An automatic parking path planning method for an unmanned vehicle

Abstract

This invention relates to the field of autonomous vehicle control and discloses an automatic parking path planning method for autonomous vehicles, comprising the following steps: acquiring parking environment information, establishing a local parking environment model, and determining the target parking posture of the vehicle; generating candidate parking paths based on vehicle kinematic constraints and path parameters, sampling the candidate parking paths, and calculating the corrected clearance for key corner points of the vehicle at each sampling point; establishing a corner clearance phase table based on the corrected clearance, and performing clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification. This invention improves the targeting of automatic parking path correction and the safety consistency of path output by coordinating the processing of corrected clearance, the corner clearance phase table, corner-boundary risk pairing, path parameter correction permissions, and the phase consequence prediction table.

Description

Technical Field

[0001] This invention relates to the field of autonomous vehicle control technology, specifically to an automatic parking path planning method for autonomous vehicles. Background Technology

[0002] With the development of autonomous vehicles and automatic parking functions, vehicles need to complete parking path planning in low-speed scenarios where environmental boundaries such as parking space boundaries, adjacent vehicles, walls, pillars, and curbs are relatively close. Since the change in distance between the vehicle's outer corner points and the environmental boundaries during parking directly affects the feasibility of the path, a method is needed that can combine the clearance status of the vehicle's key corner points for path planning and correction.

[0003] Existing automated parking path planning typically generates candidate parking paths based on the vehicle's current pose, the target parking pose, and vehicle kinematic constraints. It then determines whether the path meets traffic requirements using vehicle reference point trajectories, vehicle outline envelopes, or obstacle distance constraints. When a path does not meet the requirements, methods such as replanning, adjusting path curvature, changing path segment lengths, or increasing safety margins are often used to address the parking path issue.

[0004] While existing technologies can generate parking paths that satisfy basic kinematic constraints, some shortcomings remain: current path determinations are mostly based on vehicle reference points or a unified outer envelope, and the changes in clearance between key vehicle corner points and environmental boundaries in different path segments are not treated as continuous states; when local low clearance occurs, it is difficult to distinguish which key vehicle corner point, which environmental boundary, and which path segment the low clearance originates from, leading to path corrections that easily manifest as overall adjustments or repeated replanning; at the same time, the lack of correlation judgment between phase switching, boundary correlation changes, and path parameter correction permissions between low clearance points can easily lead to the current correction introducing subsequent phase conflicts or unmet execution constraints. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an automatic parking path planning method for autonomous vehicles. It solves the problems in existing automatic parking path planning, such as the lack of continuous representation of the clearance status of key corner points of the vehicle, the difficulty in associating low clearance risk sources with specific corner points and environmental boundaries, and the lack of phase state constraints in path parameter correction, which leads to insufficient targeted correction.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an automatic parking path planning method for autonomous vehicles, comprising the following steps: acquiring parking environment information, establishing a local parking environment model, and determining the target parking posture of the vehicle; Candidate parking paths are generated based on vehicle kinematic constraints and path parameters. The candidate parking paths are sampled, and the corrected clearance of the vehicle's key corner points at each sampling point is calculated. Based on the corrected clearance, a corner clearance phase table is established, and clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification are performed. The path parameter correction authority is determined based on the latch clearance phase, the nearest boundary type, and the current path segment type. The path parameter sensitivity is calculated, and candidate path parameter correction schemes are screened based on the phase consequence prediction table. Update the path parameters based on the filtering results or reconstruct the path segment for phase conflict segments, and output the automatic parking path.

[0007] Preferably, the local parking environment model includes a set of environmental boundaries, and each environmental boundary in the set is configured with a boundary number, boundary geometry, boundary type, boundary error, boundary valid state, and boundary update time. The boundary types include at least one of the following: parking space boundary, adjacent vehicle boundary, wall boundary, pillar boundary, curb boundary, limiter boundary, and general obstacle boundary. When the confidence level of boundary type identification is lower than the preset condition, or when the boundary category is inconsistent within a continuous planning period, the corresponding environmental boundary will be used as a general obstacle boundary in the correction clearance calculation of the vehicle's key corner points.

[0008] Preferably, the key corner points of the vehicle include at least the front left corner point, the front right corner point, the rear left corner point, and the rear right corner point; The calculation of the corrected clearance of the vehicle's key corner points at each sampling point includes: determining the coordinates of the vehicle's key corner points in the vehicle's body coordinate system based on the vehicle's external dimensions; The positions of the vehicle's key corner points in the parking local coordinate system are determined based on the vehicle reference point coordinates and the vehicle heading angle at the sampling points. Calculate the distance from the vehicle's key corner points to each environmental boundary, take the minimum distance as the original clearance, and record the nearest boundary number and the nearest boundary type; calculate the corrected clearance based on the original clearance and the clearance correction amount.

