminal lateral derivative, wherein each previous frame lateral offset, the pre-terminal lateral offset, the end-terminal lateral offset, the first-order pre-terminal lateral derivative, the first-order end-terminal lateral derivative, the second-order pre-terminal lateral derivative, and the second-order end-terminal lateral derivative are obtained based on the trajectory point data of the previous frame; cost relationships are determined based on the trajectory smoothing term, the proximity to the centerline term, the similarity term, and the end pose proximity term.

[0011] In one embodiment of this application, the condition constraints further include: applying path continuity constraints to the cost relationship based on the multi-order derivative expansion of the lateral offset data; applying curvature constraints to the cost relationship based on local curvature data, the lateral offset data, and the driving information; and applying at least one of the starting point constraints to the cost relationship based on the driving information, the lateral offset data, and the trajectory point data of the previous frame; wherein the local curvature data is obtained by sampling the initial local route.

[0012] In one embodiment of this application, constraining the cost relationship based on the driving information, the lateral offset data, and the trajectory point data of the previous frame includes: if the current planning scene state is a replanning scene, then determining the starting point constraint based on the lateral offset data and the starting point lateral offset; if the current planning scene state is not a replanning scene, then determining the starting point constraint based on the lateral offset data and the endpoint lateral offset in the trajectory point data of the previous frame; constraining the cost relationship based on the starting point constraint; wherein, the driving information also includes the current planning scene state, the starting point lateral offset is obtained based on the lateral offset data corresponding to the current pose information, and the replanning scene is used to characterize the first frame scene and the obstacle avoidance scene.

[0013] In one embodiment of this application, curvature constraints are applied to the cost relationship based on local curvature data, the lateral offset data, and the driving information, including: determining each first curvature parameter based on the vehicle's maximum steering angle, each lateral offset data, and each reference curvature; the driving information further includes the vehicle's wheelbase and the vehicle's maximum steering angle; the local curvature data includes the reference curvature corresponding to each reference path point; determining a second curvature parameter based on the vehicle's maximum steering angle; determining each third curvature parameter based on the vehicle's wheelbase and each reference curvature; determining curvature constraints based on the second curvature parameter, each first curvature parameter, and each third curvature parameter; and applying curvature constraints to the cost relationship.

[0014] In one embodiment of this application, constraining the cost relationship to obtain an intermediate local route includes: converting boundary constraints, path continuity constraints, starting point constraints, curvature constraints, and the cost relationship into matrix form based on a preset quadratic programming general form to obtain the cost plan to be determined; solving the cost plan to be determined by minimizing the cost to obtain the intermediate local route; the condition constraints also include the path continuity constraints, the starting point constraints, and the curvature constraints.

[0015] In one embodiment of this application, global path planning is performed on the path between the initial starting point information and the target parking position based on an electronic map, the initial pose information, and the obstacle information to obtain a reference route. This includes: generating a lane candidate list based on the initial pose information, the initial starting point information, the target parking position, and the lane-parking space association; performing an optimal path search on a search map based on the lane candidate list and the initial pose information to obtain an initial global path; and smoothing the initial global path based on the lane reference line and the obstacle information to obtain the reference route. The search map, the lane-parking space association, and the lane reference line are obtained based on the electronic map. The search map is used to represent the length connection relationship between roads, road segments, and lanes in the parking lot, and the lane-parking space association is used to represent the association relationship between lanes, parking spaces, and coordinate points in the parking lot.

[0016] In one embodiment of this application, generating a lane candidate list based on the initial pose information, the initial starting point information, the target parking position, and the lane-parking space association includes: if the initial starting point information is a starting point coordinate point, then generating a virtual node based on the angle between the initial heading angle in the initial pose information and the heading angle of the discrete point of the lane reference line corresponding to the starting point coordinate point, and connecting the virtual node to a first starting lane to obtain a first candidate list, wherein the discrete point of the lane reference line is obtained based on the lane-parking space association, and the first starting lane is used to represent all drivable lanes matched from the lane-parking space association based on the starting point coordinate point and the preset distance; if the initial starting point information is a starting parking space, then connecting the starting parking space to the lane-parking space association... The lane-space association is matched to obtain a second starting lane, and the second starting lane that meets the preset distance is determined as a second candidate list; if the target parking location is the destination coordinate point, then the first destination lane corresponding to the destination coordinate point in the lane-space association is determined as a third candidate list based on the preset distance; if the target parking location is the destination parking space, then the destination parking space is matched with the lane-space association to obtain a second destination lane, and the second destination lane that meets the preset distance is determined as a fourth candidate list; the starting candidate list and the destination candidate list are used as lane candidate lists, the starting candidate list includes the first candidate list or the second candidate list, and the destination candidate list includes the third candidate list or the fourth candidate list.

[0017] In one embodiment of this application, smoothing the initial global path based on the lane reference line and the obstacle information to obtain a reference route includes: assessing the collision risk of the initial global path based on the left and right boundary values ​​of the obstacles to obtain a collision risk state, wherein the left and right boundary values ​​of the obstacles are obtained based on the lane reference line and the obstacle information; if the collision risk state indicates a collision risk exists, an optimization objective is generated based on path length information, obstacle distance information, and curvature information; a penalty term is generated based on the obstacle information and added to the optimization objective; and the optimization objective is minimized based on the lane reference line to obtain the reference route, wherein the path length information, obstacle distance information, and initial curvature information are obtained based on the initial global path.

[0018] This application provides a valet parking trajectory planning device, comprising: an information acquisition module for acquiring driving information and external information of a vehicle waiting to be parked, wherein the driving information includes at least initial pose information and current pose information, and the external information includes at least initial starting point information, target parking location, and obstacle information; a path planning module for performing global path planning on the path between the initial starting point information and the target parking location based on an electronic map, the initial pose information, and the obstacle information to obtain a reference route, and performing local path planning based on the reference route and the current pose information to obtain an initial local route along the reference route; a constraint solving module for performing a first coordinate transformation on the initial local route based on the current pose information to obtain a cost relationship, and applying conditional constraints to the cost relationship to obtain an intermediate local route, wherein the conditional constraints include calculating discrete boundaries of drivable areas in the initial local route based on the obstacle information, and applying boundary constraints to the cost relationship based on the discrete boundaries; and a coordinate transformation module for performing a second coordinate transformation on the intermediate local route based on the reference route to obtain a target local route, so as to perform valet parking according to the target local route.

[0019] This application also provides an electronic device, the electronic device comprising: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the valet parking trajectory planning method as described in any of the above embodiments.

[0020] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer's processor, causes the computer to perform the valet parking trajectory planning method as described in any of the above embodiments.

