WO2022007227A1 - 一种自动泊车的方法和车辆 - Google Patents

一种自动泊车的方法和车辆 Download PDF

Info

Publication number
WO2022007227A1
WO2022007227A1 PCT/CN2020/121566 CN2020121566W WO2022007227A1 WO 2022007227 A1 WO2022007227 A1 WO 2022007227A1 CN 2020121566 W CN2020121566 W CN 2020121566W WO 2022007227 A1 WO2022007227 A1 WO 2022007227A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
parking
path
nodes
neighbor
Prior art date
Application number
PCT/CN2020/121566
Other languages
English (en)
French (fr)
Inventor
许扬
苏镜仁
许匡正
赖健明
陈盛军
Original Assignee
广东小鹏汽车科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广东小鹏汽车科技有限公司 filed Critical 广东小鹏汽车科技有限公司
Priority to EP20930646.3A priority Critical patent/EP3971047B1/en
Publication of WO2022007227A1 publication Critical patent/WO2022007227A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
    • B60W30/06—Automatic manoeuvring for parking
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
    • B62D—MOTOR VEHICLES; TRAILERS
    • B62D15/00—Steering not otherwise provided for
    • B62D15/02—Steering position indicators ; Steering position determination; Steering aids
    • B62D15/027—Parking aids, e.g. instruction means
    • B62D15/0285—Parking performed automatically
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3446—Details of route searching algorithms, e.g. Dijkstra, A*, arc-flags, using precalculated routes
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/36—Input/output arrangements for on-board computers
    • G01C21/3679—Retrieval, searching and output of POI information, e.g. hotels, restaurants, shops, filling stations, parking facilities
    • G01C21/3685—Retrieval, searching and output of POI information, e.g. hotels, restaurants, shops, filling stations, parking facilities the POI's being parking facilities

