Autonomous parking track planning method and system for irregular road obstacles

Abstract

The invention relates to the technical field of automatic driving, in particular to an autonomous parking track planning method and system for irregular road obstacles, and the method comprises the steps: constructing an obstacle map for a to-be-planned region; pre-searching the obstacle map according to the position information generated by the user side to obtain a pre-search track; performing trajectory expansion along the pre-search trajectory to form a trajectory planning corridor; and planning in the trajectory planning corridor according to the kinematic model of the vehicle to obtain an actual driving trajectory. In order to solve the problem that in the prior art, the planning process of automatic parking in a multi-obstacle scene is relatively complex, a rough pre-search track is obtained through pre-sampling and searching, then a track planning corridor is generated based on the pre-search track, and track planning is performed in corridor constraints. Therefore, the vehicle trajectory planning process is separated from irregular obstacles in the environment, the problem that the obstacle constraint in the unstructured road cannot be solved in the nonlinear solving process is solved, and the scale of the optimization problem is simplified.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method and system for autonomous parking trajectory planning for irregular road obstacles. Background Art

[0002] The Automated Valet Parking (AVP) system is used to achieve fully autonomous vehicle driving in a specific area from the parking lot entrance / exit to the parking space. Due to its low driving speed and the ability to ensure maximum safety in a specific operating site, it is recognized by the industry as the first L4 autonomous driving scenario to be implemented. Normally, the parking function requires the user to get off the vehicle at a designated drop-off point and issue a parking command through a mobile phone APP. After receiving the command, the vehicle can automatically drive to the parking space in the parking lot without the need for user operation and monitoring.

[0003] In the prior art, there are many technical solutions for realizing automatic valet parking. For example, Chinese patent CN202011432970.0 discloses an automatic parking system, which is configured to automatically move the autonomous driving vehicle from the parking space to the boarding lot based on the parking space departure time determined in a manner so that the autonomous driving vehicle can arrive at the boarding lot at a preset scheduled time for leaving the warehouse, and the automatic parking system includes: a connection control time setting unit, which is configured to set a connection control time for controlling the state of the autonomous driving vehicle to a power-on state based on the parking space departure time; and a power-on control unit, which is configured to control the state of the autonomous driving vehicle to the power-on state based on the set connection control time. The connection control time setting unit is configured to set the connection control time in a manner so that the autonomous driving vehicle becomes the power-on state before the parking space departure time.

[0004] For another example, Chinese patent CN202010151376.8 discloses an autonomous valet parking method, including the following steps: S1, when a vehicle enters the entrance drop-off area of ​​a parking lot, the backend server unit monitors whether the parking spaces in the parking lot are available through the parking lot monitoring unit, and sends the nearest target parking space information to the vehicle in real time, as well as the driving trajectory information and the corresponding vehicle actuator command information; S2, in the parking lot, the vehicle drives to the target parking space according to the driving trajectory sent by the backend server unit, and executes the vehicle actuator command corresponding to the driving trajectory information in real time; S3, when the vehicle arrives at the parking task point near the target parking space, the vehicle's original automatic parking assistance function is executed to park, and the corresponding vehicle actuator command is executed to complete the parking action. By reusing the existing parking assistance function, the modified parking lot is suitable for any vehicle with parking assistance function, reducing the cost per vehicle.

[0005] However, in the actual implementation process, the inventors found that in some parking scenarios, there will be a large number of unevenly distributed polygonal obstacles in the static site, including concave polygonal obstacles and convex polygonal obstacles. Since the vehicle needs to be collision-free at any time during driving, it is necessary to consider the avoidance of polygonal obstacles in trajectory planning. Since concave polygonal obstacles are prone to deadlock or unsolvable situations due to the vehicle's own kinematic model during the driving trajectory planning process, it is also necessary to split such obstacles during planning to form two adjacent convex obstacles. When the number of obstacles in the site is large, the demand for computing power also increases. Summary of the invention

[0006] In view of the above problems existing in the prior art, a method and system for autonomous parking trajectory planning for irregular road obstacles are provided.

[0007] The specific technical solutions are as follows:

[0008] An autonomous parking trajectory planning method for irregular road obstacles, comprising:

[0009] Step S1: Scan the planned area to obtain multiple obstacle projections and construct an obstacle map;

[0010] Step S2: Pre-search the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory;

[0011] The position information includes a vehicle starting position and a vehicle target position;

[0012] Step S3: performing trajectory expansion along the pre-search trajectory to form a trajectory planning corridor;

[0013] The boundary of the trajectory planning corridor offsets the obstacle projection;

[0014] Step S4: obtaining an actual driving trajectory in the trajectory planning corridor according to the kinematic model of the vehicle.

[0015] On the other hand, the step S1 comprises:

[0016] Step S11: acquiring digital map information associated with the area to be planned from an external data source;

[0017] Step S12: parsing the digital map information to obtain a driving area and obstacle information associated with the driving area;

[0018] Step S13: resampling the obstacle information to obtain the obstacle projection;

[0019] Step S14: generating the obstacle map according to the driving area and the obstacle projection.

[0020] On the other hand, the step S2 comprises:

[0021] Step S21: receiving the position information from the user side, marking the vehicle starting position and the vehicle target position on the obstacle map, and performing gridding processing on the obstacle map to obtain a gridded map;

[0022] Step S22: searching adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position respectively, and recording first search information constructed from the vehicle starting position and second search information from the vehicle target position, until a repeated grid node is found as a search end point;

[0023] Step S23: generating a first half track according to the search end point and the vehicle starting position, and generating a second half track according to the search end point and the vehicle target position;

[0024] Step S24: splicing the first half trajectory and the second half trajectory to obtain the pre-search trajectory.