[0009] Preferably, the clearance correction is determined based on the boundary error of the nearest boundary, the vehicle planar positioning error, the vehicle heading angle error, the steering angle error, and the steering actuator response hysteresis. Among them, the impact of vehicle heading angle error on clearance is determined based on the position of the vehicle's key corner points relative to the vehicle's reference point; the impact of steering angle error on clearance is determined based on the corrected clearance changes before and after steering angle disturbance; and the impact of steering actuator response lag on clearance is determined based on changes in vehicle speed, response lag time, and clearance along the path direction.

[0010] Preferably, the corner clearance phase table records at least the sampling point number, vehicle status, path segment type, corrected clearance of each vehicle's key corner point, candidate dominant corner point, nearest boundary number corresponding to the candidate dominant corner point, and nearest boundary type corresponding to the candidate dominant corner point; The candidate dominant corner point is the vehicle's critical corner point with the smallest corrected clearance at the same sampling point; The corner clearance phase table is used to determine the clearance phase status distributed along the candidate parking path and serves as input for clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification.

[0011] Preferably, the clearance phase latching and release includes: when the same candidate dominant corner point appears consecutively within the latching judgment window, and the current path segment belongs to the allowed latching path segment, or the current minimum corrected clearance is lower than the boundary and close to the threshold, the candidate dominant corner point is determined as the latched clearance phase; When new candidate dominant corner points appear consecutively within the release judgment window, the current path segment belongs to the allowed phase switching path segment, and both the original latched corner point and the new candidate dominant corner point meet the phase switching safety threshold at the switching position, the latch clearance phase is switched to the new candidate dominant corner point. If the phase switching conditions are not met, the original latch net phase is maintained.

[0012] Preferably, the corner-boundary risk pairing includes the latched clearance phase, the nearest boundary number corresponding to the latched clearance phase, the nearest boundary type corresponding to the latched clearance phase, and the current path segment type; When there is a latching clearance phase at adjacent sampling points, the boundary association change marker is determined based on whether the dominant corner point changes and whether the nearest boundary number changes. When no latching clear phase is formed between any of the adjacent sampling points, the nearest boundary number is not used to determine the boundary association change.

[0013] Preferably, the low headroom non-trigger zone filtering includes: determining sampling points where the minimum corrected headroom is lower than the minimum headroom threshold as low headroom points, and determining sampling points where the minimum corrected headroom is lower than the hard safety headroom threshold as hard safety points. The hard safety point does not enter the low-headroom non-triggered zone filter and serves as the trigger for replanning, path segment reconstruction, or parking control. For low clearance points that are not hard safety points, whether to trigger path correction is determined based on phase continuity, boundary correlation continuity, phase switching neighborhood, and execution sensitivity. Low clearance points that do not trigger path correction are classified into the low clearance non-triggered zone and continue to participate in phase conflict segment identification.

[0014] Preferably, the phase conflict segment identification includes: counting the non-zero latch clearance phase, the nearest boundary number, the number of phase switching, and the number of boundary association changes within the path interval to be judged; When at least two non-zero latch clearance phases appear in the path interval to be judged, and at least one of the following conditions is met: the number of nearest boundary numbers is not less than two, the number of phase switching is greater than the upper limit of the number of phase switching, and the number of boundary association changes is greater than the upper limit of the number of boundary association changes, the path interval to be judged is determined as a phase conflict segment. The number of phase switching and the number of boundary correlation changes are both based on the statistics of sampling points corresponding to the non-zero latch net phase.

[0015] Preferably, the path parameter correction permission is determined by the path parameter correction permission table based on the latch clearance phase, the nearest boundary type, and the current path segment type; When multiple corner-boundary risk pairs exist within the path correction trigger area, the trigger sampling points are grouped according to the latch clearance phase, the type of the nearest boundary, and the type of the current path segment. The same group uses the same path parameter correction authority, and different groups participate in the path parameter sensitivity calculation. The path parameter sensitivity is determined by applying a perturbation to the path parameters that are allowed to be corrected and comparing the changes in corrected clearance at the vehicle's key corner points before and after the perturbation. The phase consequence prediction table is used to record the prediction correction clearance, prediction latch clearance phase, prediction nearest boundary number, prediction boundary association change marker, prediction phase conflict state, prediction terminal attitude state, and prediction execution constraint state within the prediction window, and is used to filter candidate path parameter correction schemes that meet the constraints.

[0016] This invention provides an automatic parking path planning method for autonomous vehicles. It has the following beneficial effects: 1. This invention calculates the corrected clearance of key corner points of the vehicle at each sampling point of the candidate parking path, and establishes a corner clearance phase table based on the corrected clearance, so that the path planning process can reflect the distance changes between key positions of the vehicle outline and the environmental boundary, avoiding the use of vehicle reference point trajectory or unified safety envelope as the basis for path judgment.

[0017] 2. This invention incorporates environmental boundary error, vehicle positioning and heading error, steering angle error, and steering actuator response lag into the clearance correction amount, so that the corrected clearance can more accurately represent the safety margin in actual parking operations.