Definitions

  • the present invention relates to the technical field of parking, in particular to an automatic parking method and a vehicle.
  • the geometry-based path planning method In order to avoid planning failure, when using a geometry-based path planning method, it is usually necessary to manually set the reversing point for parking, and it is difficult for novice drivers to accurately control the reversing point for parking. Moreover, the geometry-based path planning method has poor universality, and is only suitable for parking scenarios with wide parking spaces and aisles, and cannot achieve efficient path planning for other complex scenarios.
  • a method for automatic parking comprising:
  • the node distribution information for the to-be-parked area is determined; wherein, the node distribution information includes a plurality of nodes that the vehicle can pass through;
  • an intermediate node is determined, and a neighbor node search is performed based on the intermediate node to obtain a plurality of neighbor nodes;
  • the candidate parking path is a path from an initial node to the intermediate node via the neighbor nodes;
  • a target neighbor node is determined, and automatic parking is performed according to the target neighbor node.
  • the automatic parking according to the target neighbor node includes:
  • the parking path includes a target path corresponding to the target neighbor node;
  • the determining an intermediate node according to the node distribution information includes:
  • intermediate nodes are determined.
  • performing a neighbor node search based on the intermediate node to obtain multiple neighbor nodes including:
  • a node that does not have an obstacle between the intermediate node and the intermediate node is a neighbor node.
  • the determining, respectively, the cost information of the candidate paths corresponding to the multiple neighboring nodes includes:
  • path attribute information including but not limited to path length, obstacle distance, and vehicle commutation times
  • cost information is determined.
  • connection function For any two nodes in the parking path, a preset connection function is used for connection; wherein, the connection function corresponds to a variety of connection modes.
  • the parking feature information includes any one or more of the following:
  • the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle is the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle.
  • a vehicle comprising:
  • a node distribution determination module configured to determine node distribution information for the to-be-parked area in the process of automatic parking; wherein the node distribution information includes a plurality of nodes that the vehicle can pass through;
  • an intermediate node determination module configured to determine an intermediate node according to the node distribution information, and perform a neighbor node search based on the intermediate node to obtain a plurality of neighbor nodes;
  • a cost information determination module configured to respectively determine the cost information of the candidate paths corresponding to the plurality of neighbor nodes; wherein, the candidate parking path is a path from an initial node to the intermediate node via the neighbor node;
  • the target neighbor node determination module is configured to determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node.
  • a vehicle comprising a processor, a memory, and a computer program stored on the memory and capable of running on the processor, the computer program being executed by the processor to implement the above-mentioned method for automatic parking .
  • a computer-readable storage medium storing a computer program on the computer-readable storage medium, when the computer program is executed by a processor, implements the above-mentioned automatic parking method.
  • the node distribution information for the to-be-parked area is determined, the intermediate nodes are determined according to the node distribution information, and the neighbor nodes are searched based on the intermediate nodes to obtain a plurality of neighbor nodes. , respectively determine the cost information of the candidate paths corresponding to multiple neighbor nodes, determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node, which realizes the optimization of the automatic parking path and can be applied to complex
  • the parking scene is universal, that is, the search efficiency can be improved by searching the neighbor nodes, and the cost information of the path from the starting node to the intermediate node through the neighbor node is used to select the target neighbor node. Compared with the method in which the nearest neighbor of the intermediate node is used as the parent node of the intermediate node, the cost of the overall path is fully considered, and the parking path with the least cost can be selected for parking.
  • 1a is a schematic diagram of a vertical parking space sampling area provided by an embodiment of the present invention.
  • Fig. 1b is a schematic diagram of a parallel parking space sampling area provided by an embodiment of the present invention.
  • FIG. 2 is a flowchart of steps of a method for automatic parking provided by an embodiment of the present invention
  • 3a is a schematic diagram of determining a target neighbor node according to an embodiment of the present invention.
  • FIG. 3b is a schematic diagram after determining a target neighbor node according to an embodiment of the present invention.
  • FIG. 3c is a schematic diagram of rerouting of neighboring nodes according to an embodiment of the present invention.
  • 3d is a schematic diagram of a neighboring node after rewiring provided by an embodiment of the present invention.
  • FIG. 5 is a schematic flowchart of planning a minimum cost path according to an embodiment of the present invention.
  • FIG. 6 is a schematic structural diagram of a vehicle according to an embodiment of the present invention.
  • the Rapidly Exploring Random Tree (RRT) algorithm is an incremental sampling search algorithm whose advantage is to generate the final route by connecting discrete nodes in the configuration space.
  • the RRT algorithm tends to expand into open unexplored areas, when the time is long enough and the number of iterations is enough, there are no areas that will not be explored.
  • the RRTstar algorithm adopts the iterative deepening idea to optimize the planning results obtained by the RRT algorithm.
  • sampling area of RRT is selected according to the size of the parking space, the type of parking space (which may include vertical parking spaces, parallel parking spaces, and oblique parking spaces), and the location of obstacles.
  • the shaded part in Figure 1a represents the sampling area of the vertical parking space
  • the shaded part in Figure 2a is the sampling area of the parallel parking space.
  • the straight-line distance (Euclidean distance) between two nodes can be used as the path cost between two nodes, while in an embodiment of the present invention, the path and obstacles can be comprehensively considered
  • the distance of the object, the number of paths, the direction of the path, and the curvature of the path are used to calculate the path cost, so as to ensure the safety of the planned path and reduce the number of kneading.
  • the path cost between two nodes can be calculated as:
  • Path cost a*path length-b*obstacle distance coefficient+c*commutation times
  • a, b, and c are constants greater than 0.
  • the flow of the adopted planning algorithm is as follows:
  • the starting pose can be Represented by startpos(xs, ys, ⁇ s)
  • the target pose can be represented by tarpos(xg, yg, ⁇ g)
  • multiple obstacles and parking space frame segments can pass through (x11, y11, x12, y12), (x21, y21,x22,y22),...(xn1,yn1,xn2,yn2) to indicate that after calculating all the parameters of the input, the output result obtained is from the starting pose to the target pose
  • the collision-free path, the collision-free path can be a combination of straight lines and arcs.
  • FIG. 2 a flowchart of steps of a method for automatic parking provided by an embodiment of the present invention is shown, which may specifically include the following steps:
  • Step 201 in the process of automatic parking, determine the node distribution information for the area to be parked; wherein, the node distribution information includes a plurality of nodes that the vehicle can pass through;
  • the valid nodes obtained by each operation can be recorded. These valid nodes are the nodes that the vehicle can pass through during the parking process, and all valid nodes can form nodes. Therefore, in the process of automatic parking, the node distribution information in the to-be-parked area can be determined.
  • each node contains the corresponding pose (x, y, ⁇ ) of the node, the cost of the node, the index of the node, the index of the parent node of the node, the path information from the parent node of the node to the node, the parent node of the node is from In the path from the starting point to the node, the previous node of the node.
  • Step 202 Determine an intermediate node according to the node distribution information, and perform a neighbor node search based on the intermediate node to obtain a plurality of neighbor nodes;
  • an intermediate node can be determined according to the node distribution information, and the intermediate node is a node that the vehicle passes through in the process of moving from the starting position to the target position.
  • the intermediate node is determined by the acquired node distribution information, which can avoid the problem of inaccurate judgment of the steering point caused by manually setting the steering point.
  • multiple intermediate nodes may be determined at one time, and then path planning is performed for each determined intermediate node to obtain one or more planning results; one intermediate node may also be determined at a time, and by setting an iterative loop, at the preset Within the number of iterations, an intermediate node is determined in each loop, and planning is performed for each determined intermediate node to obtain one or more planning results.
  • a preset search range can be determined based on the intermediate node, and a neighbor node search can be performed within the preset range to obtain multiple neighbor nodes.
  • the adjacent nodes are used to search within a preset range to avoid blind search, which is beneficial to improve search efficiency.
  • the preset range can be set to a circle with R as the radius, where R can be calculated by the following formula:
  • R represents the radius of the circle in the preset search range when the intermediate node is the center of the circle for the nearest neighbor search
  • n node is the number of existing nodes.
  • a to K are valid nodes distributed in the parking area, and the number on the line connecting the nodes is the path cost information between the two nodes.
  • the intermediate node determined by node K, Xnew Indicates that the K node is a new node.
  • the neighbor node search is performed along with the K node, and the neighbor nodes are determined as F and G.
  • Step 203 respectively determining the cost information of the candidate paths corresponding to the plurality of neighbor nodes; wherein, the candidate parking path is a path from the starting node to the intermediate node via the neighbor node;
  • the corresponding candidate path can be determined from the starting point via the neighbor node to the intermediate node, and then the cost information of each candidate path can be calculated.
  • the neighbor nodes of the intermediate node K within the range of R are G and F.
  • the determining, respectively, the cost information of the candidate paths corresponding to the multiple neighboring nodes may include the following sub-steps:
  • path attribute information including but not limited to path length, obstacle distance, and vehicle reversal times; use the path attribute information to determine cost information.
  • the corresponding path attribute information can be determined for the candidate path, and then the path cost information can be obtained through calculation by using the path attribute information.
  • the path attribute information may include, but is not limited to, path length, obstacle distance, and vehicle commutation times.
  • the path cost can be calculated by the following formula:
  • Path cost a*path length-b*obstacle distance coefficient+c*commutation times
  • a, b, and c are constants greater than 0.
  • Step 204 Determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node.
  • the neighbor node corresponding to the target path with the least cost among all the candidate paths can be determined as the target neighbor node according to the cost information, and the target path from the starting point to the end point via the target neighbor node is the minimum cost path, and then can automatically park according to the determined target neighbor nodes.
  • the neighbor node corresponding to the candidate path with the smallest path cost can be selected as the target neighbor node, and finally the parking path with the smallest path cost can be used for parking.
  • the corresponding calculation cost information is that the cost of ABDGK is 15, the cost of ABCFK is 10, and the cost of the candidate path corresponding to node F is is small, then ABCFK is the target path, and node F is the target neighbor node.
  • Figure 3b is a schematic diagram of the wiring after node F is the target neighbor node, and node F is used as the target neighbor node for automatic parking.
  • the step of performing automatic parking according to the target neighbor node may include the following sub-steps:
  • re-wiring is performed to obtain a parking path; automatic parking is performed according to the parking path.
  • the parking path may include the target path corresponding to the target neighbor node.
  • the target neighbor node and the intermediate node can be used for rewiring, and a parking path can be obtained according to the rerouting result, and then automatic parking can be performed according to the parking path.
  • the rewiring may be for the adjacent nodes of the intermediate node or for the rewiring between the intermediate node and the target node.
  • each neighbor node of the intermediate node may also be rerouted. Specifically, a plurality of second candidate paths of each neighboring node from the starting point to the intermediate node may be determined first, and the path cost information of each second candidate path corresponding to each neighboring node may be calculated separately. According to the second candidate path The path cost information to determine whether rerouting is required, and rerouting is performed for the neighboring nodes that need rerouting.
  • the significance of the rerouting process is that whenever a central node (new node) is generated, the path cost of some nodes can be reduced by rerouting. From an overall perspective, not every rewired node will appear in the final generated path, but in the process of generating random trees, each rerouting creates an opportunity to reduce the final path cost as much as possible. .
  • Path1:A-B-C-F, the calculated path cost is 2+1+4 7;
  • the target path is ABCFK
  • the cost of Path1 for F is less than Path2
  • ABCF is the second candidate path with the smallest path cost. Node rewiring.
  • the routing can be continuously optimized in each iteration, and the parking path can be obtained more intelligently.
  • the target neighbor node after determining the target neighbor node, it can be further determined whether there is an obstacle between the intermediate node and the target node, and when there is no obstacle between the intermediate node and the target node, the intermediate node and the target node can be rewired,
  • the parking path can be determined according to the target path corresponding to the target adjacent node and the rewiring between the intermediate node and the target node, and then automatic parking can be performed according to the parking path.
  • the step of determining the intermediate node is performed.
  • the method may further include the following steps:
  • connection function For any two nodes in the parking path, a preset connection function is used for connection; wherein, the connection function corresponds to a variety of connection modes.
  • any two nodes in the parking path can be connected through a preset connection function, wherein the preset connection function corresponds to a variety of connection methods.
  • connection may include one or more of the following:
  • the ReedsSheep curve can contain at least 3 paths, line-arc-line, arc-line-arc, arc-arc-arc.
  • connection methods that can be selected between nodes, which can improve the success rate of connection and further improve the success rate of path planning; the planned path is also more in line with human parking habits; the number of planned paths is less.
  • the nearest neighbor node of the intermediate node can be determined first, that is, the nearest neighbor node that can be connected to the intermediate node with the least cost within a preset range, and then it can be determined whether there is an obstacle between the node and the intermediate node. , when there is no obstacle between the node and the intermediate node, the cost information of the candidate path can be calculated separately through a set of neighbor nodes obtained by searching, and the target neighbor node can be determined according to the cost information of the candidate path.
  • the step of determining the intermediate node is performed.
  • the intermediate node K is within the range of a circle with a radius of R, and there are adjacent nodes G and F, where the cost of GK is 2 and the cost of FK is 3.
  • the node G can be regarded as the parent node of the intermediate node K first.
  • the candidate paths corresponding to the nodes G and F can continue to be calculated (that is, when G and F are respectively used as the parent nodes of the K point, the corresponding path from the starting point is The path cost of the path to point K), after calculating the cost of the two paths, it is obtained that when the node F is used as the parent node of the intermediate node K, the cost of the corresponding candidate path is small. Therefore, the node F can be re-determined as K The parent node of the node, that is, the target neighbor node is F. After determining that the node F is the parent node of the node K, the corresponding wiring is shown in Figure 3b.
  • the node distribution information for the to-be-parked area is determined, the intermediate nodes are determined according to the node distribution information, and the neighbor nodes are searched based on the intermediate nodes to obtain a plurality of neighbor nodes. , respectively determine the cost information of the candidate paths corresponding to multiple neighbor nodes, determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node, which realizes the optimization of the automatic parking path and can be applied to complex
  • the parking scene is universal, that is, the search efficiency can be improved by searching the neighbor nodes, and the cost information of the path from the starting node to the intermediate node through the neighbor node is used to select the target neighbor node. Compared with the method in which the nearest neighbor of the intermediate node is used as the parent node, the cost of the overall path is fully considered, and then the parking path with the least cost can be selected for parking.
  • FIG. 4 a flowchart of steps of another method for automatic parking provided by an embodiment of the present invention is shown, which may specifically include the following steps:
  • Step 401 in the process of automatic parking, determine the node distribution information for the to-be-parked area; wherein, the node distribution information includes a plurality of nodes that the vehicle can pass through;
  • Step 402 obtaining parking feature information
  • the parking feature information includes any one or more of the following:
  • the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle is the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle.
  • Step 403 using the parking feature information and the node distribution information to determine a sampling area
  • the parking feature information and the node distribution information may be used to determine the sampling area.
  • the sampling area can be applied to different types of parking environments and has universality.
  • it can be applied to rearward parking in vertical parking spaces, forward parking in vertical parking spaces, parallel parking spaces, and inclined parking spaces.
  • the sampling region can be trained offline to obtain a priority sampling region.
  • a sampling area of a preset range can be set (the area of the preset range is larger than the area of the daily sampling area, and the difference can be set to the preset value); in the pre-built parking scene library, run the RRT star algorithm, Obtain the planning result; record the valid sampling points of the planning result and the parking feature information; take the parking feature information as the input and the valid sampling points as the output, and train the classification tree containing the mapping relationship between the two.
  • the sampling area can be preferentially selected through the classification tree according to the parking feature information and node distribution information, so as to avoid a large number of invalid search nodes, improve the search efficiency, and reduce the planning time.
  • the increase of the set scale can make the path planning of parking more reasonable and intelligent;
  • sampling areas can be selected according to different parking space types: the shaded part in Figure 1a represents the sampling area of vertical parking spaces, and the shaded part in Figure 2a is the sampling area of parallel parking spaces.
  • Step 404 in the sampling area, determine an intermediate node
  • the parking pose (x, y, ⁇ ) is sampled, and a node is randomly sampled in the sampling area, and the node is determined as an intermediate node.
  • Step 405 performing a neighbor node search based on the intermediate node to obtain multiple candidate nodes
  • adjacent nodes can be searched within a preset range to obtain multiple candidate nodes.
  • the candidate nodes that can be obtained are G, F, and E.
  • Step 406 from the plurality of candidate nodes, determine that a node that does not have an obstacle with the intermediate node is a neighbor node;
  • the candidate nodes After the candidate nodes are obtained from the search, not every candidate node can be successfully connected to the intermediate node due to the existence of obstacles. Therefore, the candidate nodes can be screened to remove the unconnectable candidate nodes, and finally the intermediate nodes and adjacent nodes can be connected. Nodes with no obstacles between them are determined as neighbor nodes.
  • the candidate nodes of the intermediate node K are G, F, and E.
  • the node E and the node K cannot be successfully connected due to the existence of obstacles. Therefore, they cannot become the neighbor nodes of the intermediate node K, and finally
  • the determined neighbor nodes are G and F.
  • Step 407 Determine the cost information of the candidate paths corresponding to the plurality of neighbor nodes respectively; wherein, the candidate parking path is a path from the starting node to the intermediate node via the neighbor node;
  • Step 408 Determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node.
  • node distribution information for the to-be-parked area may be determined to obtain parking feature information; the parking feature information and the node distribution information may be used to determine the sampling area , in the sampling area, determine an intermediate node, perform a neighbor node search based on the intermediate node, obtain multiple candidate nodes, and determine from the multiple candidate nodes that there is no obstacle between the intermediate node and the intermediate node
  • the node is a neighbor node, and the cost information of the candidate paths corresponding to multiple neighbor nodes is determined respectively, and the target neighbor node can be determined according to the cost information, and automatic parking is performed according to the target neighbor node, and the sampling area is determined by the parking feature information; Avoid a large number of invalid search nodes, improve search efficiency, reduce planning time, and sample in different sampling areas for different parking characteristics, which can be applied to various parking scenarios and improve universality.
  • the cost confidence determines the target neighbor nodes, and can plan a parking path with a small path cost.
  • the starting pose is used as the starting node to start planning the parking path, and the maximum number of descending generations is set.
  • the number of iterations the number of iterations + 1, the sampling area is set according to the type of parking space, and the new node md_node (ie, the intermediate node) is randomly sampled;
  • the parking feature information may include the type of parking space, and then a sampling area may be set according to the type of parking space, and intermediate nodes may be randomly sampled in the sampling area
  • a preset function may be used to connect the intermediate node and the nearest neighbor node.
  • the connection is successful, and go to step 7; when there is an obstacle between the intermediate node and the nearest neighbor node, the connection cannot be made, and the center node is randomly sampled again.
  • the target neighbor node after determining the target neighbor node, it can be further determined whether there is an obstacle between the intermediate node and the target node, and when there is no obstacle between the intermediate node and the target node, the intermediate node and the target node can be rewired, and can The parking path is determined according to the target path corresponding to the target adjacent node and the rewiring between the intermediate node and the target node, and then automatic parking is performed according to the parking path.
  • the solution set is a set of planned paths after performing multiple iterations.
  • FIG. 6 a schematic structural diagram of a vehicle provided by an embodiment of the present invention is shown, which may specifically include the following modules:
  • a node distribution determination module 601 configured to determine the node distribution information for the to-be-parked area in the process of automatic parking; wherein, the node distribution information includes a plurality of nodes that the vehicle can pass through;
  • An intermediate node determination module 602 configured to determine an intermediate node according to the node distribution information, and perform a neighbor node search based on the intermediate node to obtain a plurality of neighbor nodes;
  • a cost information determination module 603, configured to respectively determine cost information of candidate paths corresponding to the plurality of neighbor nodes; wherein, the candidate parking path is a path from an initial node to the intermediate node via the neighbor node;
  • the target neighbor node determination module 604 is configured to determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node.
  • the intermediate node determination module 602 includes:
  • the feature information acquisition sub-module is used to obtain the parking feature information
  • a sampling area determination sub-module configured to use the parking feature information and the node distribution information to determine the sampling area
  • the intermediate node determination submodule is configured to determine the intermediate node in the sampling area.
  • the intermediate node determination module 602 further includes:
  • a candidate node determination sub-module used for performing a neighbor node search based on the intermediate node to obtain a plurality of candidate nodes
  • the neighbor node determination submodule is configured to determine, from the plurality of candidate nodes, a node that does not have an obstacle between the intermediate node and the intermediate node as a neighbor node.
  • the cost information determination module 603 includes:
  • the path attribute information determination sub-module is used to determine the path attribute information including but not limited to the path length, obstacle distance, and vehicle commutation times for the candidate path corresponding to each neighboring node;
  • the cost information determination sub-module is configured to use the path attribute information to determine the cost information.
  • the target neighbor node determination module 604 includes:
  • a parking path generation sub-module for rewiring the target neighbor node and the intermediate node to obtain a parking path; wherein the parking path includes a target path corresponding to the target neighbor node;
  • the parking processing sub-module is used to perform automatic parking according to the parking path.
  • the vehicle further includes:
  • the function connection module is used for connecting any two nodes in the parking path by using a preset connection function; wherein, the connection function corresponds to a variety of connection modes.
  • the parking feature information may include any one or more of the following:
  • the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle is the relative relationship between the type of parking space, the width of the parking space, and the starting pose of the vehicle and the target pose of the vehicle.
  • the node distribution information for the to-be-parked area is determined, the intermediate nodes are determined according to the node distribution information, and the neighbor nodes are searched based on the intermediate nodes to obtain a plurality of neighbor nodes. , respectively determine the cost information of the candidate paths corresponding to multiple neighbor nodes, determine the target neighbor node according to the cost information, and perform automatic parking according to the target neighbor node, which realizes the optimization of the automatic parking path and can be applied to complex
  • the parking scene is universal, that is, the search efficiency can be improved by searching the neighbor nodes, and the cost information of the path from the starting node to the intermediate node through the neighbor node is used to select the target neighbor node. Compared with the method in which the nearest neighbor of the intermediate node is used as the parent node, the cost of the overall path is fully considered, and then the parking path with the least cost can be selected for parking.
  • An embodiment of the present invention also provides a vehicle, which may include a processor, a memory, and a computer program stored in the memory and capable of running on the processor.
  • a vehicle which may include a processor, a memory, and a computer program stored in the memory and capable of running on the processor.
  • An embodiment of the present invention also provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the above method for automatic parking is implemented.
  • embodiments of the present invention may be provided as a method, an apparatus, or a computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media having computer-usable program code embodied therein, including but not limited to disk storage, CD-ROM, optical storage, and the like.
  • Embodiments of the present invention are described with reference to flowcharts and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the present invention. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing terminal equipment to produce a machine that causes the instructions to be executed by the processor of the computer or other programmable data processing terminal equipment Means are created for implementing the functions specified in the flow or flows of the flowcharts and/or the blocks or blocks of the block diagrams.
  • These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal equipment to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising instruction means, the The instruction means implement the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