[0025] On the other hand, the step S3 comprises:

[0026] Step S31: generating an equivalent projection according to vehicle parameters of the vehicle;

[0027] The equivalent projection includes two circular projections;

[0028] The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle respectively;

[0029] The diameter of the circular projection is the wheelbase of the vehicle;

[0030] Step S32: performing projection expansion on the obstacle projection associated with the pre-search trajectory in the obstacle map to form an expanded projection map;

[0031] Step S33: in the expanded projection map, respectively setting the equivalent projection for each grid node of the pre-search trajectory;

[0032] Step S34: performing edge expansion on each of the equivalent projections respectively until the edge expansion stops after the edge expansion reaches the obstacle projection, and recording the vehicle expansion projection;

[0033] Step S35: generating the trajectory planning corridor according to all the vehicle expansion projections.

[0034] On the other hand, the step S4 comprises:

[0035] Step S41: setting a search pointer from the vehicle starting position;

[0036] Step S42: Filtering the next travel point in the trajectory planning corridor on the search pointer according to the kinematic model;

[0037] Step S43: setting the search pointers at the travel points respectively, and then returning to step S42 until the travel points coincide with the vehicle target position;

[0038] Step S44: Optimizing the trajectory according to the travel points to obtain the actual driving trajectory.

[0039] An autonomous parking trajectory planning system for irregular road obstacles, used to implement the above autonomous parking trajectory planning method;

[0040] The autonomous parking trajectory planning system comprises:

[0041] A map modeling module, wherein the map modeling module scans the area to be planned to obtain multiple obstacle projections and construct an obstacle map;

[0042] A pre-search module, the pre-search module is connected to the map modeling module, and the pre-search module pre-searches the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory;

[0043] The position information includes a vehicle starting position and a vehicle target position;

[0044] A corridor laying module, the corridor laying module is connected to the pre-search module, and the corridor laying module performs trajectory expansion along the pre-search trajectory to form a trajectory planning corridor;

[0045] The boundary of the trajectory planning corridor offsets the obstacle projection;

[0046] A trajectory generation module is connected to the corridor laying module, and the trajectory generation module obtains an actual driving trajectory in the trajectory planning corridor according to the kinematic model planning of the vehicle.

[0047] On the other hand, the map modeling module includes:

[0048] A map acquisition module, wherein the map acquisition module acquires digital map information associated with the area to be planned from an external data source;

[0049] A map parsing module, the map parsing module is connected to the map acquisition module, and the map parsing module parses the digital map information to obtain a driving area and obstacle information associated with the driving area;

[0050] A resampling module, the resampling module is connected to the map parsing module, and the resampling module resamples the obstacle information to obtain the obstacle projection;

[0051] A map generation module is connected to the resampling module, and the map generation module generates the obstacle map according to the driving area and the obstacle projection.

[0052] On the other hand, the pre-search module comprises:

[0053] A map processing module, wherein the map processing module receives the position information from the user side, marks the vehicle starting position and the vehicle target position on the obstacle map, and performs grid processing on the obstacle map to obtain a grid map;

[0054] A bidirectional search module, the bidirectional search module is connected to the map processing module, the bidirectional search module searches for adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position, respectively, and records first search information constructed from the vehicle starting position and second search information from the vehicle target position, until a repeated grid node is found as a search end point;

[0055] A trajectory generation module, the trajectory generation module is connected to the bidirectional search module, the trajectory generation module generates a first half trajectory according to the search end point and the vehicle starting position, and generates a second half trajectory according to the search end point and the vehicle target position;

[0056] A trajectory splicing module is connected to the trajectory generating module, and the trajectory splicing module splices the first half trajectory and the second half trajectory to obtain the pre-search trajectory.

[0057] On the other hand, the corridor paving module includes:

[0058] An equivalent projection generation module, wherein the equivalent projection generation module generates an equivalent projection according to vehicle parameters of the vehicle;

[0059] The equivalent projection includes two circular projections;

[0060] The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle respectively;

[0061] The diameter of the circular projection is the wheelbase of the vehicle;

[0062] A projection expansion module, the projection expansion module is connected to the equivalent projection generation module, and the projection expansion module performs projection expansion on the obstacle projection associated with the pre-search trajectory in the obstacle map to form an expanded projection map;

[0063] An equivalent projection setting module, the equivalent projection setting module is connected to the projection expansion module, and the equivalent projection setting module sets the equivalent projection for each grid node of the pre-search trajectory in the expansion projection map;

[0064] An edge expansion module, the edge expansion module is connected to the equivalent projection setting module, and the edge expansion module performs edge expansion on each of the equivalent projections, stops after contacting the obstacle projection, and records the vehicle expansion projection;

[0065] A corridor superposition module is connected to the edge expansion module, and the corridor superposition module generates the trajectory planning corridor according to all the vehicle expansion projections.

[0066] On the other hand, the trajectory generation module includes:

[0067] A starting pointer placement module, the starting pointer placement module sets a search pointer from the starting position of the vehicle;

[0068] A kinematics screening module, the kinematics screening module is connected to the start pointer placement module, and the kinematics screening module screens the next travel point in the trajectory planning corridor on the search pointer according to the kinematic model;

[0069] A pointer search module, the pointer search module is connected to the kinematics screening module, and the pointer search module respectively sets the search pointer on the travel point until the travel point coincides with the vehicle target position;

[0070] A trajectory optimization module is connected to the pointer search module, and the trajectory optimization module performs trajectory optimization according to the travel points to obtain the actual driving trajectory.