[0021] The beneficial effects of the embodiments of the present invention are as follows: The present invention provides a valet parking trajectory planning method, device, electronic device, and storage medium. In the embodiments of the present invention, a local path planning is performed on the reference route obtained by global path planning based on the current pose information to obtain an initial local route along the reference route. A cost relationship is obtained by performing a first coordinate transformation on the initial local route according to the current pose information. Conditional constraints are applied to the cost relationship to obtain an intermediate local route. By using conditional constraints and cost relationships, the optimization of local routes after local path planning corresponding to different parking scenarios is uniformly expressed, simplifying local route optimization. Furthermore, it eliminates the need to distinguish between parking scenarios, providing a local route optimization that is universal for all parking scenarios, thus increasing the versatility of valet parking. The embodiments of the present invention optimize the initial local route by using discrete boundaries of the drivable area for boundary constraints, reducing the collision probability of waiting vehicles and improving the safety of waiting parking.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0024] Figure 1 A schematic diagram of an exemplary system architecture to which the technical solutions of the embodiments of this application can be applied is shown;

[0025] Figure 2 A flowchart illustrating a valet parking trajectory planning method according to an embodiment of this application is shown;

[0026] Figure 3 A schematic diagram of lane matching for global path planning according to an embodiment of this application is shown;

[0027] Figure 4 An architecture diagram of global path planning according to an embodiment of this application is shown;

[0028] Figure 5 A schematic diagram of the trajectory for a first coordinate transformation of an initial local route according to an embodiment of this application is shown;

[0029] Figure 6 A flowchart illustrating the solution of the upper and lower boundaries of a target according to an embodiment of this application is shown;

[0030] Figure 7 A schematic diagram of convex hull solving in Cartesian coordinates according to an embodiment of this application is shown;

[0031] Figure 8 A block diagram of a valet parking trajectory planning device according to an embodiment of this application is shown;

[0032] Figure 9 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown. Detailed Implementation

[0033] The embodiments of this application will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be understood that the preferred embodiments are only for illustrating this application and are not intended to limit the scope of protection of this application.

[0034] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0035] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0036] First, it's important to clarify that a complete valet parking system's path planning should include both global and local path planning. Global path planning involves automatically planning a route from outside the parking lot to available parking spaces after acquiring a pre-set high-precision map of the parking area, thus expanding the valet parking area and application scenarios. Local path planning, on the other hand, uses environmental information provided by sensors and, based on the global path, dynamically and in real-time plans a safe and smooth local path to ensure the safety and stability of vehicles waiting to be parked.

[0037] Please see Figure 1 , Figure 1 A schematic diagram of an exemplary system architecture to which the technical solutions of the embodiments of this application can be applied is shown. For example... Figure 1 As shown, the system architecture may include a server 101 and a vehicle waiting to park 102. The server 101 is used to provide external information to the vehicle waiting to park 102.

[0038] For example, after the vehicle 102 waiting to park obtains its driving information and the external information provided by the server 101, the driving information includes at least initial pose information and current pose information, and the external information includes at least initial starting point information, target parking position, and obstacle information. Based on the electronic map, initial pose information, and obstacle information, global path planning is performed on the path between the initial starting point information and the target parking position to obtain a reference route. Based on the reference route and current pose information, local path planning is performed to obtain an initial local route along the reference route. Based on the current pose information, a first coordinate transformation is performed on the initial local route to obtain a cost relationship. Conditional constraints are applied to the cost relationship to obtain an intermediate local route. The conditional constraints include calculating the discrete boundary of the drivable area in the initial local route based on obstacle information and applying boundary constraints to the cost relationship based on the discrete boundary. Based on the reference route, a second coordinate transformation is performed on the intermediate local route to obtain a target local route, so as to perform valet parking according to the target local route.

[0039] In related technologies, local path planning is provided for multiple scenarios such as vehicle entry and exit from parking lots and intersection turning scenarios. However, multiple local planning algorithms for different scenarios need to be maintained, and trajectory splicing is required when switching scenarios, which results in high maintenance costs and is prone to errors.

[0040] To address the aforementioned technical problems, this application provides a valet parking trajectory planning method, apparatus, electronic device, and storage medium.

[0041] Please see Figure 2 , Figure 2 A flowchart illustrating a valet parking trajectory planning method according to an embodiment of this application is shown. Figure 2 As shown, in an exemplary embodiment, the valet parking trajectory planning method includes at least steps S210 to S240, which are described in detail below:

[0042] Step S210: Obtain the driving information and external information of the vehicle waiting to be parked.

[0043] The driving information includes at least initial pose information and current pose information, and the external information includes at least initial starting point information, target parking position and obstacle information.

[0044] In one embodiment of this application, the initial pose information includes initial position information and initial attitude information. The initial attitude information includes the initial heading angle corresponding to the driving direction of the vehicle waiting to be parked. The current pose information includes current position information and current attitude information.

[0045] In one embodiment of this application, the obstacle information includes lane line associated boundary information and static obstacles on the left and right sides of the lane line. The lane reference line is represented in the form of a discrete point array [P1, P2, ... P...].n Each point P contains a floor, coordinates, and associated lane. Static obstacles include walls, curbs, and pillars, and are represented as a set of bounding boxes or a set of line segments.

[0046] Step S220: Based on the electronic map, initial pose information and obstacle information, perform global path planning on the path between the initial starting point information and the target parking position to obtain a reference route, and perform local path planning based on the reference route and the current pose information to obtain an initial local route along the reference route.

[0047] In one embodiment of this application, global path planning is performed on the path between the initial starting point information and the target parking position based on an electronic map, initial pose information, and obstacle information to obtain a reference route. This includes: generating a lane candidate list based on the initial pose information, initial starting point information, target parking position, and lane-parking space association; performing optimal path search on the search map based on the lane candidate list and initial pose information to obtain an initial global path; and smoothing the initial global path based on lane reference lines and obstacle information to obtain the reference route. The search map, lane-parking space association, and lane reference lines are obtained based on an electronic map. The search map is used to represent the length connection relationship between roads, road segments, and lanes in the parking lot, and the lane-parking space association is used to represent the association relationship between lanes, parking spaces, and coordinate points in the parking lot.

[0048] In one embodiment of this application, the electronic map is a high-precision map with an accuracy at the centimeter level, such as 10 to 20 centimeters. This is merely an example, and this application does not impose any limitations on the accuracy of the high-precision map.

[0049] In one embodiment of this application, the search map is obtained by extracting roads (Links) and road segments (LinkSections) from the electronic map corresponding to the parking lot. Road segments are used as the smallest unit of the search algorithm, and road endpoints are used as search nodes to establish road topology relationships. Left and right lanes are then associated with corresponding roads or road segments based on the relationships provided by the electronic map. All road segments are treated as topology nodes, and a weighted directed graph is formed based on the adjacency and sequential connection relationships between them. The weight of each topology node corresponding to a road segment is the path length weight value, resulting in the search map. The search map includes the lengths and connection relationships of all roads within the parking lot area, as well as the correspondence between each road segment and lane.

[0050] In one embodiment of this application, key map feature information within the parking lot area is represented by polyline features or polygon features. Key map feature information includes parking spaces, intersections, and road boundary lines. Furthermore, using floor levels, coordinate points, or geometric shape information provided by the electronic map, geometric calculations are performed to match and correspond with lanes, obtaining the lane-parking space relationship. This reduces the complexity of lane matching when the starting point (or parking space) and ending point (or parking space) of global path planning are near intersections, reduces computational load, and improves the rationality of the matching results.