Landscapes

  • Engineering & Computer Science (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Traffic Control Systems (AREA)
  • Navigation (AREA)

Abstract

一种自动泊车的方法和车辆，自动泊车的方法包括：在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，节点分布信息包括多个车辆可途经的节点；根据节点分布信息，确定中间节点，并基于中间节点进行近邻节点搜索，得到多个近邻节点；分别确定多个近邻节点对应的候选路径的代价信息；其中，候选泊车路径为从起始节点经近邻节点至中间节点的路径；根据代价信息，确定目标近邻节点，并按照目标近邻节点，进行自动泊车。该方法实现了对自动泊车路径的优化，能够适用于复杂的泊车场景，具有普适性。

Description

一种自动泊车的方法和车辆 技术领域
本发明涉及泊车技术领域，特别是涉及一种自动泊车的方法和车辆。
背景技术
目前，多数的自动泊车采用基于几何的路径规划方法，该方法主要是根据经验进行设置规划。但是，当泊车场景中的障碍物较多时，基于几何规划方法所规划的路径由于无法绕开障碍物，容易导致规划失败。
为了避免规划失败，在采用基于几何的路径规划方法时，通常需要人工地设置泊车的换向点，而对于新手驾驶员，很难准确把控泊车的换向点。而且，基于几何的路径规划方法普适性差，只适用车位和过道都较宽的泊车场景，对于其他复杂场景不能实现高效规划路径。
发明内容
鉴于上述问题，提出了以便提供克服上述问题或者至少部分地解决上述问题的一种自动泊车的方法和车辆，包括：
一种自动泊车的方法，所述方法包括：
在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
可选地，所述按照所述目标近邻节点，进行自动泊车，包括：
采用所述目标近邻节点和所述中间节点，进行重新布线，得到泊车路径；其中，所述泊车路径包括所述目标近邻节点对应的目标路径；
按照所述泊车路径，进行自动泊车。
可选地，所述根据所述节点分布信息，确定中间节点，包括：
获取泊车特征信息；
采用所述泊车特征信息和所述节点分布信息，确定采样区域；
在所述采样区域中，确定中间节点。
可选地，所述基于所述中间节点进行近邻节点搜索，得到多个近邻节点，包括：
基于所述中间节点进行近邻节点搜索，得到多个候选节点；
从所述多个候选节点中，确定与所述中间节点之间不存在障碍物的节点为近邻节点。
可选地，所述分别确定所述多个近邻节点对应的候选路径的代价信息，包括：
针对每个近邻节点对应的候选路径，确定包括但不限于路径长度、障碍物距离、车辆换向次数的路径属性信息；
采用所述路径属性信息，确定代价信息。
可选地，还包括：
针对所述泊车路径中的任一两个节点，采用预置的连接函数进行连接；其中，所述连接函数对应有多种连接方式。
可选地，所述泊车特征信息包括以下任一项或多项：
车位类型、车位宽度、车辆起始位姿与车辆目标位姿的相对关系。
一种车辆，所述车辆包括：
节点分布确定模块，用于在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
中间节点确定模块，用于根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
代价信息确定模块，用于分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
目标近邻节点确定模块，用于根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
一种车辆，包括处理器、存储器及存储在所述存储器上并能够在所述处理器上运行的计算机程序，所述计算机程序被所述处理器执行时实现如上所述的自动泊车的方法。
一种计算机可读存储介质，所述计算机可读存储介质上存储计算机程序，所述计算机程序被处理器执行时实现如上所述的自动泊车的方法。
本发明实施例具有以下优点：
在本发明实施例中，通过在自动泊车的过程中，确定针对待泊车区域的节点分布信息，根据节点分布信息，确定中间节点，并基于中间节点进行近邻节点搜索，得到多个近邻节点，分别确定多个近邻节点对应的候选路径的代价信息，根据代价信息，确定目标近邻节点，并按照目标近邻节点，进行自动泊车，实现了对自动泊车路径的优化，能够适用于复杂的泊车场景，且具有普适性，即通过近邻节点搜索，可以提高搜索效率，又采用了从起始节点经近邻节点至中间节点的路径的代价信息来选择目标近邻节点的方式，与直接将与中间节点最近的邻节点作为中间节点的父节点的方式相比，充分考虑的整体路径的代价，进而可以选择代价最小的泊车路径进行泊车。
附图说明
为了更清楚地说明本发明的技术方案，下面将对本发明的描述中所需要使用的附图作简单地介绍， 显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1a是本发明一实施例提供的一种垂直车位采样区域的示意图；
图1b是本发明一实施例提供的一种平行车位采样区域的示意图；
图2是本发明一实施例提供的一种自动泊车的方法的步骤流程图；
图3a是本发明一实施例提供的一种确定目标近邻节点的示意图；
图3b是本发明一实施例提供的一种确定目标近邻节点后的示意图；
图3c是本发明一实施例提供的一种近邻节点重新布线的示意图；
图3d是本发明一实施例提供的一种近邻节点重新布线后的示意图；
图4是本发明一实施例提供的另一种自动泊车的方法的步骤流程图；
图5是本发明一实施例提供的一种规划最小代价路径的流程示意图；
图6是本发明一实施例提供的一种车辆的结构示意图。
具体实施方式
为使本发明的上述目的、特征和优点能够更加明显易懂，下面结合附图和具体实施方式对本发明作进一步详细的说明。显然，所描述的实施例是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
快速探索随机树(Rapidly Exploring Random Tree，RRT)算法是一种递增采样搜索算法，其优势在于通过在位形空间中连接离散的节点来产生最终的线路。RRT算法倾向于拓展到开放的未探索区域，当时间足够长，迭代次数足够多，没有不会被探索到的区域。
作为RRT算法的改进算法，RRTstar算法采用迭代深化思想对RRT算法得到的规划结果进行优化，其重要的思想在于近邻节点搜索以及重新布线。
在本发明实施例中，根据泊车场景对RRTstar算法做出以下的改进：
1)采样区域的选择：根据车位的大小，车位的类型(可以包括垂直车位，平行车位，斜列车位)，障碍物的位置选择RRT的采样区域。例如，如图1a中的阴影部分表示垂直车位的采样区域；如图2a中的阴影部分为平行车位的采样区域。
2)基础连接方式：对于采样区域中的任一点(即中间节点)，从车辆起始位置经由中间节点至目的位置，可以构建很多种路径。通过使用ReedsSheep曲线，Dubins曲线，直线-圆弧，圆弧-圆弧进行连接，可以保证最终成功规划的路径的多样性。
3)布线时路径代价计算：在实际应用中，可以将两节点之间的直线距离(欧式距离)作为两节点之间的路径代价，而在本发明一实施例中，可以综合考虑路径与障碍物的距离、路径的数量、路径的方向、路径的曲率计算路径代价，从而保证规划的路径安全，揉库次数少。例如，两个节点间的路 径代价可以通过下式进行计算：
路径代价＝a*路径长度-b*障碍物距离系数+c*换向次数
(Cost＝a*Path_Length-b*Distan_Obs+c*Turns_Num)
其中，a、b、c为大于0的常数。
作为一种示例，采用的规划算法的流程如下：
当输入车辆的起始位姿、目标位姿、和多个障碍物与车位框线段的横坐标(x)、纵坐标(y)、转向角参数(θ)时，例如，起始位姿可以用startpos(xs,ys,θs)表示，目标位姿可以用tarpos(xg,yg,θg)表示，多个障碍物与车位框线段可以分别通过(x11,y11,x12,y12),(x21,y21,x22,y22),......(xn1,yn1,xn2,yn2)来表示，在对输入的所有参数进行计算后，得到的输出结果为从起始位姿到目标位姿的无碰撞路径，得到无碰撞路径可以是直线，圆弧的组合。
以下对本发明实施例进行详细说明：
参照图2，示出了本发明一实施例提供的一种自动泊车的方法的步骤流程图，具体可以包括如下步骤：
步骤201，在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
在实际应用中，由于在泊车场景中通过RRT运算进行训练时，可以记录每一次运算得到的有效的节点，这些有效节点为车辆在泊车过程中可以途径的节点，所有有效节点可以构成节点分布信息，因而，在自动泊车的过程中，可以确定待泊车区域内的节点分布信息。
其中，每个节点都包含了节点对应的位姿(x,y,θ)，节点的代价，节点的索引，节点父节点的索引，节点父节点到本节点的路径信息，节点父节点为从起点到节点的路径中，该节点的上一节点。
通过获取节点分布信息，对有效节点进行处理，无需筛选无效节点，进而可以减少规划时间，提高搜索效率。
步骤202，根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
在确定节点分布信息后，可以根据节点分布信息来确定中间节点，中间节点为车辆从起始位置移动到目标位置过程中所经过的一个节点。通过获取的节点分布信息确定中间节点，可以避免人工设置转向点造成的转向点判断不准确的问题。
在一示例中，可以一次确定多个中间节点，然后针对确定的每个中间节点分别进行路径规划，得到一个或多个规划结果；也可以一次确定一个中间节点，通过设置迭代循环，在预设迭代次数内，在每个循环内确定中间节点，并针对每次确定的中间节点进行规划，得到一个或多个规划结果。
在选取中间节点之后，基于中间节点可以通过确定预设搜索范围，并在预设范围内进行近邻节点 搜索，得到多个近邻节点。采用邻近节点在预设范围内进行搜索，避免盲目搜索，有利于提高搜索效率。
例如，可以设置预设范围为R为半径的圆，其中R可以如下公式计算得到：
Figure PCTCN2020121566-appb-000001