[0071] The above technical solution has the following advantages or beneficial effects:

[0072] In view of the problem that the planning process of automatic parking in multi-obstacle scenarios in the prior art is relatively complex and requires high computing power, in this solution, a rough pre-search trajectory is obtained through pre-sampling and searching, and then a trajectory planning corridor is generated based on the pre-search trajectory. Trajectory planning is performed within the corridor constraints, thereby separating the vehicle trajectory planning process from the irregular obstacles in the environment, solving the problem that obstacle constraints in unstructured roads cannot be solved in the process of nonlinear solution, and simplifying the scale of the optimization problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The embodiments of the present invention will be described more fully with reference to the attached drawings, which are provided for illustration and description only and are not intended to limit the scope of the present invention.

[0074] Figure 1 is an overall schematic diagram of an embodiment of the present invention;

[0075] Figure 2 This is a schematic diagram of step S1 in an embodiment of the present invention;

[0076] Figure 3 This is a schematic diagram of step S2 in an embodiment of the present invention;

[0077] Figure 4 This is a schematic diagram of step S3 in an embodiment of the present invention;

[0078] Figure 5 This is a schematic diagram of step S4 in an embodiment of the present invention;

[0079] Figure 6 A schematic diagram of a system in an embodiment of the present invention;

[0080] Figure 7 This is a schematic diagram of a map modeling module in an embodiment of the present invention;

[0081] Figure 8 This is a schematic diagram of a pre-search module in an embodiment of the present invention;

[0082] Fig. 9 This is a schematic diagram of a corridor paving module in an embodiment of the present invention;

[0083] Fig.10 Schematic diagram of a trajectory generation module in an embodiment of the present invention. DETAILED DESCRIPTION

[0084] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0085] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0086] Some of the blocks shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0087] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0088] The present invention comprises:

[0089] An autonomous parking trajectory planning method for irregular road obstacles, such as Figure 1 As shown, including:

[0090] Step S1: Scan the planned area to obtain multiple obstacle projections and construct an obstacle map;

[0091] Step S2: Pre-search the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory;

[0092] The position information includes the vehicle starting position and the vehicle target position;

[0093] Step S3: performing trajectory expansion along the pre-search trajectory to form a trajectory planning corridor;

[0094] The boundaries of the trajectory planning corridor are offset by the obstacle projection;

[0095] Step S4: obtaining the actual driving trajectory in the trajectory planning corridor according to the kinematic model of the vehicle.

[0096] Specifically, in view of the problem that the planning process of automatic parking in multi-obstacle scenarios in the prior art is relatively complex and requires high computing power, in this solution, a rough pre-search trajectory is obtained through pre-sampling and searching, and a trajectory planning corridor is generated based on the pre-search trajectory. Trajectory planning is performed within the corridor constraints, thereby separating the vehicle trajectory planning process from obstacles in the environment, solving the problem that obstacle constraints in unstructured roads cannot be solved in the process of nonlinear solution, and simplifying the scale of the optimization problem.

[0097] During implementation, the above autonomous parking trajectory planning method is mainly configured as a software embodiment in a computer system, such as an automatic driving platform of a vehicle, for trajectory planning of a vehicle with an autonomous valet parking function.

[0098] Among them, the area to be planned refers to the physical area where vehicles need to park autonomously. It is usually planned with corresponding parking spaces and roads for vehicles to drive on, but it is not necessarily a common parking lot with a rectangular distribution. It may be an open space or an area with curved roads according to specific design requirements.

[0099] In this area, there are many unspecified obstacles, such as landscape objects, trash cans, monitoring systems, load-bearing structures, etc. placed in the venue. According to the needs, they may appear as irregular polygons in the projection direction, including convex polygons and concave polygons.

[0100] Since trajectory planning is performed for the vehicle, the obstacle is equivalent to a two-dimensional projection in the parking lot coordinate system during the processing, and the projection contour depends on the outer edge of the obstacle in the top-down direction.

[0101] The obstacle map is a map formed by collecting, measuring and aligning the digital map and obstacle information to the same coordinate system. Subsequent operations such as trajectory planning are based on the obstacle map.

[0102] The location information is obtained through the vehicle-side computer input. The vehicle can determine the vehicle's starting position based on a positioning system, such as a GPS positioning system, and map it to the coordinate system of the obstacle map. Users usually input corresponding instructions through the vehicle computer to indicate the target position of the vehicle to be parked.

[0103] The pre-search trajectory is designed based on the obstacle map, and the trajectory line that can connect the vehicle's starting position and the vehicle's target position in the obstacle map is found through sampling and searching.

[0104] Since this step does not take into account issues such as obstacle contour optimization, vehicle-obstacle collision and vehicle kinematic feasibility, the pre-search trajectory is usually not sufficient for directly controlling the parking process, and its computing power requirements are correspondingly reduced.

[0105] The trajectory planning corridor is an irregular polygonal corridor formed on the basis of the pre-search trajectory. It has a certain width, and the corridor projection will not overlap with obstacles.

[0106] Based on the trajectory planning corridor, the existing trajectory optimization problem based on obstacle contours can be converted into a driving process on a structured road, thereby effectively reducing the amount of calculation for vehicle trajectory planning.

[0107] In one embodiment, Figure 2 As shown, step S1 includes:

[0108] Step S11: acquiring digital map information associated with the area to be planned from an external data source;

[0109] Step S12: parsing the digital map information to obtain the driving area and obstacle information associated with the driving area;

[0110] Step S13: Resample the obstacle information to obtain obstacle projection;

[0111] Step S14: Generate an obstacle map according to the driving area and obstacle projection.