[0051] In one embodiment of this application, since the start and end points of global path planning can be specific coordinate points or parking spaces, and the lane connections in the area near the intersection are relatively complex, it is necessary to achieve lane matching at the start and end points (or parking spaces) while considering the initial pose information of the vehicles waiting to park and the rationality of the path connections, so as to avoid the problem that the current driving direction of the vehicle is inconsistent with the planned lane direction or the lane is discontinuous.

[0052] In one embodiment of this application, generating a lane candidate list based on initial pose information, initial starting point information, target parking position, and lane-parking space association includes: if the initial starting point information is a starting point coordinate point, then a virtual node is generated based on the angle between the initial heading angle in the initial pose information and the heading angle of the discrete point of the lane reference line corresponding to the starting point coordinate point, and the virtual node is connected to a first starting lane to obtain a first candidate list. The discrete point of the lane reference line is obtained based on the lane-parking space association, and the first starting lane is used to represent all drivable lanes matched from the lane-parking space association based on the starting point coordinate point and a preset distance; if the initial starting point information is a starting parking space, then the starting parking space is associated with a lane parking space. The relationship is matched to obtain the second starting lane, and the second starting lane that meets the preset distance is determined as the second candidate list; if the target parking location is the end point coordinate point, the first end point lane corresponding to the end point coordinate point in the lane parking space association relationship is used as the third candidate list based on the preset distance; if the target parking location is the end point parking space, the end point parking space and the lane parking space association relationship are matched to obtain the second end point lane, and the second end point lane that meets the preset distance is used as the fourth candidate list; the starting candidate list and the end point candidate list are used as the lane candidate list, the starting candidate list includes the first candidate list or the second candidate list, and the end point candidate list includes the third candidate list or the fourth candidate list.

[0053] In one embodiment of this application, please refer to Figure 3 , Figure 3 A schematic diagram illustrating lane matching for global path planning according to one embodiment of this application is shown. Figure 3As shown, the start and end point information includes initial start point information and target parking location. If the type of start and end point information is a start and end parking space, the first start and end lane is obtained by matching the start and end parking spaces with the lane parking space association. If the distance between the first start and end lane and the start and end parking space meets the preset distance, the first start and end lane is used as the first candidate matching lane. If the distance between the first start and end lane and the start and end parking space does not meet the preset distance, the unreasonable first start and end lane is discarded. If the type of start and end point information is not a start and end parking space, the start and end coordinates and the lane parking space association are matched based on the preset distance to obtain the second start and end lane. If the heading angle between the second start and end lane and the initial vehicle heading angle meets the preset angle, the second start and end lane is determined as the second candidate matching lane. If the heading angle between the second start and end lane and the initial vehicle heading angle does not meet the preset angle, the unreasonable second start and end lane is discarded.

[0054] In one embodiment of this application, the first start-end lane includes a second start lane and a second end lane; the second start-end lane includes a first start lane and a first end lane. The first alternative matching lane includes a second alternative list and a fourth alternative list, and the second alternative matching lane includes a first alternative list and a third alternative list. The start-end coordinate points include a start coordinate point and an end coordinate point, and the start-end parking spaces include a start parking space and an end parking space.

[0055] In one embodiment of this application, for the start and end points, the information is first classified according to its type: If the start point information is a start point coordinate point, a virtual node is generated by considering the angle between the initial heading angle provided by vehicle positioning and the heading angle of the discrete point of the lane reference line corresponding to the start point coordinate point. This virtual node is then connected to all drivable lanes near the start point coordinate point to obtain a first candidate list of drivable start point coordinate points. All drivable lanes are obtained by matching the start point coordinate point and lane parking space association relationship within a preset distance. If the start point information is a start point parking space, i.e., a parking exit scenario, the start point parking space is matched to the corresponding second start point lane according to the lane parking space association relationship. The second start point lane includes lanes on the same side and lanes on the opposite side, achieving the purpose of parking exiting from both sides. If the distance between the second start point lane and the start point parking space meets the preset distance, the second start point lane is determined as the second candidate list. If the end point information is an end point coordinate point, all lanes within the preset distance matched by the end point coordinate point are directly used as candidates to obtain a third candidate list. If the termination information is the destination parking space, i.e. the parking scenario, in order to ensure the success rate of planning, the destination parking space is matched to the corresponding second destination lane as a candidate based on the lane-parking space association relationship. The second destination lane includes the lane on the same side and the opposite lane. If the distance between the second destination lane and the destination parking space meets the preset distance, the second destination lane is determined as the fourth candidate list.

[0056] In one embodiment of this application, a virtual road segment is generated based on the distance between the initial position information in the initial pose information and the start and end lane points in the lane candidate list, and the search map is updated based on the virtual road segment; the shortest path in the updated search map that matches the initial heading angle is determined as the initial global path.

[0057] In one embodiment of this application, for the lane candidate list, virtual road segments are formed by calculating the distances from the initial position information in the initial pose information to the corresponding start and end lane points, and then added to the search map. To achieve smooth and reasonable path connections, all road segments in the new search map will be accompanied by intersection attribute information to avoid the problem of discontinuity between adjacent road segments belonging to the same intersection in the search results. When the next road segment found is within the same intersection range as the current road, it indicates that the current path connection scheme is not feasible, and the next intersection road segment is skipped for selection.

[0058] In one embodiment of this application, the matching lane in the updated search map whose lane point heading angle matches the initial heading angle is determined as the path to be searched, and the shortest path in the path to be searched is determined as the initial global path, so as to obtain a continuous path with high traffic efficiency and short distance at both the start and end points.

[0059] In one embodiment of this application, smoothing the initial global path based on lane reference lines and obstacle information to obtain a reference route includes: assessing the collision risk of the initial global path based on the left and right boundary values ​​of obstacles to obtain a collision risk state, wherein the left and right boundary values ​​of obstacles are obtained based on lane reference lines and obstacle information; if the collision risk state indicates a collision risk exists, an optimization objective is generated based on path length information, obstacle distance information, and curvature information; a penalty term is generated based on the left and right boundary values ​​of obstacles and added to the optimization objective; and the optimization objective is minimized based on lane reference lines to obtain the reference route, wherein the path length information, obstacle distance information, and initial curvature information are obtained based on the initial global path.

[0060] In one embodiment of this application, left and right boundary values ​​are obtained based on lane line associated boundary information and static obstacles. The left and right boundary values ​​include at least one of the obstacle left boundary value (LeftBoundary) and obstacle right boundary value (RightBoundary).

[0061] In one embodiment of this application, a smoothing method based on optimization control is employed to smooth the lane reference lines in the initial local route, transforming the smoothing problem into a non-convex optimization problem. Optimization objectives are generated based on path length information (Cost1), obstacle distance information (Cost2), and curvature information (Cost3). The bounding boxes around the initial global path are transformed into collision constraints, and the lane reference lines of the initial global path serve as a smoothing warmstart. To achieve stable solutions, collision constraints are not given as hard constraints but are added to the optimization objective as penalty terms. After obtaining the smoothed path, the smoothed path is stored as the path reference lines for each road segment, and the overall output is the global planned path, i.e., the reference route.