R表示以中间节点为圆心进行近邻搜索时的预设搜索范围的圆的半径；
n node为已有的节点数目。
例如，如图3a，A至K为分布在待泊车区域内的有效节点，节点与节点之间连线上的数字为两节点之间的路径代价信息，将K节点确定的中间节点，Xnew表示K节点为一个新节点，在以R为半径的圆中，随K节进行近邻节点搜索，确定近邻节点为F、G。
步骤203，分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
在确定中间节点的多个近邻节点后，可以针对每个近邻节点，根据从起点经由该近邻节点到中间节点确定其对应的候选路径，进而可以计算每条候选路径的代价信息。
例如，如图3a，中间节点K在R范围内的近邻节点有G、F。对于节点G，从起点A经由节点G点到中间节点K点的候选路径为A-B-D-G-K，可以通过计算得到路径代价为2+5+6+2＝15；对于节点F，从A点经由节点F点到中间节点K点的候选路径为：A-B-C-F-K，可以通过计算得到路径代价2+1+4+3＝10。
在本发明一实施例中，所述分别确定所述多个近邻节点对应的候选路径的代价信息可以包括以下子步骤：
针对每个近邻节点对应的候选路径，确定包括但不限于路径长度、障碍物距离、车辆换向次数的路径属性信息；采用所述路径属性信息，确定代价信息。
在实际应用中，可以通过针对候选路径确定相应路径属性信息，进而采用路径属性信息，通过计算得到路径代价信息。
其中，路径属性信息可以包括但不限于路径长度、障碍物距离、车辆换向次数。
考虑到泊车场景的安全性，泊车路径的效率，可以通过下述公式计算路径代价：
路径代价＝a*路径长度-b*障碍物距离系数+c*换向次数
(Cost＝a*Path_Length-b*Distan_Obs+c*Turns_Num)
其中，a、b、c为大于0的常数。
由该公式可知，路径越长，代价越大；路径与障碍物的距离越大，代价越小；换向次数越多，代价越大。
步骤204，根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
在确定候选路径的代价信息后，可以根据代价信息，将所有候选路径中，代价最小的目标路径所对应的近邻节点确定为目标近邻节点，从起点经由目标近邻节点至终点的目标路径为最小代价的路径， 进而可以根据确定的目标近邻节点进行自动泊车。
由于确定的目标近邻节点是根据候选路径的代价确定的，可以选择路径代价最小的候选路径所对应的近邻节点作为目标近邻节点，最终可以按路径代价最小的泊车路径进行泊车。
例如，如图3a中，针对中间节点K的邻近节点G、F的候选路径A-B-D-G-K、A-B-C-F-K，对应计算代价信息分别为A-B-D-G-K的代价为15、A-B-C-F-K的代价为10，节点F对应的候选路径的代价小，则A-B-C-F-K为目标路径，节点F为目标近邻节点，图3b为节点F为目标近邻节点后的布线示意图，以节点F为目标近邻节点进行自动泊车。
在本发明一实施例中，所述按照所述目标近邻节点，进行自动泊车的步骤可以包括如下子步骤：
采用所述目标近邻节点和所述中间节点，进行重新布线，得到泊车路径；按照所述泊车路径，进行自动泊车。
其中，泊车路径可以包括目标近邻节点对应的目标路径。
在确定目标近邻节点之后，可以采用目标近邻节点和所述中间节点进行重新布线，根据重新布线的结果可以得到泊车路径，然后可以按照泊车路径进行自动泊车。
其中，重新布线可以是针对中间节点的邻近节点也可以是针对中间节点与目标节点进行重新布线。
在一示例中，在确定目标近邻节点后，还可以对中间节点的每个近邻节点进行重新布线。具体地，可以先确定从起点到中间节点的每个近邻节点的多条第二候选路径，针对每个近邻节点分别计算其对应的每条第二候选路径的路径代价信息，根据第二候选路径的路径代价信息确定是否需要重新布线，针对需要重新布线的近邻节点进行重新布线。
重布线过程的意义在于每当生成了中心节点(新节点)后，可以通过重新布线，使得某些节点的路径代价减少。如果以整体的眼光看，并不是每一个重新布线的节点都会出现在最终生成的路径中，但在生成随机树的过程中，每一次的重布线都尽可能的为最终路径代价减小创造机会。
例如，在图3c中，在为K点重新选择父节点F后。可以对K点的近邻节点F、G进行重新布线。具体重新布线方法如下：
针对邻节点F，从A到F的第二候选路径有两条：
Path1:A-B-C-F，计算得到路径代价为2+1+4＝7；
Path2:A-B-D-G-K-F，计算得到路径代价为2+5+6+2+3＝17。
在确定目标近邻节点后，可以确定目标路径为A-B-C-F-K，针对F的Path1的代价小于Path2，A-B-C-F为路径代价小的第二候选路径，该路径布线在目标路径中已存在，因此，不需要对F节点重新布线。
针对邻节点F，从A到G点的路径有两条：
Path1:A-B-D-G，计算得到路径代价为2+5+6＝13；
Path2:A-B-C-F-K-G，计算得到路径代价为2+1+4+3+2＝11；
在确定目标近邻节点后，可以确定目标路径为A-B-C-F-K，而针对G的Path1的代价大于Path2， A-B-C-F-K-G为路径代价小的第二候选路径，目前的布线中并没有K-G的布线，因此，G点需要重新布线，如图3d，示出了对G点重新布线后的布线图。
通过对近邻节点的重新布线，可以在每次迭代时，不断优化布线，更加智能的得到泊车路径。
在一示例中，在确定目标近邻节点后，可以进一步确定中间节点与目标节点之间是否存在障碍物，当中间节点与目标节点不存在障碍物时，可以对中间节点与目标节点进行重新布线，并可以根据目标邻近节点对应的目标路径以及中间节点与目标节点的重新布线确定泊车路径，进而按照泊车路径进行自动泊车。
当中间节点与目标节点存在障碍物时，执行确定中间节点的步骤。
在本发明一实施例中，该方法还可以包括如下步骤：
针对所述泊车路径中的任一两个节点，采用预置的连接函数进行连接；其中，所述连接函数对应有多种连接方式。
在对中间节点的重新布线后，可以通过预置的连接函数连接泊车路径中的任一两个节点，其中预置的连接函数对应有多种连接方式，
作为一种示例，连接方式可以包括以下一项或多项：
ReedsSheep曲线，直线与圆弧连接，双圆弧连接；
ReedsSheep曲线可以至少包含3段路径，直线-圆弧-直线，圆弧-直线-圆弧，圆弧-圆弧-圆弧。
通过如上预置函数进行连接时，节点间可以选择的连接方式众多，从而可以提高连接成功率，进而提高路径规划成功率；规划的路径也更符合人类泊车习惯；规划路径数量更少。
在一示例中，可以先确定中间节点的最近的近邻节点，即在预设范围内，可以与中间节点代价最小连接的最近的近邻节点，进而可以判断该节点与中间节点之间是否存在障碍物，当该节点与中间节点之间不存在障碍物，可以进一步通过搜索得到的一组邻节点，分别计算候选路径的代价信息，根据候选路径的代价信息确定目标近邻节点。
当该节点与中间节点之间存在障碍物时，执行确定中间节点的步骤。
例如，在图3a中，中间节点K在半径为R的圆的范围内，存在近邻节点G，F，其中，G-K的代价值为2、F-K的代价值为3，则可以根据两节点之间的代价，将节点G作为中间节点K最近的近邻节点，在布线图中，可以先将节点G作为中间节点K的父节点。
进一步的，由于节点G点与中间节点K点之间不存在障碍物，可以继续计算节点G、F对应的候选路径(即当G与F分别作为K点的父节点时，所对应的从起点到K点的路径)的路径代价，在计算两条路径的代价后，得到当F节点作为中间节点K的父节点时，对应的候选路径的代价小，因而，可以重新将F节点确定为K节点的父节点，即目标近邻节点为F，在确定F节点为K节点的父节点后，此时，对应的布线如图3b所示。
在本发明实施例中，通过在自动泊车的过程中，确定针对待泊车区域的节点分布信息，根据节点分布信息，确定中间节点，并基于中间节点进行近邻节点搜索，得到多个近邻节点，分别确定多个近 邻节点对应的候选路径的代价信息，根据代价信息，确定目标近邻节点，并按照目标近邻节点，进行自动泊车，实现了对自动泊车路径的优化，能够适用于复杂的泊车场景，且具有普适性，即通过近邻节点搜索，可以提高搜索效率，又采用了从起始节点经近邻节点至中间节点的路径的代价信息来选择目标近邻节点的方式，与直接将与中间节点最近的邻节点作为父节点的方式相比，充分考虑的整体路径的代价，进而可以选择代价最小的泊车路径进行泊车。
参照图4，示出了本发明一实施例提供的另一种自动泊车的方法的步骤流程图，具体可以包括如下步骤：
步骤401，在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
步骤402，获取泊车特征信息；
其中，所述泊车特征信息包括以下任一项或多项：
车位类型、车位宽度、车辆起始位姿与车辆目标位姿的相对关系。
步骤403，采用所述泊车特征信息和所述节点分布信息，确定采样区域；
在获取泊车特征信息后，可以采用所述泊车特征信息和所述节点分布信息，确定采样区域。
通过上述方式采样区域综合考虑泊车特征信息及节点分布信息，可以适用于不同类型的泊车环境，具有普适性。
例如，可以适用于垂直车位后向泊入，垂直车位前向泊车，平行车位，斜列车位。
在一示例中，可以对采样区域进行离线训练，得到一个优先采样区域。
具体的，可以设置一个预设范围的采样区域(预设范围的面积大于日常采样区域面积，其差值可以设置为预设值)；在预先构建的泊车场景库中，运行RRT star算法，得到规划结果；记录规划结果的有效采样点，泊车特征信息；将泊车特征信息作为输入，有效采样点作为输出，训练包含二者映射关系的分类树。
在实际泊车过程中，可以根据泊车特征信息和节点分布信息，通过分类树优先选择采样区域，从而可以避免大量无效的搜索节点，提高搜索效率，减少规划时间，并且，随着泊车数据集规模的增加，可以让泊车的路径规划更合理更智能；
例如：可以根据不同的车位类型选择不同采样的区域：如图1a中的阴影部分表示垂直车位的采样区域，如图2a中的阴影部分为平行车位的采样区域。
步骤404，在所述采样区域中，确定中间节点；
在确定采样区域后，在设定的采样区域中，采样得到的是泊车的位姿(x,y,θ)，在采样区内随机采样某一节点，并将该节点确定为中间节点。
步骤405，基于所述中间节点进行近邻节点搜索，得到多个候选节点；
在确定中间节点后，可以在预设范围内进行邻近节点搜索，得到多个候选节点。