[0018] 3. This invention associates low clearance status with key corner points, nearest boundary types, and path segment types of specific vehicles by using clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification. This enables path correction to address stable risk sources and reduces frequent corrections caused by clearance fluctuations at individual sampling points. Attached Figure Description

[0019] Figure 1 This is a flowchart of the automatic parking path planning method for driverless cars according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0021] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0022] Example: See attached document Figure 1 This invention provides an automatic parking path planning method for autonomous vehicles. The method is executed by a parking controller, an autonomous driving domain controller, or an onboard control unit with parking path planning capabilities. It generates an automatic parking path based on parking environment boundaries, vehicle kinematic parameters, vehicle critical corner clearances, and path execution constraints. The method includes the following steps: S100. Obtain parking environment information, establish a local parking environment model, and determine the target parking posture of the vehicle. S200. Generate candidate parking paths based on vehicle kinematic constraints and path parameters, sample the candidate parking paths, and calculate the corrected clearance of the vehicle's key corner points at each sampling point. S300: Establish a corner clearance phase table based on the corrected clearance of the vehicle's key corner points, and perform clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification based on the corner clearance phase table. S400. Determine the path parameter correction authority based on the latch clearance phase, the nearest boundary type, and the current path segment type. Calculate the path parameter sensitivity within the scope of the path parameter correction authority and screen candidate path parameter correction schemes based on the phase consequence prediction table. S500: Update the path parameters according to the screening results, or perform path segment reconstruction for phase conflict segments, and output the automatic parking path when the corner clearance constraint, phase switching constraint, boundary association constraint, vehicle kinematics constraint and target parking posture constraint are satisfied; during the process of the vehicle executing the automatic parking path, update the vehicle pose, environmental boundary and vehicle key corner correction clearance, and perform path parameter correction, path segment reconstruction, replanning or parking control according to the update results.

[0023] The technical implementation details of the above steps are explained in detail below with reference to specific embodiments: Step S100 generates the input data required for subsequent candidate parking path generation, vehicle key corner point correction clearance calculation, clearance phase judgment, and path parameter correction.

[0024] S101. Obtain environmental information within the vehicle parking area.

[0025] The vehicle acquires environmental information within the parking area through onboard sensing devices. These devices include one or more of the following: surround-view cameras, ultrasonic radar, millimeter-wave radar, and lidar. The vehicle simultaneously acquires its position, heading angle, wheel speed, steering angle, and gear status.

[0026] Parking environment information includes parking space boundaries, adjacent vehicle boundaries, wall boundaries, pillar boundaries, curb boundaries, limiter boundaries, and other obstacle boundaries. Parking space boundaries are used to define the target parking area, while adjacent vehicle boundaries, wall boundaries, pillar boundaries, curb boundaries, limiter boundaries, and other obstacle boundaries are used to define intrusive areas.

[0027] For parking line recognition, obstacle detection, point cloud clustering, and boundary fitting, those skilled in the art can use existing perception algorithms based on the vehicle sensor configuration. Sensor calibration and coordinate transformation are well-known technologies in this field and will not be elaborated here.

[0028] S102. Establish the local coordinate system for parking.

[0029] A parking local coordinate system is established using the vehicle reference point when the vehicle enters the parking planning state as the origin. The vehicle reference point can be the rear axle center, the vehicle's geometric center, or a preset vehicle control reference point. Environmental data from different sensors are converted to the vehicle's body coordinate system via calibration parameters, and then further converted to the parking local coordinate system based on the vehicle's current pose.

[0030] S103, Forming a set of environmental boundaries.

[0031] The environmental data in the parking local coordinate system is organized according to boundary categories and spatial continuity to form an environmental boundary set: In the formula, Represents the set of environmental boundaries. Indicates the first An environmental boundary, Indicates the number of environmental boundaries.

[0032] Each environmental boundary must include at least the boundary number, boundary geometry, boundary type, boundary error, boundary valid state, and boundary update time, represented as: In the formula, Indicates the first Boundary number of each environmental boundary. Indicates the first The boundary geometry of an environment boundary. Indicates the first The boundary type of an environment boundary. Indicates the first Boundary error of an environmental boundary, Indicates the first The effective state of an environmental boundary Indicates the first The update time of each environment boundary Indicates the environmental boundary number.

[0033] Boundary types include at least parking space boundaries, adjacent vehicle boundaries, wall boundaries, column boundaries, curb boundaries, limit switch boundaries, and general obstacle boundaries. When the confidence level of a boundary type is lower than the preset condition, or when the boundary category is inconsistent within consecutive planning periods, the boundary type is set to a general obstacle boundary, and the boundary is retained for subsequent clearance calculations.

[0034] Boundary geometry is represented using line segments, polylines, or local curves. When using polylines... Specifically, it is expressed as follows: In the formula, Indicates the first The first environmental boundary A boundary point, Indicates the first The index of the endpoint of an environment boundary. This indicates the index of the point on the boundary.

[0035] The valid boundary state is determined based on boundary length, boundary point continuity, update time, sensor confidence, and boundary position stability. When a boundary is invalid, it is not used as the target parking space boundary; however, the boundary can still be used as a general obstacle boundary in safety distance judgment. Boundary error is used to represent the positional uncertainty of the environmental boundary during perception, fitting, or coordinate transformation, and is used in subsequent calculations of the vehicle's critical corner point clearance correction.