[0062] In one embodiment of this application, please refer to Figure 4 , Figure 4 An architecture diagram of global path planning according to one embodiment of this application is shown. Figure 4 As shown, the road topology is obtained through key map element information of the high-precision map. The road topology includes the search map and the relationship between lanes and parking spaces. Lane matching is performed through the relationship between lanes and parking spaces to obtain the start and end lane combination. The start and end lane combination is the lane candidate list. A global path search is performed based on the start and end lane combination and the search map to obtain the initial global path. For narrow road scenarios, global path optimization is performed based on lane reference lines and obstacle information. Global path optimization is also known as smoothing the initial global path to obtain the global planned path, which is the reference route.

[0063] In one embodiment of this application, local path planning is performed based on a reference route and current pose information to obtain an initial local route that travels along the reference route. This includes: matching the point closest to the reference route based on the current pose information of the vehicle waiting to park, and extracting the path of that point and a preset number of subsequent points as the initial local route. The preset number is obtained based on a preset planning frequency period.

[0064] Step S230: Perform a first coordinate transformation on the initial local route based on the current pose information to obtain the cost relationship, and apply conditional constraints to the cost relationship to obtain the intermediate local route. The conditional constraints include calculating the discrete boundary of the drivable area in the initial local route based on obstacle information, and applying boundary constraints to the cost relationship based on the discrete boundary.

[0065] In one embodiment of this application, the first coordinate transformation of the initial local route based on the current pose information to obtain the cost relationship includes: uniformly sampling the initial local route to obtain multiple reference path points; obtaining lateral offset data and channel lateral offset data based on the offset state of each reference path point relative to the current pose information; and determining the cost relationship based on the lateral offset data, channel lateral offset data, and the trajectory point data of the previous frame in the initial local route.

[0066] In one embodiment of this application, please refer to Figure 5 , Figure 5 A schematic diagram of the trajectory for a first coordinate transformation of an initial local route according to an embodiment of this application is shown. Figure 5 As shown, the line corresponding to the initial local route is also the s-axis of the Frenet coordinate system. The initial local route is uniformly sampled along the preset sampling interval Δs to obtain multiple reference path points Pr in the current frame. The lateral offset data includes the lateral offset l′ of the current pose information relative to the i-th reference path point. i and its first-order transverse partial derivative l′ i and second-order transverse partial derivative l″ i The transformation from the Cartesian coordinate system to the Frenet coordinate system, i.e., the first coordinate transformation, is achieved through uniform sampling and offset states.

[0067] In one embodiment of this application, the lateral offset l′ of the i-th reference path point i and its first-order transverse partial derivative l′ i and second-order transverse partial derivative l″ l As shown in equation (1):

[0068]

[0069] Where l0 is the lateral offset of the first reference path point, l1 is the lateral offset of the second reference path point, and l n-2 l is the lateral offset of the (n-1)th reference path point. n-1 Let l' be the lateral offset of the nth reference path point, l'0 be the first-order lateral partial derivative of the 1st reference path point, Δs be the preset sampling interval, l'1 be the first-order lateral partial derivative of the 2nd reference path point, l'2 be the first-order lateral partial derivative of the 3rd reference path point, and l' ... n-2 Let l′ be the first-order lateral partial derivative of the (n-1)th reference path point. n-1 Let l″0 be the first-order lateral partial derivative of the nth reference path point, l″1 be the second-order lateral partial derivative of the 1st reference path point, l″2 be the second-order lateral partial derivative of the 2nd reference path point, and l″3 be the third reference path point. n-2 Let l″ be the second-order lateral partial derivative of the (n-1)th reference path point. n-1Let n be the second-order lateral partial derivative of the nth reference path point, where n is the total number of reference path points.

[0070] In one embodiment of this application, the third derivative jerk of each local discrete path point is constant, and the third lateral partial derivative is shown in equation (2):

[0071]

[0072] Among them, l i→i+1 Let l be the third-order lateral partial derivative of the (i+2)th reference path point. i+1 The second-order lateral partial derivative of the (i+2)th reference path point, l″ i Let be the second-order lateral partial derivative of the (i+1)th reference path point, Δs be the preset sampling interval, i = 0, 1, ..., n-2, n-1, and n be the total number of reference path points.

[0073] In one embodiment of this application, determining the cost relationship based on lateral offset data, channel lateral offset data, and previous frame trajectory point data in the initial local route includes: determining a trajectory smoothing term based on the square of each lateral offset, the square of each first-order lateral partial derivative, the square of each second-order lateral partial derivative, and the square of each third-order lateral partial derivative, wherein each lateral offset, its first-order lateral partial derivative, second-order lateral partial derivative, and third-order lateral partial derivative are obtained based on the lateral offset data; and determining a proximity to the centerline term based on the squared difference between each lateral offset and each channel lateral offset, wherein each channel lateral offset is used to characterize the narrow channel centerline relative to each... The lateral offset of the reference path point is used to determine the similarity term of the trajectory in the previous frame based on the squared difference between each lateral offset and the lateral offset of the previous frame. The end pose proximity term is determined based on the squared difference between the lateral offset before the end and the lateral offset at the end, the squared difference between the first-order lateral offset before the end and the first-order lateral offset at the end, and the squared difference between the second-order lateral offset before the end and the second-order lateral offset at the end. The lateral offset of the previous frame, the lateral offset before the end, the lateral offset at the end, the first-order lateral offset before the end, the first-order lateral offset at the end, the second-order lateral offset before the end, and the second-order lateral offset at the end are obtained based on the trajectory point data of the previous frame. The cost relationship is determined based on the trajectory smoothing term, the proximity to the centerline term, the similarity term, and the end pose proximity term.

[0074] In one embodiment of this application, each previous frame lateral offset is used to characterize the lateral offset of the previous frame trajectory point relative to the path reference point corresponding to the previous frame trajectory point, the pre-terminal lateral offset is used to characterize the lateral offset of the previous reference path point corresponding to the local endpoint, the first-order pre-terminal lateral derivative is used to characterize the first-order lateral derivative of the previous reference path point corresponding to the local endpoint, the second-order pre-terminal lateral derivative is used to characterize the second-order lateral derivative of the previous reference path point corresponding to the local endpoint, the endpoint lateral offset is used to characterize the lateral offset of the reference path point corresponding to the local endpoint, the first-order endpoint lateral derivative is used to characterize the first-order lateral derivative of the reference path point corresponding to the local endpoint, and the second-order endpoint lateral derivative is used to characterize the second-order lateral derivative of the reference path point corresponding to the local endpoint.

[0075] In one embodiment of this application, the cost relationship is as shown in equation (3):

[0076] J = CosL smooth +Cost ref +CosL similar +Cost end Equation (3)

[0077] Where J represents the cost, and Cost is the value. smooth Cost is the trajectory smoothing term. ref For items closer to the center line, Cost similar Similar terms to the trajectory of the previous frame, Cost end The term is the closest to the endpoint pose.

[0078] In one embodiment of this application, the trajectory smoothing term is as shown in equation (4):

[0079]

[0080] Among them, CosL smooth w is the trajectory smoothing term. l w is the first weighting coefficient. l ' represents the initial first-order derivative weighting coefficient, w l " represents the initial second derivative weighting coefficients and w l "' represents the initial third derivative weighting coefficient, l l Let l′ be the lateral offset of the (i+1)th reference path point. i Let l″ be the first-order lateral partial derivative of the (i+1)th reference path point. i Let l″′ be the second-order lateral partial derivative of the (i+1)th reference path point. i Let be the third-order lateral partial derivative of the (i+2)th reference path point, and n be the total number of reference path points.