[0112] Specifically, in order to achieve a better map construction effect, in this embodiment, digital map information associated with the area to be planned is firstly obtained from an external data source.

[0113] For example, the external data source may be a parking system or a provider of a digital map that marks the locations of various structures of the building, including load-bearing structures, decorative structures, fire-fighting and ventilation facilities, etc.

[0114] In some embodiments, an external monitoring system may be introduced to perform real-time three-dimensional modeling of the site to form a digital twin model corresponding to the parking lot, and the available driving area and irregular obstacles may be identified in the self-modeling to form complete digital map information.

[0115] Subsequently, the digital map information is parsed, including two-dimensional processing, removal of redundant annotations, and alignment to a coordinate system, so that only the driving area and obstacle information associated with the driving area are retained.

[0116] On this basis, the obstacle information is resampled to obtain the obstacle projection, which mainly includes removing other descriptive information of the obstacle and obtaining the maximum outer contour of the obstacle in the projection direction.

[0117] For some obstacles with more complex shapes, resampling can be performed by setting a lower sampling rate, and contour optimization can be performed to remove concave surfaces.

[0118] At the same time, for obstacles that are too close, the areas between the obstacles are excluded from the driving area in advance.

[0119] Finally, the driving area and obstacle projections are simultaneously registered to the same coordinate system to form an obstacle map.

[0120] In one embodiment, Figure 3 As shown, step S2 includes:

[0121] Step S21: receiving position information from the user side, marking the vehicle starting position and the vehicle target position on the obstacle map, and performing gridding processing on the obstacle map to obtain a gridded map;

[0122] Step S22: searching for adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position respectively, and recording first search information constructed from the vehicle starting position and second search information from the vehicle target position, until a repeated grid node is found as the search end point;

[0123] Step S23: generating a first half track according to the search end point and the vehicle starting position, and generating a second half track according to the search end point and the vehicle target position;

[0124] Step S24: splicing the first half trajectory and the second half trajectory to obtain a pre-search trajectory.

[0125] Specifically, in order to achieve faster trajectory search efficiency, a bidirectional search algorithm is introduced in this embodiment.

[0126] Specifically, after the obstacle map is constructed, the location information is first received from the user side, and the vehicle starting position and the vehicle target position are marked on the obstacle map.

[0127] To facilitate subsequent searches, the obstacle map is gridded to obtain a gridded map, thereby replacing the continuous space with multiple adjacent network nodes.

[0128] Then, the grid nodes at the starting position and the target position of the vehicle are searched for the adjacent grid nodes in the eight directions in the grid map respectively. When an adjacent grid point is found, the next set of adjacent grid nodes in the eight directions is searched for the adjacent grid point.

[0129] During the search process, the grid point information of the adjacent grid points traversed in each direction is retained and stored in a tree structure. Each point in the tree structure corresponds to a grid node, and the connection corresponds to the adjacent direction.

[0130] When both sides of the tree structure have searched for a repeated grid node, the search process stops. Since the bidirectional search algorithm performs a global search in each direction according to a fixed step size, after finding the first set of grid nodes, it can usually be considered that the current search is the shortest path.

[0131] Then, use this grid node as the search end point.

[0132] In the tree structure, the first half trajectory is generated by searching the connections and traversed grid nodes of the tree structure according to the search end point and the vehicle start position;

[0133] And, in the tree structure, connections of the tree structure and traversed grid nodes are searched according to the search end point and the target position of the vehicle to generate a second half trajectory.

[0134] Finally, the first half trajectory and the second half trajectory are concatenated to obtain the pre-search trajectory.

[0135] In one embodiment, Figure 4 As shown, step S3 includes:

[0136] Step S31: generating an equivalent projection according to vehicle parameters of the vehicle;

[0137] The equivalent projections include two circular projections;

[0138] The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle;

[0139] The diameter of the circular projection is the wheelbase of the vehicle;

[0140] Step S32: performing projection expansion on obstacle projections associated with the pre-search trajectory in the obstacle map to form an expanded projection map;

[0141] Step S33: in the expanded projection map, setting an equivalent projection for each grid node of the pre-search trajectory;

[0142] Step S34: Perform edge expansion on each equivalent projection respectively until it contacts the obstacle projection and stops, and record the vehicle expansion projection;

[0143] Step S35: Generate a trajectory planning corridor based on all vehicle expansion projections.

[0144] Specifically, in order to achieve a better corridor paving effect, in this embodiment, the vehicle is first replaced by an equivalent projection. The equivalent projection uses two circles to cover a rectangular body to replace the vehicle. Specifically, the two circular projections are centered at the front axle center and the rear axle center of the corresponding vehicle, and the edges of the two circular projections are offset, and the radius is half of the wheelbase of the vehicle, which is recorded as R.

[0145] Accordingly, in order to effectively avoid the collision problem, in the obstacle map, according to the coordinates of the pre-search trajectory, the obstacle projections whose edge coordinates are within a certain range are found, and the projection expansion of this part of the obstacle projections is performed, that is, the edge is expanded outward by a distance of R to form an expanded projection map.

[0146] Subsequently, in the expansion projection map, an equivalent projection is placed for each grid node in the pre-search trajectory, and then the outer contour of the equivalent projection is expanded in four directions. When an overlap with an obstacle is detected, the expansion is stopped. The step size of each expansion is a constant size s, and the length of the expanded rectangular box is L.

[0147] When the obstacle is touched, the expansion process is stopped and the grid nodes covered by the vehicle's expanded projection are recorded.