[0036] S104. Determine the parking areas and non-intrusive areas.

[0037] Parkable and intrusive areas are determined based on the set of environmental boundaries. The area enclosed by the parking space boundaries serves as the base area for the target parking area; the areas corresponding to adjacent vehicle boundaries, wall boundaries, pillar boundaries, curb boundaries, limiter boundaries, and general obstacle boundaries are designated as intrusive areas.

[0038] Parkable areas and inaccessible areas satisfy the following: In the formula, This indicates the area that vehicles are permitted to traverse when parking. This indicates the area where the vehicle's outline must not be encroached upon.

[0039] In subsequent path planning, the area swept by the vehicle's outer contour satisfies: In the formula, Represents the path parameter vector The generated vehicle outline sweep area, This represents the path parameter vector of the candidate parking path.

[0040] S105. Maintain environmental boundary numbering.

[0041] To support subsequent corner-boundary risk pairing, environmental boundary numbers are maintained across consecutive planning cycles. Environmental boundaries detected in the current planning cycle are matched with those from the previous planning cycle based on boundary location, direction, type, and endpoint distance. Successfully matched environmental boundaries retain the boundary number from the previous planning cycle; unmatched boundaries are assigned new boundary numbers.

[0042] By maintaining the boundary number, subsequent steps can directly determine whether the nearest boundary corresponding to the vehicle's key corner point has changed in adjacent sampling points or adjacent planning periods, thereby supporting corner point-boundary risk pairing and phase conflict segment identification.

[0043] S106. Determine the target parking posture of the vehicle.

[0044] Based on the target parking space boundaries, parking direction, and vehicle parking requirements, determine the target parking posture of the vehicle. The target parking posture of the vehicle is represented as follows: In the formula, Indicates the vehicle's target parking posture. This represents the lateral coordinates of the target vehicle's reference point in the parking local coordinate system. This represents the longitudinal coordinate of the target vehicle's reference point in the parking local coordinate system. Indicates the heading angle of the target vehicle.

[0045] When the target parking space has a longitudinal centerline, the target heading angle is determined by the direction of the longitudinal centerline. The target reference point position is determined based on the center point of the target parking space, the vehicle's external dimensions, and the offset of the vehicle reference point relative to the vehicle's geometric center. In the formula, Indicates the center point of the target parking space. This represents a two-dimensional rotation matrix determined by the heading angle of the target vehicle. This represents the offset vector from the vehicle's geometric center to the vehicle's reference point.

[0046] S107, Output the data results of S100.

[0047] After processing by S100, the output includes the local coordinate system for parking, the set of environmental boundaries, the parking area, the non-intrusive area, the current state of the vehicle, and the target parking posture of the vehicle.

[0048] The current status of the vehicle is represented as follows: In the formula, Indicates the current status of the vehicle. and This represents the coordinates of the vehicle reference point in the parking local coordinate system. Indicates the vehicle's current heading angle. Indicates the vehicle's current speed. This indicates the vehicle's current steering angle.

[0049] In S100, the parking environment boundary is organized into a data structure with boundary number, boundary type, boundary error, and effective state. The target parking attitude of the vehicle is represented as the coordinates of the target reference point and the target heading angle in the local parking coordinate system. The above results are used in S200 for candidate parking path generation and vehicle critical corner point correction clearance calculation.

[0050] Step S200 generates candidate parking paths and calculates the original clearance, nearest boundary, and corrected clearance for each sampling point of the vehicle's key corner points. The corrected clearance serves as input for step S300 to establish a corner clearance phase table.

[0051] S201. Determine the vehicle kinematic constraints and path parameters.

[0052] Vehicle kinematic constraints include wheelbase constraints, maximum steering angle constraints, steering angle change rate constraints, forward direction constraints, reverse direction constraints, and end-point attitude constraints. For vehicle kinematic modeling in low-speed parking, those skilled in the art can complete it using existing parking kinematic models based on vehicle chassis parameters. The specific models are well-known in the field and will not be elaborated upon here.

[0053] The path parameter vector is represented as: In the formula, Represents a path parameter vector. Indicates the first One path parameter.

[0054] The path parameters include at least one of the following: reversing start position, first steering segment length, curvature gradually increasing segment length, fixed curvature segment length, return-to-center start point, end alignment segment length, forward correction segment length, and secondary reversing segment length. These path parameters are used to define the length, start and end positions, steering hold status, and return-to-center position of each segment in the candidate parking path.

[0055] S202. Generate candidate parking paths and sample them.

[0056] Based on the vehicle's current state, target parking posture, path parameter vector, and vehicle kinematic constraints, candidate parking paths are generated. The candidate parking paths are represented as follows: In the formula, This represents the candidate parking path generated from the path parameter vector. Indicates the initial vehicle status. Indicates the status of the vehicle at the end of the path. Indicates the sampling point number at the end of the path.