例如，在图3a中，对中间节点K在半径为R的圆的范围内进行搜索后，可以得到的候选节点为G、F、E。
步骤406，从所述多个候选节点中，确定与所述中间节点之间不存在障碍物的节点为近邻节点；
在搜索得到候选节点后，由于存在障碍物，并不是每个候选节点均可以与中间节点成功连接，因此，可以对候选节点进行筛选，去掉不可连接的候选节点，最终可以将中间节点和邻近节点之间不存在障碍物的节点确定为近邻节点。
例如，如图3a中，中间节点K的候选节点为G、F、E，其中，节点E与节点K之间由于存在障碍物，无法成功连接，因此，不能成为中间节点K的近邻节点，最终确定的近邻节点为G、F。
步骤407，分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
步骤408，根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
在本发明实施例中，可以在自动泊车的过程中，确定针对待泊车区域的节点分布信息，获取泊车特征信息；采用所述泊车特征信息和所述节点分布信息，确定采样区域，在所述采样区域中，确定中间节点，基于所述中间节点进行近邻节点搜索，得到多个候选节点，从所述多个候选节点中，确定与所述中间节点之间不存在障碍物的节点为近邻节点，分别确定多个近邻节点对应的候选路径的代价信息，可以根据代价信息，确定目标近邻节点，并按照目标近邻节点，进行自动泊车，通过泊车特征信息确定采样区；可以避开大量无效的搜索节点，提高搜索效率，减少规划时间，针对不同泊车特征，在不同的采样区域进行采样，可以适用于各种泊车场景，提高了普适性，通过对候选路径的代价信心确定目标近邻节点，可以规划出路径代价小的泊车路径。
以下结合图5对本发明实施例进行示例性说明：
1、将起始位姿作为Tree的根节点，设置最大迭代次数N，迭代次数为0；
具体的，RRTstar算法中，以起始位姿作为起始节点，开始规划泊车路径，设置最大跌代次数。
2、判断迭代次数是否达到最大迭代次数。如果没有达到最大迭代次数，进行步骤3，如果达到迭代次数进行步骤，进行步骤11；
3、迭代次数＝迭代次数+1，根据车位类型设定采样区域，随机采样新节点md_node(即中间节点)；
具体地，在获取泊车特征信息后，泊车特征信息可以包括车位类型，进而可以根据车位类型设定采样区域，并在采样区域内随机采样中间节点
4、搜索md_node的最近邻节点nearst_node(即最近的近邻节点)；
5、使用基础连接曲线尝试连接md_node与nearst_node；
具体地，中间节点与最近的近邻节点之间可能存在障碍物，因此，可以在确定最近的邻近节点后，常试通过预设的函数连接中间节点与最近的近邻节点。
6、判断是否连接成功且无碰撞，如果连接成功且无碰撞，进行步骤7；否则，返回步骤2；
当中间节点与最近的近邻节点之间不存在障碍物时，连接成功，进行步骤7；当中间节点与最近的近邻节点之间存在障碍物时，无法连接，重新随机采样中心节点。
7、将md_node加入Tree中，以设定的半径R为圆，搜索md_node的最近一组节点near_nodes(即近邻节点)；
8、在near_nodes中，为md_node重新选择父节点(即目标近邻节点)；
9、为near_nodes重新布线；
10、尝试连接md_node与目标位姿(即目标节点)，如果规划成功且无碰撞，将path加入到解的集合中，返回步骤2；
具体地，在确定目标近邻节点后，可以进一步确定中间节点与目标节点之间是否存在障碍物，当中间节点与目标节点不存在障碍物时，可以对中间节点与目标节点进行重新布线，并可以根据目标邻近节点对应的目标路径以及中间节点与目标节点的重新布线确定泊车路径，进而按照泊车路径进行自动泊车。
11、判断解集合是否为空，如果解集合为空，则规划失败，如果解集合不为空，返回代价最小的路径。
具体地，解集合中为进行多次迭代后的所规划的路径集合。
需要说明的是，对于方法实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本发明实施例并不受所描述的动作顺序的限制，因为依据本发明实施例，某些步骤可以采用其他顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作并不一定是本发明实施例所必须的。
参照图6，示出了本发明一实施例提供的一种车辆的结构示意图，具体可以包括如下模块：
节点分布确定模块601，用于在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
中间节点确定模块602，用于根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
代价信息确定模块603，用于分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
目标近邻节点确定模块604，用于根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
在本发明一实施例中，中间节点确定模块602包括：
特征信息获取子模块，用于获取泊车特征信息；
采样区域确定子模块，用于采用所述泊车特征信息和所述节点分布信息，确定采样区域；
中间节点确定子模块，用于在所述采样区域中，确定中间节点。
在本发明一实施例中，中间节点确定模块602还包括：
候选节点确定子模块，用于基于所述中间节点进行近邻节点搜索，得到多个候选节点；
近邻节点确定子模块，用于从所述多个候选节点中，确定与所述中间节点之间不存在障碍物的节点为近邻节点。
在本发明一实施例中，代价信息确定模块603包括：
路径属性信息确定子模块，用于针对每个近邻节点对应的候选路径，确定包括但不限于路径长度、障碍物距离、车辆换向次数的路径属性信息；
代价信息确定子模块，用于采用所述路径属性信息，确定代价信息。
在本发明一实施例中，目标邻节点确定模块604包括：
泊车路径生成子模块，用于采用所述目标近邻节点和所述中间节点，进行重新布线，得到泊车路径；其中，所述泊车路径包括所述目标近邻节点对应的目标路径；
泊车处理子模块，用于按照所述泊车路径，进行自动泊车。
在本发明一实施例中，所述车辆还包括：
函数连接模块，用于针对所述泊车路径中的任一两个节点，采用预置的连接函数进行连接；其中，所述连接函数对应有多种连接方式。
在本发明一实施例中，泊车特征信息可以包括以下任一项或多项：
车位类型、车位宽度、车辆起始位姿与车辆目标位姿的相对关系。
在本发明实施例中，通过在自动泊车的过程中，确定针对待泊车区域的节点分布信息，根据节点分布信息，确定中间节点，并基于中间节点进行近邻节点搜索，得到多个近邻节点，分别确定多个近邻节点对应的候选路径的代价信息，根据代价信息，确定目标近邻节点，并按照目标近邻节点，进行自动泊车，实现了对自动泊车路径的优化，能够适用于复杂的泊车场景，且具有普适性，即通过近邻节点搜索，可以提高搜索效率，又采用了从起始节点经近邻节点至中间节点的路径的代价信息来选择目标近邻节点的方式，与直接将与中间节点最近的邻节点作为父节点的方式相比，充分考虑的整体路径的代价，进而可以选择代价最小的泊车路径进行泊车。
本发明一实施例还提供了一种车辆，可以包括处理器、存储器及存储在存储器上并能够在处理器上运行的计算机程序，计算机程序被处理器执行时实现如上自动泊车的方法。
本发明一实施例还提供了一种计算机可读存储介质，计算机可读存储介质上存储计算机程序，计算机程序被处理器执行时实现如上自动泊车的方法。
对于装置实施例而言，由于其与方法实施例基本相似，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
本说明书中的各个实施例均采用递进的方式描述，每个实施例重点说明的都是与其他实施例的不同之处，各个实施例之间相同相似的部分互相参见即可。
本领域内的技术人员应明白，本发明实施例可提供为方法、装置、或计算机程序产品。因此，本发明实施例可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明实施例可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明实施例是参照根据本发明实施例的方法、终端设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理终端设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理终端设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理终端设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理终端设备上，使得在计算机或其他可编程终端设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程终端设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
尽管已描述了本发明实施例的优选实施例，但本领域内的技术人员一旦得知了基本创造性概念，则可对这些实施例做出另外的变更和修改。所以，所附权利要求意欲解释为包括优选实施例以及落入本发明实施例范围的所有变更和修改。
最后，还需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者终端设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者终端设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者终端设备中还存在另外的相同要素。
以上对所提供的一种自动泊车的方法和车辆，进行了详细介绍，本文中应用了具体个例对本发明的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本发明的方法及其核心思想；同时，对于本领域的一般技术人员，依据本发明的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本发明的限制。