[0148] Finally, the covered grid nodes are spliced ​​to obtain the trajectory planning corridor.

[0149] In one embodiment, Figure 5 As shown, step S4 includes:

[0150] Step S41: Setting a search pointer from the vehicle's starting position;

[0151] Step S42: Filtering the next travel point in the trajectory planning corridor on the search pointer according to the kinematic model;

[0152] Step S43: setting search pointers at the travel points respectively, and then returning to step S42 until the travel point coincides with the vehicle target position;

[0153] Step S44: Optimize the trajectory according to the travel points to obtain the actual driving trajectory.

[0154] Specifically, in order to achieve a better trajectory planning process, in this embodiment, the above processing flow is set based on the kinematic model of the vehicle. Specifically, a search pointer is first set for the vehicle's starting position, and then the next travel point in the trajectory planning corridor is screened on the search pointer according to the kinematic model.

[0155] Specifically, the trajectory planning corridor is a directed area from the vehicle's starting position to the vehicle's target position. After the kinematic model is set on the search pointer, the kinematic model will naturally always move toward the vehicle's target position.

[0156] The kinematic model is an equivalent model designed around the vehicle's suspension system characteristics, mainly including parameters such as wheelbase, front track, rear track, vehicle width, vehicle length, and maximum front wheel turning angle. It is mainly used for the driving actions that the equivalent vehicle can perform in a two-dimensional plane, while omitting irrelevant elements such as suspension geometry changes, the operation process of the vehicle's power system, etc.

[0157] After determining the search pointer and the heading on the current search pointer, the kinematic model can be combined to easily predict several grid nodes that the vehicle can drive to as the next travel point. At the same time, the heading angle of the vehicle after driving to the next travel point can also be calculated.

[0158] Subsequently, a new search pointer is placed for each filtered travel point, and the above-mentioned search process based on the kinematic model is repeated until the vehicle reaches the destination position.

[0159] It is generally believed that since the distance traveled by the vehicle in each step is the same, the first driving trajectory obtained is the shorter trajectory. By combining this trajectory with the collision volume of the vehicle and properly optimizing the concave surface, the optimal driving trajectory can be obtained.

[0160] An autonomous parking trajectory planning system for irregular road obstacles, used to implement the above autonomous parking trajectory planning method;

[0161] like Figure 6 As shown, the autonomous parking trajectory planning system includes:

[0162] A map modeling module 1, which scans the planned area to obtain multiple obstacle projections and constructs an obstacle map;

[0163] Pre-search module 2, the pre-search module 2 is connected to the map modeling module 1, and the pre-search module 2 pre-searches the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory;

[0164] The position information includes the vehicle starting position and the vehicle target position;

[0165] A corridor laying module 3, the corridor laying module 3 is connected to the pre-search module 2, and the corridor laying module 3 performs trajectory expansion along the pre-search trajectory to form a trajectory planning corridor;

[0166] The boundaries of the trajectory planning corridor are offset by the obstacle projection;

[0167] The trajectory generation module 4 is connected to the corridor laying module 3. The trajectory generation module 4 obtains the actual driving trajectory in the trajectory planning corridor according to the kinematic model planning of the vehicle.

[0168] Specifically, in view of the problem that the planning process of automatic parking in multi-obstacle scenarios in the prior art is relatively complex and requires high computing power, in this solution, the pre-search module 2 obtains a rough pre-search trajectory by pre-sampling and searching the obstacle map constructed by the map modeling module 1, and the corridor laying module 3 then generates a trajectory planning corridor based on the pre-search trajectory. The trajectory generation module 4 performs trajectory planning within the corridor constraints, thereby separating the vehicle trajectory planning process from the obstacles in the environment, solving the problem that the obstacle constraints in unstructured roads cannot be solved in the process of nonlinear solution, and simplifying the scale of the optimization problem.

[0169] In one embodiment, Figure 7 As shown, the map modeling module 1 includes:

[0170] A map acquisition module 11, which acquires digital map information related to the area to be planned from an external data source;

[0171] A map parsing module 12, the map parsing module 12 is connected to the map acquisition module 11, and the map parsing module 12 parses the digital map information to obtain the driving area and the obstacle information associated with the driving area;

[0172] A resampling module 13, the resampling module 13 is connected to the map parsing module 12, and the resampling module 13 resamples the obstacle information to obtain obstacle projection;

[0173] The map generation module 14 is connected to the resampling module 13 , and generates an obstacle map according to the driving area and obstacle projection.

[0174] Specifically, in order to achieve a better map construction effect, in this embodiment, the map acquisition module 11 first obtains digital map information related to the area to be planned from an external data source.

[0175] For example, the external data source may be a parking system or a provider of a digital map that marks the locations of various structures of the building, including load-bearing structures, decorative structures, fire-fighting and ventilation facilities, etc.

[0176] In some embodiments, an external monitoring system may be introduced to perform real-time three-dimensional modeling of the site to form a digital twin model corresponding to the parking lot, and the available driving area and irregular obstacles may be identified in the self-modeling to form complete digital map information.

[0177] Subsequently, the map parsing module 12 parses the digital map information, including two-dimensionalizing it, removing redundant annotations, and aligning it to a coordinate system, so as to retain only the driving area and obstacle information associated with the driving area.

[0178] On this basis, the resampling module 13 resamples the obstacle information to obtain obstacle projection, mainly including removing other description information of the obstacle, obtaining the maximum outer contour of the obstacle in the projection direction, etc.