[0081] In one embodiment of this application, for narrow passage scenarios such as lane changing and obstacle avoidance, and passing other vehicles, a term close to the centerline is added, as shown in equation (5):

[0082]

[0083] Cost ref For items closer to the center line, l i w is the lateral offset of the (i+1)th reference path point. center The weighting coefficient is the one closest to the center line of the narrow channel. i_ref This represents the lateral offset of the (i+1)th channel, and n is the total number of reference path points.

[0084] In one embodiment of this application, to enhance trajectory stability, except for scenarios requiring rapid and significant trajectory changes such as obstacle avoidance, a similarity term to the trajectory of the previous frame is considered, as shown in equation (6):

[0085]

[0086] Cost similar For the similarity terms of the trajectory in the previous frame, l i w is the lateral offset of the (i+1)th reference path point. similar l represents the similarity weighting coefficient of the previous frame. i_last It represents the horizontal offset of the (i+1)th previous frame.

[0087] In one embodiment of this application, in order to make the local endpoint close to the reference path, an endpoint pose proximity term is considered, as shown in equation (7):

[0088]

[0089] Cost end The final pose is close to the head. The first derivative weighting coefficients at the endpoint. The weighting coefficients are the second derivatives of the endpoint. The third derivative weighting coefficient at the endpoint, l endref l is the lateral offset of the local endpoint. endref ′ is the first-order lateral partial derivative of the reference path point corresponding to the local endpoint, l endref " is the first-order lateral partial derivative of the reference path point corresponding to the local endpoint, l n-1 Let l′ be the lateral offset of the nth reference path point. n-1 Let l″ be the first-order lateral partial derivative of the nth reference path point. n-1 It is the second-order lateral partial derivative of the nth reference path point.

[0090] In one embodiment of this application, calculating the discrete boundary of the drivable area in the initial local route based on obstacle information and constraining the cost relationship based on the discrete boundary includes: determining the pose information of multiple path guidance points based on the initial local route; establishing an initial obstacle boundary corresponding to each path guidance point based on the pose information of each guidance point, wherein the initial size of the initial obstacle boundary is the same as the rectangular outline size of the vehicle waiting to park; extending the four sides of each initial obstacle boundary outward in a preset order and a preset fixed compensation interval until all four sides are target boundaries to obtain the drivable area, wherein the target boundary is determined as the target boundary if any extended edge collides with an obstacle in the obstacle information, or if the extended edge has been extended to a preset maximum extension distance; performing a third coordinate transformation on each target boundary of the drivable area based on the width of the vehicle waiting to park to obtain a discrete boundary; and constraining the cost relationship based on the discrete boundary.

[0091] In one embodiment of this application, please refer to Figure 6 , Figure 6 A flowchart illustrating the solution process for the upper and lower boundaries of a target according to an embodiment of this application is shown. Figure 6 As shown, firstly, in the Frenet coordinate system discrete module, the initial local route is uniformly sampled at a preset sampling interval Δs to obtain multiple reference path points. A grid map is established along the normal direction of the reference path points at a preset grid spacing Δl. Static obstacles in the obstacle information are projected onto the grid map and expanded by half the vehicle width as impassable grid points, and the grid map is output to the path search module. Secondly, the path search module searches and generates sample local routes in the drivable grids. Thirdly, the historical path processing module processes the previous frame's local route corresponding to the trajectory point data, calculates the matching point ID of the vehicle in the previous frame's local path based on the current pose information, discards the path before the matching point, and outputs the matching point and the path points after it as the historical matching path. Fourthly, the guiding trajectory arbitration generation module arbitrates the driving scene, selecting either the sample local route output by the path search module or the historical matching path output by the historical path processing module as the guiding trajectory for calculating the convex hull, obtaining the guiding point pose information of the path guiding point, and outputting it to the Cartesian coordinate system drivable area calculation module. Fifth, the discrete boundary (convex hull) of the drivable region in the Cartesian coordinate system is calculated by the drivable region calculation module. Sixth, the Frenet coordinate system pipeline boundary mapping module transforms the discrete boundary of the drivable region in the Cartesian coordinate system to the Frenet coordinate system and outputs it to the local path smoothing module as the boundary constraint for the cost planning to be determined.

[0092] In one embodiment of this application, please refer to Figure 7 , Figure 7A schematic diagram illustrating the convex hull solution in Cartesian coordinates according to an embodiment of this application is shown. Figure 7 As shown, taking the convex hull calculation of a single path guide point as an example, an initial obstacle boundary (boundingbox) is established based on the guide point pose information (X,Y,θ) of path guide point P1. The initial size of the initial boundingbox is the size of the rectangular outline of the vehicle waiting to park, and the rear axle center point of the vehicle waiting to park coincides with the path guide point P1; the orientation of the initial boundingbox is consistent with the orientation θ of point P1. From the four sides of the initial boundingbox, the boundaries are expanded outward in a clockwise order with a preset fixed compensation (ΔD), and a preset maximum expansion distance (MAX_D) is set for each direction. When the expanded edge collides with an obstacle in the obstacle information, or when the direction expands to MAX_D, the direction that can no longer be expanded is removed from the potential expansion directions, and the expanded edge is determined as the target boundary. Other directions continue to expand in a clockwise order until all directions cannot be expanded, or all directions have expanded to the maximum distance, thus obtaining the drivable area. The third coordinate transformation involves projecting all target boundaries corresponding to the drivable areas generated by all path guide points from the Cartesian coordinate system to the Frenet coordinate system. Each target boundary is discretized into (s,l) value pairs, where s is the ordinate and l is the abscissa in the Frenet coordinate system. Since the drivable areas generated by adjacent path guide points will inevitably overlap, if there are value pairs with the same s value, only the L values ​​corresponding to the left and right closest path guide points are retained for the l value. The discrete boundaries are obtained by subtracting half the width of the waiting vehicle from all (S,L) value pairs corresponding to L values, thus completing an approximately linear mapping from the drivable area in the Cartesian coordinate system to the Frenet coordinate system.

[0093] In one embodiment of this application, the condition constraints further include: applying path continuity constraints to the cost relationship based on the multi-order derivative expansion of the lateral offset data, applying curvature constraints to the cost relationship based on local curvature data, lateral offset data, and driving information, and applying at least one of the starting point constraints to the cost relationship based on driving information, lateral offset data, and trajectory point data from the previous frame; wherein the local curvature data is obtained by sampling the initial local route.

[0094] In one embodiment of this application, by the lateral offset l′ i and its first-order transverse partial derivative l′ i and second-order transverse partial derivative l″ l This optimizes the trajectory shape of the initial local route and connects the local discrete path points corresponding to each local planning path using the third-order lateral partial derivative.