Claims (10)

  1. 一种自动泊车的方法，其特征在于，所述方法包括：
    在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
    根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
    分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
    根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
  2. 根据权利要求1所述的方法，其特征在于，所述按照所述目标近邻节点，进行自动泊车，包括：
    采用所述目标近邻节点和所述中间节点，进行重新布线，得到泊车路径；其中，所述泊车路径包括所述目标近邻节点对应的目标路径；
    按照所述泊车路径，进行自动泊车。
  3. 根据权利要求1或2所述的方法，其特征在于，所述根据所述节点分布信息，确定中间节点，包括：
    获取泊车特征信息；
    采用所述泊车特征信息和所述节点分布信息，确定采样区域；
    在所述采样区域中，确定中间节点。
  4. 根据权利要求1所述的方法，其特征在于，所述基于所述中间节点进行近邻节点搜索，得到多个近邻节点，包括：
    基于所述中间节点进行近邻节点搜索，得到多个候选节点；
    从所述多个候选节点中，确定与所述中间节点之间不存在障碍物的节点为近邻节点。
  5. 根据权利要求1或2所述的方法，其特征在于，所述分别确定所述多个近邻节点对应的候选路径的代价信息，包括：
    针对每个近邻节点对应的候选路径，确定包括但不限于路径长度、障碍物距离、车辆换向次数的路径属性信息；
    采用所述路径属性信息，确定代价信息。
  6. 根据权利要求1或2所述的方法，其特征在于，还包括：
    针对所述泊车路径中的任一两个节点，采用预置的连接函数进行连接；其中，所述连接函数对应有多种连接方式。
  7. 根据权利要求3所述的方法，其特征在于，所述泊车特征信息包括以下任一项或多项：
    车位类型、车位宽度、车辆起始位姿与车辆目标位姿的相对关系。
  8. 一种车辆，其特征在于，所述车辆包括：
    节点分布确定模块，用于在自动泊车的过程中，确定针对待泊车区域的节点分布信息；其中，所述节点分布信息包括多个车辆可途经的节点；
    中间节点确定模块，用于根据所述节点分布信息，确定中间节点，并基于所述中间节点进行近邻节点搜索，得到多个近邻节点；
    代价信息确定模块，用于分别确定所述多个近邻节点对应的候选路径的代价信息；其中，所述候选泊车路径为从起始节点经所述近邻节点至所述中间节点的路径；
    目标近邻节点确定模块，用于根据所述代价信息，确定目标近邻节点，并按照所述目标近邻节点，进行自动泊车。
  9. 一种车辆，其特征在于，包括处理器、存储器及存储在所述存储器上并能够在所述处理器上运行的计算机程序，所述计算机程序被所述处理器执行时实现如权利要求1至8中任一项所述的自动泊车的方法的步骤。
  10. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储计算机程序，所述计算机程序被处理器执行时实现如权利要求1至8中任一项所述的自动泊车的方法的步骤。
PCT/CN2020/121566 2020-07-10 2020-10-16 一种自动泊车的方法和车辆 WO2022007227A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP20930646.3A EP3971047B1 (en) 2020-07-10 2020-10-16 Automatic parking method and vehicle