[0179] For some obstacles with more complex shapes, resampling can be performed by setting a lower sampling rate, and contour optimization can be performed to remove concave surfaces.

[0180] At the same time, for obstacles that are too close, the areas between the obstacles are excluded from the driving area in advance.

[0181] Finally, the map generation module 14 aligns the driving area and the obstacle projection into the same coordinate system to form an obstacle map.

[0182] In one embodiment, Figure 8 As shown, the pre-search module 2 includes:

[0183] The map processing module 21 receives the position information from the user side, marks the vehicle starting position and the vehicle target position on the obstacle map, and performs grid processing on the obstacle map to obtain a grid map;

[0184] A bidirectional search module 22, the bidirectional search module 22 is connected to the map processing module 21, and the bidirectional search module 22 searches for adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position, respectively, and records the first search information constructed from the vehicle starting position and the second search information from the vehicle target position, until a repeated grid node is found as the search end point;

[0185] A trajectory generation module 23, the trajectory generation module 23 is connected to the bidirectional search module 22, and the trajectory generation module 23 generates a first half trajectory according to the search end point and the vehicle starting position, and generates a second half trajectory according to the search end point and the vehicle target position;

[0186] The track splicing module 24 is connected to the track generating module 23 , and the track splicing module 24 splices the first half track and the second half track to obtain a pre-search track.

[0187] Specifically, in order to achieve faster trajectory search efficiency, a bidirectional search algorithm is introduced in this embodiment.

[0188] Specifically, after constructing the obstacle map, the map processing module 21 first receives position information from the user side, and marks the vehicle starting position and the vehicle target position on the obstacle map.

[0189] To facilitate subsequent searches, the obstacle map is gridded to obtain a gridded map, thereby replacing the continuous space with multiple adjacent network nodes.

[0190] Then, the bidirectional search module 22 searches for the grid nodes in the eight adjacent directions in the grid map from the grid nodes where the vehicle's starting position and the vehicle's target position are located. When an adjacent grid point is found, the adjacent grid point is searched for the next set of adjacent grid nodes in the eight adjacent directions.

[0191] During the search process, the grid point information of the adjacent grid points traversed in each direction is retained and stored in a tree structure. Each point in the tree structure corresponds to a grid node, and the connection corresponds to the adjacent direction.

[0192] When both sides of the tree structure have searched for a repeated grid node, the search process stops. Since the bidirectional search algorithm performs a global search in each direction according to a fixed step size, after finding the first set of grid nodes, it can usually be considered that the current search is the shortest path.

[0193] Then, use this grid node as the search end point.

[0194] In the tree structure, the trajectory generation module 23 searches for connections and traversed grid nodes in the tree structure according to the search end point and the vehicle start position to generate a first half trajectory;

[0195] And, in the tree structure, connections of the tree structure and traversed grid nodes are searched according to the search end point and the target position of the vehicle to generate a second half trajectory.

[0196] Finally, the trajectory concatenation module 24 concatenates the first half trajectory and the second half trajectory to obtain a pre-search trajectory.

[0197] In one embodiment, Fig. 9 As shown, the corridor paving module 3 includes:

[0198] An equivalent projection generation module 31, the equivalent projection generation module 31 generates an equivalent projection according to vehicle parameters of the vehicle;

[0199] The equivalent projections include two circular projections;

[0200] The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle;

[0201] The diameter of the circular projection is the wheelbase of the vehicle;

[0202] A projection expansion module 32, the projection expansion module 32 is connected to the equivalent projection generation module 31, and the projection expansion module 32 performs projection expansion on obstacle projections associated with the pre-search trajectory in the obstacle map to form an expanded projection map;

[0203] An equivalent projection setting module 33, the equivalent projection setting module 33 is connected to the projection expansion module 32, and the equivalent projection setting module 33 sets an equivalent projection for each grid node of the pre-search trajectory in the expansion projection map;

[0204] The edge expansion module 34 is connected to the equivalent projection setting module 33. The edge expansion module 34 performs edge expansion on each equivalent projection respectively until it contacts the obstacle projection and stops, and records the vehicle expansion projection.

[0205] The corridor superposition module 35 is connected to the edge expansion module 34, and the corridor superposition module 35 generates a trajectory planning corridor according to the expansion projections of all vehicles.

[0206] Specifically, in order to achieve a better corridor paving effect, in this embodiment, the equivalent projection generation module 31 first replaces the vehicle with an equivalent projection. The equivalent projection uses two circles to cover a rectangular body to replace the vehicle. Specifically, the two circular projections are centered at the front axle center and the rear axle center of the corresponding vehicle, and the edges of the two circular projections are offset, and the radius is half of the wheelbase of the vehicle, which is recorded as R.

[0207] Accordingly, in order to effectively avoid the collision problem, the projection expansion module 32 searches for obstacle projections whose edge coordinates are within a certain range in the obstacle map according to the coordinates of the pre-search trajectory, and performs projection expansion on these obstacle projections, that is, the edge is expanded outward by a distance R to form an expanded projection map.

[0208] Subsequently, the equivalent projection setting module 33 places an equivalent projection for each grid node in the pre-search trajectory in the expanded projection map, and then the edge expansion module 34 expands from the outer contour of the equivalent projection in four directions. When an overlap with an obstacle is detected, the expansion is stopped. The step size of each expansion is a constant size s, and the length of the expanded rectangular box is L.

[0209] When the obstacle is touched, the expansion process is stopped and the grid nodes covered by the vehicle's expanded projection are recorded.

[0210] Finally, the corridor overlay module 35 splices the covered grid nodes to obtain the trajectory planning corridor.