[0057] Candidate parking paths are sampled according to path arc length or control cycle. Each sampling point records the vehicle reference point coordinates, vehicle heading angle, vehicle speed, vehicle steering angle, vehicle direction of movement, and current path segment type. The current path segment type includes at least one of the following: straight segment, gradually increasing curvature segment, fixed curvature segment, gradually decreasing curvature segment, straightening segment, end alignment segment, forward correction segment, and secondary reversing segment.

[0058] S203. Calculate the positions of the vehicle's key corner points.

[0059] The key corner points of a vehicle include at least the front left corner, the front right corner, the rear left corner, and the rear right corner. The coordinates of each key corner point of the vehicle in the vehicle body coordinate system are represented as follows: In the formula, Indicates the first The coordinates of each key corner point of the vehicle in the vehicle body coordinate system. This represents the longitudinal coordinate of the critical corner point relative to the vehicle reference point. This represents the lateral coordinates of the critical corner point relative to the vehicle reference point. Indicates the critical corner point number of the vehicle.

[0060] No. The key corner points of the vehicle at the first The position of each sampling point is represented as follows: In the formula, Indicates the first The key corner points of the vehicle at the first The location of each sampling point and Indicates the vehicle reference point at the 1st d'h. The coordinates of each sampling point Indicates the vehicle is in The heading angle at each sampling point This represents a two-dimensional rotation matrix determined by the vehicle's heading angle.

[0061] S204. Calculate the original clearance and nearest boundary of the vehicle's key corner points.

[0062] For each sampling point, calculate the distance from each vehicle's key corner point to each environmental boundary in the environmental boundary set. The calculation of the distance from a point to a line segment, polyline, or local curve is a conventional geometric calculation, and its specific formula will not be elaborated here.

[0063] No. The key corner points of the vehicle at the first The original clearance at each sampling point is represented as follows: In the formula, Indicates the first The key corner points of the vehicle at the first The original clearance at each sampling point Indicates the first The key corner points of the vehicle at the first From the sampling point to the first The distance to the environmental boundary Indicates the number of environmental boundaries. Indicates the environmental boundary number.

[0064] The nearest boundary number is represented as: In the formula, Indicates the first The key corner points of the vehicle at the first The nearest boundary number corresponding to each sampling point.

[0065] The nearest boundary type is represented as: In the formula, Indicates the first The key corner points of the vehicle at the first The nearest boundary type corresponding to each sampling point This indicates the environment boundary type corresponding to the most recent boundary number.

[0066] S205. Calculate the clearance correction for key corner points of the vehicle.

[0067] The clearance correction is used to correct for the effects of environmental boundary errors, vehicle horizontal positioning errors, vehicle heading angle errors, steering angle errors, and steering actuator response hysteresis on the original clearance. In one specific implementation, the clearance correction is expressed as: In the formula, Indicates the first The key corner points of the vehicle at the first Net clearance correction at each sampling point This indicates the boundary error corresponding to the nearest boundary number. Indicates the first Vehicle planar positioning error at each sampling point Indicates the first The distance vector from each vehicle's key corner point to the nearest boundary is a unit vector. Represents a two-dimensional antisymmetric matrix. Indicates the vehicle's heading angle error. Indicates the vehicle's steering angle. This indicates the amount of steering angle disturbance. Indicates steering angle error. Indicates the first The cumulative path arc length corresponding to each sampling point Indicates vehicle speed. Represents the steering actuator response lag time, where the two-dimensional antisymmetric matrix... for .

[0068] S206. Calculate the corrected clearance at the critical corner points of the vehicle.

[0069] The corrected clearance at the critical corner points of the vehicle is expressed as follows: In the formula, Indicates the first The key corner points of the vehicle at the first Corrected clearance at each sampling point This indicates the original clearance of the vehicle's key corner points at the sampling point. This indicates the clearance correction amount for the vehicle's critical corner points at the sampling point.

[0070] After processing by S200, each candidate parking path corresponds to a set of path sampling point data. This path sampling point data includes vehicle status, path segment type, vehicle key corner point position, original clearance at the vehicle key corner point, nearest boundary number, nearest boundary type, clearance correction amount, and corrected clearance at the vehicle key corner point. This data is used by S300 to establish a corner point clearance phase table.

[0071] Step S300 converts the corrected clearance at each sampling point into the clearance phase state distributed along the candidate parking path, and identifies the sources of risk, low clearance triggering states, and phase conflict segments.

[0072] S301. Determine the candidate dominant corner point.

[0073] At each sampling point, the corrected clearance of each vehicle's critical corner point is compared, and the vehicle's critical corner point with the smallest corrected clearance is determined as the candidate dominant corner point. In the formula, Indicates the first Candidate dominant corner points at each sampling point Represents the set of critical corner points of the vehicle. Indicates the critical corner point number of the vehicle. Indicates the first The key corner points of the vehicle at the first Corrected clearance at each sampling point.

[0074] The minimum corrected clearance is the minimum value of the corrected clearance at each vehicle's critical corner point, used to determine the low clearance point and the boundary proximity status.

[0075] S302. Establish a corner clearance phase table.