[0095] In one embodiment of this application, the path continuity constraint is as shown in equation (8):

[0096]

[0097] Among them, l i+1 Let l be the lateral offset of the (i+2)th reference path point. i Let l′ be the lateral offset of the (i+1)th reference path point. i Let l″ be the first-order lateral partial derivative of the (i+1)th reference path point, Δs be the preset sampling interval, and l″ be the second-order partial derivative of the reference path point. i Let l″′ be the second-order lateral partial derivative of the (i+1)th reference path point. i→i+1 Let l′ be the third-order lateral partial derivative of the (i+2)th reference path point. i+1 Let l″ be the first-order lateral partial derivative of the (i+2)th reference path point. i+1 Let be the second-order lateral partial derivative of the (i+2)th reference path point, where i = 0, 1, ..., n-2, n-1, and n is the total number of reference path points.

[0098] In one embodiment of this application, constraining the cost relationship based on driving information, lateral offset data, and trajectory point data from the previous frame includes: if the current planning scene state is a replanning scene, then determining the starting point constraint based on the lateral offset data and the starting point lateral offset; if the current planning scene state is not a replanning scene, then determining the starting point constraint based on the lateral offset data and the endpoint lateral offset in the trajectory point data from the previous frame; constraining the cost relationship based on the starting point constraint; wherein, the driving information also includes the current planning scene state, the starting point lateral offset is obtained based on the lateral offset data corresponding to the current pose information, and the replanning scene is used to characterize the first frame scene and the obstacle avoidance scene.

[0099] In one embodiment of this application, to ensure the continuity and comfort of the trajectory in each frame, a starting point constraint is considered, as shown in equation (9):

[0100]

[0101] Where l0 is the lateral offset of the first reference path point, l1 is the lateral offset of the second reference path point, and l0 = l cur_last In scenarios where replanning is not required, l cur_last To match the endpoint of the trajectory from the previous frame, l cur+1_last Δs is the lateral offset of the next sampling point of the trajectory matching endpoint of the previous frame; Δs is the preset sampling interval. The lateral offsets of the trajectory matching endpoint of the previous frame and the next sampling point of the trajectory matching endpoint of the previous frame are obtained from the endpoint lateral offset in the trajectory point data of the previous frame.

[0102] In one embodiment of this application, for a replanning scenario in a driving scene, the replanning scenario includes a first frame scenario and an obstacle avoidance scenario. The obstacle avoidance scenario includes an obstacle avoidance scenario, an escape scenario, and a scenario where the vehicle reverses to avoid an obstacle and then switches back to local path planning. The l in equation (9) cur_last This refers to the lateral offset of the vehicle's current position and location, i.e., the lateral offset of the first reference path point. cur+1_last This is the lateral offset at the next sampling point in the current heading angle direction, i.e., the lateral offset of the second reference path point.

[0103] In one embodiment of this application, applying curvature constraints to the cost relationship based on local curvature data, lateral offset data, and driving information includes: determining each first curvature parameter based on the vehicle's maximum steering angle, each lateral offset data, and each reference curvature; the driving information further includes the vehicle's wheelbase and maximum steering angle; and the local curvature data includes the reference curvature corresponding to each reference path point; determining a second curvature parameter based on the vehicle's maximum steering angle; determining each third curvature parameter based on the vehicle's wheelbase and each reference curvature; determining curvature constraints based on the second curvature parameter, each first curvature parameter, and each third curvature parameter; and applying curvature constraints to the cost relationship.

[0104] In one embodiment of this application, considering the curvature constraint of the lateral offset of the initial local route, the curvature constraint corresponding to the (i+1)th reference path point is as shown in equation (10):

[0105]

[0106] Where, α max This is the vehicle's maximum steering angle. Let l be the reference curvature of the i-th reference path point. i Let L be the lateral offset of the i-th reference path point, and L be the vehicle wheelbase.

[0107] In one embodiment of this application, the process of constraining the cost relationship to obtain an intermediate local route includes: converting boundary constraints, path continuity constraints, starting point constraints, curvature constraints, and cost relationships into matrix form based on a preset general form of quadratic programming to obtain the cost plan to be determined; solving the cost plan to be determined by minimizing the cost to obtain the intermediate local route; the constraints also include path continuity constraints, starting point constraints, and curvature constraints.

[0108] In one embodiment of this application, the path continuity constraint, cost relationship, curvature constraint, starting point constraint, and upper and lower boundary constraints are converted into the general form of quadratic programming according to equations (3) to (10) in the above embodiments, resulting in the cost programming problem as shown in equation (11):

[0109]

[0110] Where x is the optimization variable, x = {l0, l1, ..., l n-1 l′0, l′1, ..., l′ n-1 ,l″0,l″1,...,l″ n-1}, l0 is the lateral offset of the first reference path point, l1 is the lateral offset of the second reference path point, l n-1 Let l' be the lateral offset of the nth reference path point, l'0 be the first-order lateral partial derivative of the 1st reference path point, Δs be the preset sampling interval, l'1 be the first-order lateral partial derivative of the 2nd reference path point, and l'' be the lateral offset of the nth reference path point. n-1 Let l″0 be the first-order lateral partial derivative of the nth reference path point, l″1 be the second-order lateral partial derivative of the 1st reference path point, and l″2 be the second-order lateral partial derivative of the 2nd reference path point. n-1 Let H be the second-order lateral partial derivative of the nth reference path point, H be the quadratic coefficient matrix of the optimization variable in the cost relation, f be the linear coefficient matrix of the optimization variable in the cost relation, A be the inequality constraint coefficient matrix of the optimization variable in the curvature constraint, b be the constant vector of the inequality constraint in the curvature constraint, Aeq be the equality constraint coefficient matrix of the optimization variable in the path continuity constraint and the starting point constraint, beq be the constant vector of the equality constraint in the path continuity constraint and the starting point constraint, lb be the lower boundary of the optimization variable in the boundary constraint, and ub be the upper boundary of the optimization variable in the boundary constraint.

[0111] In one embodiment of this application, a Sequential Quadratic Programming solver (SQP solver) is used to solve the cost programming problem by minimizing the cost, thereby obtaining the lateral deviation l in the Frenet coordinate system. i .

[0112] Step S240: Perform a second coordinate transformation on the intermediate local route based on the reference route to obtain the target local route, so as to perform valet parking according to the target local route.

[0113] In one embodiment of this application, the lateral offset in the intermediate local route is transformed by the reference Cartesian coordinates in the reference route to obtain the target Cartesian coordinates, thereby obtaining the target local route, which is to say, the second coordinate transformation is realized.

[0114] In one embodiment of this application, a general local path planning method is proposed. This method incorporates the stability, safety, and feasibility of locally planned paths into the optimization problem, ensuring path stability in various scenarios and making it applicable to local path planning on roads with varying curvatures. It generates a safe and comfortable locally planned path within each preset planning frequency cycle. Furthermore, this application addresses the challenges faced by local path planning in valet parking tasks, such as large variations in road curvature, complex static obstacle layouts in the environment, and uncertainties in the dynamic target position and trajectory. Currently, local path planning in parking lot road scenarios is typically performed in the Frenet coordinate system. By converting irregular roads into straight roads within the Frenet framework, and transforming roads of any shape into standard straight pipelines with left and right boundaries, the nonlinear collision avoidance constraints are transformed into linear constraints within the pipeline. However, local path planning methods based on the Frenet coordinate system are not suitable for curves with high curvature, and collision avoidance constraints become inaccurate within the linearized pipeline, making it difficult to guarantee the safety of the planned trajectory. The embodiments of this application solve this problem.