[0211] In one embodiment, Fig.10 As shown, the trajectory generation module 4 includes:

[0212] A starting pointer placement module 41, the starting pointer placement module 41 sets a search pointer from the starting position of the vehicle;

[0213] A kinematics screening module 42, the kinematics screening module 42 is connected to the start pointer placement module 41, and the kinematics screening module 42 screens the next travel point in the trajectory planning corridor on the search pointer according to the kinematics model;

[0214] A pointer search module 43, the pointer search module 43 is connected to the kinematics screening module 42, and the pointer search module 43 sets a search pointer on each travel point until the travel point coincides with the target position of the vehicle;

[0215] The trajectory optimization module 44 is connected to the pointer search module 43. The trajectory optimization module 44 performs trajectory optimization according to the travel points to obtain the actual driving trajectory.

[0216] Specifically, in order to achieve a better trajectory planning process, in this embodiment, the above processing flow is set based on the kinematic model of the vehicle. Specifically, the starting pointer placement module 41 first sets a search pointer for the vehicle starting position, and then selects the next travel point in the trajectory planning corridor according to the kinematic model on the search pointer.

[0217] Specifically, the trajectory planning corridor is a directed area from the vehicle's starting position to the vehicle's target position. After the kinematic model is set on the search pointer, the kinematic model will naturally always move toward the vehicle's target position.

[0218] The kinematic model is an equivalent model designed around the vehicle's suspension system characteristics, mainly including parameters such as wheelbase, front track, rear track, vehicle width, vehicle length, and maximum front wheel turning angle. It is mainly used for the driving actions that the equivalent vehicle can perform in a two-dimensional plane, while omitting irrelevant elements such as suspension geometry changes, the operation process of the vehicle's power system, etc.

[0219] After determining the search pointer and the heading on the current search pointer, the kinematics screening module 42 can easily predict several grid nodes that the vehicle can travel to as the next travel point in combination with the kinematics model. At the same time, the heading angle of the vehicle after traveling to the next travel point can also be calculated.

[0220] Subsequently, the pointer search module 43 places a new search pointer for each of the selected travel points, and repeats the above-mentioned search process based on the kinematic model until the vehicle reaches the destination position.

[0221] It is generally believed that since the distance traveled by the vehicle in each step is the same, the first driving trajectory obtained is a shorter trajectory. The trajectory optimization module 44 optimizes the concave surface appropriately based on the trajectory and the collision volume of the vehicle to obtain the optimal driving trajectory.

[0222] Those skilled in the art will appreciate that various aspects of the present invention, or possible implementations of various aspects, may be specifically implemented as systems, methods, or computer program products. Therefore, various aspects of the present invention, or possible implementations of various aspects, may take the form of complete hardware embodiments, complete software embodiments (including firmware, resident software, etc.), or embodiments of combined software and hardware aspects, all collectively referred to herein as "circuits," "modules," or "systems." In addition, various aspects of the present invention, or possible implementations of various aspects, may take the form of computer program products, which refer to computer instructions stored in a memory.

[0223] The memory may be a computer-readable signal medium or a computer-readable storage medium. Computer-readable storage media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or apparatuses, or any suitable combination of the foregoing, such as random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable read-only memory (CD-ROM).

[0224] The processor in the computer reads the computer instructions stored in the memory, so that the processor can execute the functional actions specified in each step or the combination of steps in the flowchart; and generate a device for implementing the functional actions specified in each block or the combination of blocks in the block diagram.

[0225] It should be understood that the processor in the computer can be understood as one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components implemented to execute the aforementioned computer instructions.

[0226] Computer instructions can be executed completely on the user's local computer, partially on the user's local computer, as a separate software package, partially on the user's local computer and partially on a remote computer, or completely on a remote computer or server. It should also be noted that in some alternative embodiments, the functions noted in each step in the flow chart or each block in the block diagram may not occur in the order noted in the figure. For example, depending on the functions involved, two steps or two blocks shown in succession may actually be executed roughly simultaneously, or these blocks may sometimes be executed in reverse order.

[0227] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. An autonomous parking trajectory planning method for irregular road obstacles, characterized in that: include: Step S1: Scan the planned area to obtain multiple obstacle projections and construct an obstacle map; Step S2: Pre-search the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory; The position information includes a vehicle starting position and a vehicle target position; Step S3: performing trajectory expansion along the pre-search trajectory to form a trajectory planning corridor; The boundary of the trajectory planning corridor offsets the obstacle projection; Step S4: obtaining an actual driving trajectory in the trajectory planning corridor according to the kinematic model of the vehicle.

2. The autonomous parking trajectory planning method according to claim 1, characterized in that: The step S1 comprises: Step S11: acquiring digital map information associated with the area to be planned from an external data source; Step S12: parsing the digital map information to obtain a driving area and obstacle information associated with the driving area; Step S13: resampling the obstacle information to obtain the obstacle projection; Step S14: generating the obstacle map according to the driving area and the obstacle projection.

3. The autonomous parking trajectory planning method according to claim 1, characterized in that: The step S2 comprises: Step S21: receiving the position information from the user side, marking the vehicle starting position and the vehicle target position on the obstacle map, and performing gridding processing on the obstacle map to obtain a gridded map; Step S22: searching adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position respectively, and recording first search information constructed from the vehicle starting position and second search information from the vehicle target position, until a repeated grid node is found as a search end point; Step S23: generating a first half track according to the search end point and the vehicle starting position, and generating a second half track according to the search end point and the vehicle target position; Step S24: splicing the first half trajectory and the second half trajectory to obtain the pre-search trajectory.