[0076] The corner clearance phase table records at least the sampling point number, vehicle status, path segment type, corrected clearance for each vehicle's critical corner, candidate dominant corner, and the nearest boundary number and type corresponding to the candidate dominant corner. This corner clearance phase table is used for clearance phase latching and release, corner-boundary risk pairing, low clearance non-triggering zone filtering, and phase conflict segment identification.

[0077] S303, Perform clearance phase latch and release.

[0078] The latched clearance phase represents the vehicle's critical corner point that plays a dominant role in clearance constraints within a continuous path interval. The latched clearance phase value is taken as a corner point number in the set of vehicle critical corner points; when no latched clearance phase is formed, it is recorded as zero.

[0079] When the same candidate dominant corner point appears consecutively within the latch decision window, and the current path segment belongs to an allowed latching path segment, or the current minimum correction clearance is lower than the boundary and close to the threshold, the candidate dominant corner point is determined as the latch clearance phase. In the formula, Indicates the first Candidate dominant corner points at each sampling point Indicates the critical corner point number of the vehicle. Indicates the length of the latching decision window. Indicates the first The path segment type to which each sampling point belongs. This represents the set of path segment types that are allowed to be latched. Indicates the first Minimum corrected clearance at each sampling point This indicates that the boundary is close to the threshold. When the above conditions are met, the latch clearance phase takes the value. .

[0080] When the previous sampling point already has a latched clearance phase, and new candidate dominant corner points appear consecutively within the release judgment window, the current path segment belongs to a phase-switching allowed path segment, and both the original latched corner point and the new candidate dominant corner point meet the phase-switching safety conditions at the switching position, the latched clearance phase is switched to the new candidate dominant corner point: In the formula, Indicates a new candidate dominant corner. Indicates the length of the release judgment window. This represents the set of path segment types that allow phase switching. This indicates the corrected clearance at the original latched corner point. This indicates the corrected net clearance of the new candidate dominant corner point. This represents the phase switching safety threshold. If the above conditions are not met, the latch clearance phase of the previous sampling point is maintained.

[0081] S304. Establish corner point-boundary risk pairing relationships.

[0082] When the latched clearance phase is not zero, the latched clearance phase, the nearest boundary number corresponding to the latched clearance phase, the nearest boundary type, and the current path segment type are combined to form a corner-boundary risk pair: In the formula, Indicates the first Corner-boundary risk pairing at each sampling point Indicates the latch clearance phase. Indicates the nearest boundary number corresponding to the latch clearance phase. Indicates the nearest boundary type corresponding to the latch net phase. Indicates the type of the current path segment.

[0083] When latch clearance phase exists at adjacent sampling points, boundary correlation change is determined. If the dominant corner point remains unchanged but the nearest boundary changes, or vice versa, the boundary correlation change flag is set to one; otherwise, it is set to zero. When the latch clearance phase at any adjacent sampling point is zero, the boundary correlation change flag is set to zero, and the nearest boundary number is not invoked.

[0084] S305, Perform low headroom non-trigger zone filtering.

[0085] Low clearance points are sampling points where the minimum corrected clearance is lower than the minimum clearance threshold. Hard safety points are sampling points where the minimum corrected clearance is lower than the hard safety clearance threshold. Hard safety points are not included in non-trigger filtering but are used as triggers for replanning, path segment reconstruction, or parking control.

[0086] For low clearance points that are not hard safety points, filtering is performed based on path correction trigger markers: In the formula, Indicates the path correction trigger flag. Indicates phase continuity triggering flag, Indicates the boundary association trigger flag, Indicates the phase switching neighborhood trigger flag, Indicates the execution sensitivity trigger flag, This indicates a logical OR.

[0087] The phase continuity trigger flag is determined based on whether the same latched clearance phase continuously appears within the judgment window before and after the low clearance point; the boundary association trigger flag is determined based on whether the same nearest boundary continuously appears within the judgment window; the phase switching neighborhood trigger flag is determined based on whether the low clearance point is located in the neighborhood before and after the phase switching point; the execution sensitivity trigger flag is determined based on whether the low clearance point is located in the curvature change path segment or the actuator hysteresis influence path segment.

[0088] When the path correction trigger flag is zero, the sampling point is assigned to the low clearance non-triggered zone; when the path correction trigger flag is one, the sampling point is assigned to the path correction trigger zone. The low clearance non-triggered zone does not directly trigger path parameter correction, but continues to participate in phase continuity judgment, boundary association judgment, and phase conflict segment identification.

[0089] S306, Identify phase conflict segments.

[0090] A phase conflict segment indicates that within the same path interval, multiple vehicle critical corner points alternately become latching clearance phases, or multiple nearest boundaries alternately become sources of risk within the same path interval. Let the path interval to be judged be: In the formula, This represents the path interval to be determined. Indicates the starting sampling point number. Indicates the endpoint sampling point number.