[0115] Please see Figure 8 , Figure 8 A block diagram of a valet parking trajectory planning device according to an embodiment of this application is shown. This device can be applied to... Figure 1 The implementation environment shown is specifically configured in the parking vehicle 102. This device can also be applied to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which the device is applicable.

[0116] like Figure 8 As shown, the exemplary valet parking trajectory planning device 800 includes: an information acquisition module 801, a path planning module 802, a constraint solving module 804, and a coordinate transformation module 804.

[0117] Among them, the information acquisition module 801 is used to acquire the driving information and external information of the vehicle waiting to be parked. The driving information includes at least the initial pose information and the current pose information, and the external information includes at least the initial starting point information, the target parking position and obstacle information.

[0118] The path planning module 802 is used to perform global path planning on the path between the initial starting point information and the target parking position based on the electronic map, initial pose information and obstacle information to obtain a reference route, and to perform local path planning based on the reference route and the current pose information to obtain an initial local route along the reference route.

[0119] The constraint solving module 803 is used to perform a first coordinate transformation on the initial local route based on the current pose information to obtain the cost relationship, and to impose conditional constraints on the cost relationship to obtain the intermediate local route. The conditional constraints include calculating the discrete boundary of the drivable area in the initial local route based on obstacle information, and imposing boundary constraints on the cost relationship based on the discrete boundary.

[0120] The coordinate transformation module 804 is used to perform a second coordinate transformation on the intermediate local route based on the reference route to obtain the target local route, so as to perform valet parking according to the target local route.

[0121] It should be noted that the valet parking trajectory planning device and the valet parking trajectory planning method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the valet parking trajectory planning device provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0122] Embodiments of this application also provide an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the valet parking trajectory planning method provided in the above embodiments.

[0123] Please see Figure 9 , Figure 9 A schematic diagram of a computer system suitable for implementing the embodiments of this application is shown. It should be noted that... Figure 9 The computer system 900 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0124] like Figure 9As shown, the computer system 900 includes a Central Processing Unit (CPU) 901, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 902 or programs loaded from storage portion 908 into Random Access Memory (RAM) 903. The RAM 903 also stores various programs and data required for system operation. The CPU 901, ROM 902, and RAM 903 are interconnected via a bus 904. An Input / Output (I / O) interface 905 is also connected to the bus 904.

[0125] The following components are connected to I / O interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to I / O interface 905 as needed. Removable media 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 910 as needed so that computer programs read from them can be installed into storage section 908 as needed.

[0126] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 909, and / or installed from removable medium 911. When the computer program is executed by central processing unit (CPU) 901, it performs various functions defined in the system of this application.

[0127] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0129] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0130] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the valet parking trajectory planning method provided in the various embodiments described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not assembled into the electronic device.

[0131] In the above embodiments, unless otherwise specified, the use of ordinal numbers such as "first" and "second" to describe common objects only indicates that they refer to different instances of the same object, rather than indicating that the objects being described must be in a given order, whether temporally, spatially, sequentially, or in any other way.

[0132] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A valet parking trajectory planning method, characterized in that, The valet parking trajectory planning method includes: Acquire driving information and external information of the vehicle waiting to be parked. The driving information includes at least initial pose information and current pose information, and the external information includes at least initial starting point information, target parking position and obstacle information. Based on the electronic map, the initial pose information, and the obstacle information, global path planning is performed on the path between the initial starting point information and the target parking position to obtain a reference route. Then, based on the reference route and the current pose information, local path planning is performed to obtain an initial local route that travels along the reference route. The initial local route is transformed using the current pose information to obtain a cost relationship. Conditional constraints are then applied to the cost relationship to obtain an intermediate local route. These constraints include calculating discrete boundaries of the drivable area within the initial local route based on the obstacle information, and applying boundary constraints to the cost relationship based on these discrete boundaries. Specifically, the pose information of multiple path guide points is determined based on the initial local route. An initial obstacle boundary is established for each path guide point based on its pose information. The initial size of the initial obstacle boundary is the same as the rectangular outline size of the vehicle waiting to park. The four sides of each initial obstacle boundary are extended outwards in a preset order and with a preset fixed compensation interval until all four sides are target boundaries, resulting in a drivable area. The target boundary is determined if any extended edge collides with an obstacle in the obstacle information, or if the extended edge has reached a preset maximum extension distance. Each target boundary of the drivable area is transformed using a third coordinate system based on the width of the vehicle waiting to park, resulting in a discrete boundary. Boundary constraints are then applied to the cost relationship based on these discrete boundaries. Based on the reference route, a second coordinate transformation is performed on the intermediate local route to obtain the target local route, so as to perform valet parking according to the target local route.

2. The valet parking trajectory planning method according to claim 1, characterized in that, Based on the current pose information, a first coordinate transformation is performed on the initial local route to obtain the cost relationship, including: The initial local route is uniformly sampled to obtain multiple reference path points; Based on the offset state of the current pose information relative to each of the reference path points, lateral offset data and channel lateral deviation data are obtained, wherein the channel lateral deviation is used to characterize the lateral deviation of the narrow channel centerline relative to each reference path point. The cost relationship is determined based on the lateral offset data, the channel lateral offset data, and the previous frame trajectory point data in the initial local route.

3. The valet parking trajectory planning method according to claim 2, characterized in that, Determining the cost relationship based on the lateral offset data, the channel lateral offset data, and the previous frame trajectory point data in the initial local route includes: The trajectory smoothing term is determined based on the square of each lateral offset, the square of each first-order lateral partial derivative, the square of each second-order lateral partial derivative, and the square of each third-order lateral partial derivative. Each lateral offset, the first-order lateral partial derivative, the second-order lateral partial derivative, and the third-order lateral partial derivative are obtained based on the lateral offset data. The term closest to the centerline is determined based on the squared difference between each lateral offset and each channel lateral offset, wherein each channel lateral offset is used to characterize the lateral offset of the narrow channel centerline relative to each of the reference path points. The similarity terms of the trajectory in the previous frame are determined based on the squared difference between each horizontal offset and the horizontal offset of each previous frame. The endpoint pose proximity term is determined based on the squared difference between the pre-terminal lateral offset and the endpoint lateral offset, the squared difference between the first-order pre-terminal lateral derivative and the first-order endpoint lateral derivative, and the squared difference between the second-order pre-terminal lateral derivative and the second-order endpoint lateral derivative. The lateral offset of each previous frame, the pre-terminal lateral offset, the endpoint lateral offset, the first-order pre-terminal lateral derivative, the first-order endpoint lateral derivative, the second-order pre-terminal lateral derivative, and the second-order endpoint lateral derivative are obtained based on the trajectory point data of the previous frame. The cost relationship is determined based on the trajectory smoothing term, the proximity to the center line term, the similarity term, and the endpoint pose proximity term.