4. The autonomous parking trajectory planning method according to claim 3, characterized in that: The step S3 comprises: Step S31: generating an equivalent projection according to vehicle parameters of the vehicle; The equivalent projection includes two circular projections; The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle respectively; The diameter of the circular projection is the wheelbase of the vehicle; Step S32: performing projection expansion on the obstacle projection associated with the pre-search trajectory in the obstacle map to form an expanded projection map; Step S33: in the expanded projection map, respectively setting the equivalent projection for each grid node of the pre-search trajectory; Step S34: performing edge expansion on each of the equivalent projections respectively until the edge expansion stops after the edge expansion reaches the obstacle projection, and recording the vehicle expansion projection; Step S35: generating the trajectory planning corridor according to all the vehicle expansion projections.

5. The autonomous parking trajectory planning method according to claim 1, characterized in that: The step S4 comprises: Step S41: setting a search pointer from the vehicle starting position; Step S42: Filtering the next travel point in the trajectory planning corridor on the search pointer according to the kinematic model; Step S43: setting the search pointers at the travel points respectively, and then returning to step S42 until the travel points coincide with the vehicle target position; Step S44: Optimizing the trajectory according to the travel points to obtain the actual driving trajectory.

6. An autonomous parking trajectory planning system for irregular road obstacles, characterized in that: Used to implement the autonomous parking trajectory planning method as described in any one of claims 1 to 5; The autonomous parking trajectory planning system comprises: A map modeling module, wherein the map modeling module scans the area to be planned to obtain multiple obstacle projections and construct an obstacle map; A pre-search module, the pre-search module is connected to the map modeling module, and the pre-search module pre-searches the obstacle map according to the position information generated by the user side to obtain a pre-search trajectory; The position information includes a vehicle starting position and a vehicle target position; A corridor laying module, the corridor laying module is connected to the pre-search module, and the corridor laying module performs trajectory expansion along the pre-search trajectory to form a trajectory planning corridor; The boundary of the trajectory planning corridor offsets the obstacle projection; A trajectory generation module is connected to the corridor laying module, and the trajectory generation module obtains an actual driving trajectory in the trajectory planning corridor according to the kinematic model planning of the vehicle.

7. The autonomous parking trajectory planning system according to claim 6, characterized in that: The map modeling module includes: A map acquisition module, wherein the map acquisition module acquires digital map information associated with the area to be planned from an external data source; A map parsing module, the map parsing module is connected to the map acquisition module, and the map parsing module parses the digital map information to obtain a driving area and obstacle information associated with the driving area; A resampling module, the resampling module is connected to the map parsing module, and the resampling module resamples the obstacle information to obtain the obstacle projection; A map generation module is connected to the resampling module, and the map generation module generates the obstacle map according to the driving area and the obstacle projection.

8. The autonomous parking trajectory planning method according to claim 6, characterized in that: The pre-search module comprises: A map processing module, wherein the map processing module receives the position information from the user side, marks the vehicle starting position and the vehicle target position on the obstacle map, and performs grid processing on the obstacle map to obtain a grid map; A bidirectional search module, the bidirectional search module is connected to the map processing module, the bidirectional search module searches for adjacent grid nodes in the grid map from the vehicle starting position and the vehicle target position, respectively, and records first search information constructed from the vehicle starting position and second search information from the vehicle target position, until a repeated grid node is found as a search end point; A trajectory generation module, the trajectory generation module is connected to the bidirectional search module, the trajectory generation module generates a first half trajectory according to the search end point and the vehicle starting position, and generates a second half trajectory according to the search end point and the vehicle target position; A trajectory splicing module is connected to the trajectory generating module, and the trajectory splicing module splices the first half trajectory and the second half trajectory to obtain the pre-search trajectory.

9. The autonomous parking trajectory planning system according to claim 8, characterized in that: The corridor laying module includes: An equivalent projection generation module, wherein the equivalent projection generation module generates an equivalent projection according to vehicle parameters of the vehicle; The equivalent projection includes two circular projections; The center of each circular projection corresponds to the center of the front axle and the center of the rear axle of the vehicle respectively; The diameter of the circular projection is the wheelbase of the vehicle; A projection expansion module, the projection expansion module is connected to the equivalent projection generation module, and the projection expansion module performs projection expansion on the obstacle projection associated with the pre-search trajectory in the obstacle map to form an expanded projection map; An equivalent projection setting module, the equivalent projection setting module is connected to the projection expansion module, and the equivalent projection setting module sets the equivalent projection for each grid node of the pre-search trajectory in the expansion projection map; An edge expansion module, the edge expansion module is connected to the equivalent projection setting module, and the edge expansion module performs edge expansion on each of the equivalent projections, stops after contacting the obstacle projection, and records the vehicle expansion projection; A corridor superposition module is connected to the edge expansion module, and the corridor superposition module generates the trajectory planning corridor according to all the vehicle expansion projections.

10. The autonomous parking trajectory planning method according to claim 6, characterized in that: The trajectory generation module comprises: A starting pointer placement module, the starting pointer placement module sets a search pointer from the starting position of the vehicle; A kinematics screening module, the kinematics screening module is connected to the start pointer placement module, and the kinematics screening module screens the next travel point in the trajectory planning corridor on the search pointer according to the kinematic model; A pointer search module, the pointer search module is connected to the kinematics screening module, and the pointer search module respectively sets the search pointer on the travel point until the travel point coincides with the vehicle target position; A trajectory optimization module is connected to the pointer search module, and the trajectory optimization module performs trajectory optimization according to the travel points to obtain the actual driving trajectory.

Images