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010664638.0A CN111824131B (zh) 2020-07-10 2020-07-10 一种自动泊车的方法和车辆
CN202010664638.0 2020-07-10

Publications (1)

Publication Number Publication Date
WO2022007227A1 true WO2022007227A1 (zh) 2022-01-13

Family

ID=72901333

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/121566 WO2022007227A1 (zh) 2020-07-10 2020-10-16 一种自动泊车的方法和车辆

Country Status (3)

Country Link
EP (1) EP3971047B1 (zh)
CN (1) CN111824131B (zh)
WO (1) WO2022007227A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112550280B (zh) * 2020-12-21 2022-05-13 广州小鹏自动驾驶科技有限公司 一种车辆泊出的方法和装置
CN115410409A (zh) * 2022-09-30 2022-11-29 浙江吉利控股集团有限公司 一种选择停车位的方法、系统及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150025732A1 (en) * 2013-07-17 2015-01-22 Electronics And Telecommunications Research Institute Apparatus and method for controlling automatic parking
CN109733384A (zh) * 2018-12-25 2019-05-10 科大讯飞股份有限公司 泊车路径设置方法及系统
US10365110B2 (en) * 2014-09-30 2019-07-30 Nec Corporation Method and system for determining a path of an object for moving from a starting state to an end state set avoiding one or more obstacles
CN110139794A (zh) * 2017-01-10 2019-08-16 三菱电机株式会社 用于停放车辆的系统、方法及非暂时计算机可读存储介质
CN110703768A (zh) * 2019-11-08 2020-01-17 福州大学 一种改进型动态rrt*的移动机器人运动规划方法
CN111307152A (zh) * 2020-02-18 2020-06-19 中国科学院合肥物质科学研究院 一种自主泊车路径反向生成规划方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10012984B2 (en) * 2015-12-14 2018-07-03 Mitsubishi Electric Research Laboratories, Inc. System and method for controlling autonomous vehicles
US11332141B2 (en) * 2017-06-20 2022-05-17 Mitsubishi Electric Corporation Path estimation device and path estimation method
US11117569B2 (en) * 2018-06-27 2021-09-14 Baidu Usa Llc Planning parking trajectory generation for self-driving vehicles using optimization method
US10809732B2 (en) * 2018-09-25 2020-10-20 Mitsubishi Electric Research Laboratories, Inc. Deterministic path planning for controlling vehicle movement
CN109131318B (zh) * 2018-10-19 2020-03-27 清华大学 一种基于拓扑地图的自主泊车路径协调方法
CN111098852B (zh) * 2019-12-02 2021-03-12 北京交通大学 一种基于强化学习的泊车路径规划方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150025732A1 (en) * 2013-07-17 2015-01-22 Electronics And Telecommunications Research Institute Apparatus and method for controlling automatic parking
US10365110B2 (en) * 2014-09-30 2019-07-30 Nec Corporation Method and system for determining a path of an object for moving from a starting state to an end state set avoiding one or more obstacles
CN110139794A (zh) * 2017-01-10 2019-08-16 三菱电机株式会社 用于停放车辆的系统、方法及非暂时计算机可读存储介质
CN109733384A (zh) * 2018-12-25 2019-05-10 科大讯飞股份有限公司 泊车路径设置方法及系统
CN110703768A (zh) * 2019-11-08 2020-01-17 福州大学 一种改进型动态rrt*的移动机器人运动规划方法
CN111307152A (zh) * 2020-02-18 2020-06-19 中国科学院合肥物质科学研究院 一种自主泊车路径反向生成规划方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
KLEMM SEBASTIAN; OBERLANDER JAN; HERMANN ANDREAS; ROENNAU ARNE; SCHAMM THOMAS; ZOLLNER J. MARIUS; DILLMANN RUDIGER: "RRT∗-Connect: Faster, asymptotically optimal motion planning", 2015 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND BIOMIMETICS (ROBIO), IEEE, 6 December 2015 (2015-12-06), pages 1670 - 1677, XP032873296, DOI: 10.1109/ROBIO.2015.7419012 *
ZHENG KAIYU; LIU SHAN: "RRT based Path Planning for Autonomous Parking of Vehicle", 2018 IEEE 7TH DATA DRIVEN CONTROL AND LEARNING SYSTEMS CONFERENCE (DDCLS), IEEE, 25 May 2018 (2018-05-25), pages 627 - 632, XP033432461, DOI: 10.1109/DDCLS.2018.8515940 *

Also Published As

Publication number Publication date
EP3971047B1 (en) 2023-10-04
EP3971047C0 (en) 2023-10-04
CN111824131B (zh) 2021-10-12
EP3971047A4 (en) 2022-06-22
EP3971047A1 (en) 2022-03-23
CN111824131A (zh) 2020-10-27

Similar Documents

Publication Publication Date Title
CN109115226B (zh) 基于跳点搜索的多机器人冲突避免的路径规划方法
CN111678523B (zh) 一种基于star算法优化的快速bi_rrt避障轨迹规划方法
CN108549385B (zh) 一种结合a*算法和vfh避障算法的机器人动态路径规划方法
CN110285802B (zh) 快速扩展随机树路径光滑方法
CN106444740B (zh) 基于mb-rrt的无人机二维航迹规划方法
WO2022007227A1 (zh) 一种自动泊车的方法和车辆
CN112904869B (zh) 基于多树rrt的无人艇加权迭代路径规划方法及设备
KR102242653B1 (ko) 지상 로봇의 움직임을 고려한 강인한 레이저 스캐너 매칭 방법 및 그 장치
US9443334B2 (en) Routing edges with ordered bundles
CN113449910B (zh) 一种基于顺序存储二叉树的航线自动生成方法
CN113867336B (zh) 一种适用于复杂场景下移动机器人路径导航及规划方法
CN113124875A (zh) 一种路径导航方法
CN114527761A (zh) 一种基于融合算法的智能汽车局部路径规划方法
CN116700268A (zh) 一种改进人工势场法引导的双向rrt*路径规划方法
CN115167398A (zh) 一种基于改进a星算法的无人船路径规划方法
CN112197783B (zh) 一种考虑车头指向的两阶段多抽样的rrt路径规划方法
CN109855640B (zh) 一种基于自由空间与人工蜂群算法的路径规划方法
Wang et al. Target-biased informed trees: Sampling-based method for optimal motion planning in complex environments
CN117451070A (zh) 基于改进rrt*算法的光伏机器人规划算法及规划系统
CN116560360A (zh) 面向复杂动态场景的医用护理机器人实时动态路径规划方法及系统
KR101585374B1 (ko) 유전자 알고리즘을 이용한 로봇 경로 탐색 장치 및 방법
CN115933749A (zh) 一种面向设备目标驱动的变电站巡检航线生成方法
CN109855622A (zh) 用于移动机器人的路径搜索方法及设备
WO2023137938A1 (zh) 虚拟角色的路径规划方法及装置、电子设备、存储介质
Wang et al. Research on path planning of mobile robot based on improved jump point search algorithm

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2020930646

Country of ref document: EP

Effective date: 20211018

NENP Non-entry into the national phase

Ref country code: DE