4. The valet parking trajectory planning method according to claim 2, characterized in that, The condition constraints also include at least one of the following: applying path continuity constraints to the cost relationship based on the multi-order derivative expansion of the lateral offset data, applying curvature constraints to the cost relationship based on local curvature data, the lateral offset data, and the driving information, and applying start point constraints to the cost relationship based on the driving information, the lateral offset data, and the trajectory point data of the previous frame. The local curvature data is obtained by sampling the initial local route.

5. The valet parking trajectory planning method according to claim 4, characterized in that, The starting point constraint on the cost relationship based on the driving information, the lateral offset data, and the trajectory point data from the previous frame includes: If the current planning scenario is a replanning scenario, then the starting point constraint is determined based on the lateral offset data and the starting point lateral offset. If the current planning scenario is not a replanning scenario, then the starting point constraint is determined based on the lateral offset data and the endpoint lateral offset in the trajectory point data of the previous frame. Apply a starting point constraint to the cost relationship; The driving information also includes the current planning scene status, the starting point lateral offset is obtained based on the lateral offset data corresponding to the current pose information, and the replanning scene is used to characterize the first frame scene and the obstacle avoidance scene.

6. The valet parking trajectory planning method according to claim 4, characterized in that, Curvature constraints are applied to the cost relationship based on local curvature data, the lateral offset data, and the driving information, including: Each first curvature parameter is determined based on the vehicle's maximum steering angle, each lateral offset data, and each reference curvature. The driving information also includes the vehicle's wheelbase and the vehicle's maximum steering angle. The local curvature data includes the reference curvature corresponding to each reference path point. The second curvature parameter is determined based on the vehicle's maximum steering angle; Each third curvature parameter is determined based on the vehicle wheelbase and each reference curvature. Curvature constraints are determined based on the second curvature parameter, each of the first curvature parameter, and each of the third curvature parameter; Curvature constraints are applied to the cost relationship.

7. The valet parking trajectory planning method according to any one of claims 1-6, characterized in that, By applying conditional constraints to the cost relationship, the intermediate local routes are obtained, including: Based on the preset general form of quadratic programming, the boundary constraints, path continuity constraints, starting point constraints, curvature constraints and the cost relationship are converted into matrix form to obtain the cost program to be solved. The minimum cost solution is performed on the cost planning problem to obtain the intermediate local route; The conditional constraints also include the path continuity constraint, the starting point constraint, and the curvature constraint.

8. The valet parking trajectory planning method according to any one of claims 1-6, characterized in that, Based on the electronic map, the initial pose information, and the obstacle information, global path planning is performed on the path between the initial starting point information and the target parking position to obtain a reference route, including: A lane candidate list is generated based on the initial pose information, the initial starting point information, the target parking position, and the lane-parking space association. Based on the lane candidate list and the initial pose information, the optimal path is searched on the search map to obtain the initial global path; The initial global path is smoothed based on the lane reference lines and the obstacle information to obtain a reference route; The search map, the lane-parking space association, and the lane reference line are obtained based on the electronic map. The search map is used to represent the length connection relationship of roads, road segments, and lanes in the parking lot, and the lane-parking space association is used to represent the association relationship of lanes, parking spaces, and coordinate points in the parking lot.

9. The valet parking trajectory planning method according to claim 8, characterized in that, The lane candidate list generated based on the initial pose information, the initial starting point information, the target parking position, and the lane-parking space association includes: If the initial starting point information is a starting point coordinate point, then a virtual node is generated based on the angle between the initial heading angle in the initial pose information and the heading angle of the discrete point of the lane reference line corresponding to the starting point coordinate point, and the virtual node is connected to the first starting point lane to obtain a first candidate list. The discrete point of the lane reference line is obtained based on the lane parking space association relationship. The first starting point lane is used to represent all drivable lanes matched from the lane parking space association relationship based on the starting point coordinate point and a preset distance. If the initial starting point information is a starting parking space, then the relationship between the starting parking space and the lane parking space is matched to obtain a second starting lane, and the second starting lane that meets the preset distance is determined as the second candidate list; If the target parking location is the endpoint coordinate point, then the first endpoint lane corresponding to the endpoint coordinate point in the lane parking space association relationship is used as the third candidate list based on the preset distance; If the target parking location is the destination parking space, then the relationship between the destination parking space and the lane parking space is matched to obtain the second destination lane, and the second destination lane that meets the preset distance is used as the fourth candidate list; The starting point alternative list and the ending point alternative list are used as lane alternative lists. The starting point alternative list includes a first alternative list or a second alternative list, and the ending point alternative list includes a third alternative list or a fourth alternative list.

10. The valet parking trajectory planning method according to claim 8, characterized in that, The initial global path is smoothed based on the lane reference lines and the obstacle information to obtain a reference route, including: The initial global path is assessed for collision risk based on the left and right boundary values ​​of the obstacle to obtain the collision risk status. The left and right boundary values ​​of the obstacle are obtained based on the lane reference line and the obstacle information. If the collision risk status is that there is a collision risk, an optimization target is generated based on the path length information, obstacle distance information, and curvature information. A penalty term is generated based on the obstacle information and added to the optimization target. The optimization target is then minimized based on the lane reference line to obtain a reference route. The path length information, obstacle distance information, and initial curvature information are obtained based on the initial global path.

11. A valet parking trajectory planning device, characterized in that, The valet parking trajectory planning device includes: The information acquisition module is used to acquire the driving information and external information of the vehicle waiting to be parked. The driving information includes at least the initial pose information and the current pose information, and the external information includes at least the initial starting point information, the target parking position, and obstacle information. The path planning module is used to perform global path planning on the path between the initial starting point information and the target parking position based on the electronic map, the initial pose information and the obstacle information to obtain a reference route, and to perform local path planning based on the reference route and the current pose information to obtain an initial local route along the reference route. The constraint solving module is used to perform a first coordinate transformation on the initial local route based on the current pose information to obtain a cost relationship, and to apply conditional constraints to the cost relationship to obtain an intermediate local route. The conditional constraints include calculating discrete boundaries of drivable areas in the initial local route based on the obstacle information, and applying boundary constraints to the cost relationship based on the discrete boundaries. Specifically, the module determines the pose information of multiple path guidance points based on the initial local route; and establishes an initial obstacle boundary corresponding to each path guidance point based on the pose information of each guidance point. The initial size of the initial obstacle boundary is related to the initial coordinates of the path guidance points. The rectangular outlines of the vehicles waiting to park are all the same size. The four sides of each initial obstacle boundary are extended outwards in a preset order and with a preset fixed compensation interval until all four sides are target boundaries, thus obtaining a drivable area. The target boundary is determined by whether any outwardly extended edge collides with an obstacle in the obstacle information, or whether the extended edge has extended to a preset maximum extension distance. Based on the width of the vehicles waiting to park, each target boundary of the drivable area is transformed using a third coordinate system to obtain discrete boundaries. Boundary constraints are applied to the cost relationship based on these discrete boundaries. The coordinate transformation module is used to perform a second coordinate transformation on the intermediate local route based on the reference route to obtain the target local route, so as to perform valet parking according to the target local route.

12. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the valet parking trajectory planning method as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by the computer's processor, causes the computer to perform the valet parking trajectory planning method according to any one of claims 1 to 10.

Images