[0091] When at least two non-zero latch clearance phases appear within a path interval, and at least one of the following conditions is met: the number of nearest boundaries is not less than two, the number of phase switching is greater than the upper limit of the number of phase switching, and the number of boundary association changes is greater than the upper limit of the number of boundary association changes, the path interval is determined as a phase conflict segment: In the formula, This represents the set of non-zero latch net void phases occurring within the path interval. This represents the set of nearest boundary numbers that occur within the path interval. Indicates the number of phase switches within the path interval. Indicates the number of boundary association changes within the path interval. Indicates the upper limit of the number of phase switching. This indicates the upper limit of the number of boundary-related changes. Both the number of phase switching and the number of boundary-related changes are based on the statistics of sampling points corresponding to non-zero latch clearance phases.

[0092] After processing by S300, the following are obtained: corner clearance phase table, latched clearance phase, corner-boundary risk pairing relationship, low clearance non-triggering zone, path correction triggering zone, and phase conflict segment. These results are used by S400 to determine path parameter correction authority, calculate path parameter sensitivity, and screen candidate path parameter correction schemes.

[0093] Step S400 is used to limit the path parameters that can be adjusted under the current risk status and to predict and filter candidate path parameter correction schemes. Hard safety points are not included in the regular parameter correction process of this step, but are directly used as the trigger for replanning, path segment reconstruction, or parking control.

[0094] S401, Determine the permissions for path parameter correction.

[0095] For sampling points within the path correction trigger zone where the latch clearance phase is not zero, the path parameter correction permission table is queried based on the latch clearance phase, nearest boundary type, and current path segment type corresponding to the sampling point. The input to the path parameter correction permission table includes the latch clearance phase, nearest boundary type, and current path segment type, and the output is the set of path parameters that are allowed to be corrected.

[0096] The path parameter correction permission table is represented as follows ,in Indicates the first The set of path parameters that can be corrected at each sampling point. Indicates the first Latch clearance phase at each sampling point Indicates the nearest boundary type corresponding to the latch net phase. Indicates the first The path segment type to which each sampling point belongs. This represents the set of path parameter numbers.

[0097] When multiple corner-boundary risk pairs exist within the path correction trigger zone, the trigger sampling points are grouped according to the latch clearance phase, the nearest boundary type, and the current path segment type. The same permission set is used within the same group, while permission sets corresponding to different groups participate in sensitivity calculations separately. When a candidate path parameter correction scheme takes effect across multiple groups, its non-zero correction parameters should belong to the allowed correction set of at least one group and must not belong to the prohibited correction parameters of the corresponding group.

[0098] When the latched clearance phase corresponds to the front outer corner point, the nearest boundary type is the adjacent vehicle boundary, and the current path segment is the initial reversing turning segment, the path parameters that can be corrected include at least one of the following: the reversing starting position, the length of the first turning segment, and the length of the curvature gradually increasing segment.

[0099] When the latched clearance phase corresponds to the inner corner point, the nearest boundary type is the inner boundary of the parking space, and the current path segment is a fixed curvature segment, the path parameters that can be corrected include at least one of the following: fixed curvature segment length, return to center starting point, and low curvature transition segment parameters.

[0100] When the latched clearance phase corresponds to the rear corner point, the nearest boundary type is a limiter boundary, wall boundary, or general obstacle boundary, and the current path segment is an end-aligned segment, the path parameters that can be corrected include at least one of the following: end-aligned segment length, target reference point longitudinal position correction amount, and end-alignment parameter.

[0101] When the sampling point is located within a phase conflict segment, the path parameters that can be corrected also include path segment structure parameters. Path segment structure parameters include at least one of the following: fixed curvature segment split position, low curvature transition segment length, straightening segment split position, short forward correction segment length, and secondary reversing segment length.

[0102] S402, Calculate path parameter sensitivity.

[0103] Within the scope of path parameter correction permissions, calculate the impact of path parameters on the corrected clearance of critical corner points of vehicles corresponding to the latched clearance phase. Path parameter sensitivity is used to represent the direction and degree of change in the corrected clearance of the dominant risk corner point within the trigger zone after a change in path parameters.

[0104] The permission mask is represented as: In the formula, Indicates the first The path parameter in the th... Permission mask at each sampling point Indicates the path parameter number. Indicates the latch clearance phase. Indicates the nearest boundary type corresponding to the latch net phase. Indicates the type of the current path segment.

[0105] For path parameters with a permission mask of one, sensitivity is calculated using a small perturbation method: In the formula, Indicates the first The path parameter in the th... The sensitivity of the latched clearance phase at each sampling point to the correction clearance at the vehicle's critical corner points. Indicates the first The corrected clearance for the vehicle's key corner points corresponding to the latched clearance phase at each sampling point. Represents a path parameter vector. Indicates the first The perturbation of each path parameter Indicates the first Unit vectors in the directions of each path parameter.

[0106] Overall sensitivity is expressed as: In the formula, Indicates the first The overall sensitivity of each path parameter This represents the set of trigger sampling points where the latch clearance phase is not zero. Indicates the first The weight of each sampling point Indicates the permission mask. This indicates the sensitivity of path parameters.

[0107] The sampling point weights are determined based on the low clearance level, whether it is located in the phase switching neighborhood, whether it is located in the phase conflict segment, and the path segment type. The overall sensitivity is used to generate candidate corrections for increasing or decreasing directions, and whether to use candidate corrections is based on the screening results of the phase consequence prediction table.

[0108] S403. Generate candidate path parameter correction schemes.

[0109] Candidate parameters are selected from the allowed path parameters based on overall sensitivity, and one or more candidate path parameter correction schemes are generated. Each candidate path parameter correction scheme includes a correction amount for at least one path parameter. Path parameters not allowed by the permission table have their correction amount set to zero.

[0110] When multiple trigger groups give the same correction direction for the same path parameter, the path parameter is included in the candidate correction scheme; when multiple trigger groups give opposite correction directions for the same path parameter, the correction amount of the path parameter is not directly adopted, or the correction amount of the parameter is set to a small amplitude and then submitted to the phase consequence prediction table for screening.

[0111] S404. Establish a phase consequence prediction table.

[0112] For each candidate path parameter correction scheme, a local candidate path is regenerated within the prediction window, and the corrected clearance at vehicle key corner points, candidate dominant corner points, latched clearance phase, nearest boundary number, boundary association change marker, and phase conflict status are recalculated. The prediction window starts from the beginning of the path correction trigger area or the beginning of the phase conflict segment, covering several subsequent sampling points or several subsequent path segments.

[0113] The phase consequence prediction table records at least the prediction correction clearance, prediction latch clearance phase, prediction nearest boundary number, prediction boundary association change marker, prediction phase conflict status, prediction terminal attitude status, and prediction execution constraint status for each sampling point within the prediction window.

[0114] S405, Filter candidate path parameter correction scheme.

[0115] The candidate path parameter correction scheme is determined based on the phase consequence prediction table to ensure it meets the constraints. These constraints include subsequent correction clearance constraints, phase switching constraints, boundary association constraints, phase conflict constraints, vehicle execution constraints, and target parking attitude constraints.

[0116] The consequence evaluation value of the candidate path parameter correction scheme is expressed as follows: In the formula, Indicates the first The consequence evaluation value of each candidate path parameter correction scheme. This indicates that the net airspace assessment item will be revised subsequently. This indicates the phase switching evaluation item. Indicates the boundary correlation evaluation item. This indicates the phase conflict evaluation item. This indicates the vehicle's performance constraint evaluation item. This indicates the target parking posture evaluation item. to This indicates the weight of each evaluation item.

[0117] If a candidate path parameter correction scheme results in a hard safety point within the prediction window or causes the vehicle execution constraints to be unmet, that candidate path parameter correction scheme will not be adopted. When multiple candidate path parameter correction schemes satisfy the constraints, the candidate path parameter correction scheme with the smaller consequence evaluation value will be selected as the screening result.

[0118] After processing in S400, the filtered candidate path parameter correction schemes are obtained. If a candidate path parameter correction scheme that satisfies the constraints exists, the path parameters are updated in S500 according to the scheme; if no candidate path parameter correction scheme that satisfies the constraints exists, and phase conflict segments have been identified in S300, path segment reconstruction is performed on the phase conflict segments in S500.

[0119] S500 executes the candidate path parameter correction scheme obtained in S400. If a candidate path parameter correction scheme that satisfies the constraints exists, the path parameters are updated according to the scheme; if no candidate path parameter correction scheme that satisfies the constraints exists, and there are phase conflict segments, path segment reconstruction is performed on the phase conflict segments. If the updated or reconstructed path satisfies the output constraints, it is output as the automatic parking path.

[0120] S501. Update the path parameters based on the filtering results.

[0121] When S400 selects candidate path parameter correction schemes, it applies these schemes to the current path parameters to obtain the updated path parameters: In the formula, This represents the updated path parameter vector. This represents the path parameter vector before the update. This indicates the selected candidate path parameter correction scheme.

[0122] After the path parameters are updated, candidate parking paths are regenerated according to S200, and the vehicle's key corner positions, original clearance, nearest boundary, clearance correction, and corrected clearance are recalculated. The recalculated data continues to be used for phase state updates and consequence predictions according to S300 and S400. Path parameters not permitted by the path parameter correction permission table remain unchanged.

[0123] S502, Perform path segment reconstruction on phase conflict segments.

[0124] When the candidate path parameter correction scheme fails to meet the constraints, and the candidate parking path contains phase conflict segments, path segment reconstruction is performed on the phase conflict segments. Path segment reconstruction is used to change the combination of path segments within the phase conflict segments, thereby redistributing the dominant relationship between the vehicle's critical corner points and the environmental boundary.

[0125] Path segment reconstruction includes at least one of the following: fixed curvature segment splitting, insertion of low curvature transition segments, straightening segment splitting, insertion of short forward correction segments, and generation of secondary reversing segments. Fixed curvature segment splitting refers to dividing the original fixed curvature segment into two or more curvature-preserving intervals; low curvature transition segments refer to inserting transitional path segments with lower absolute curvature values ​​between adjacent path segments; straightening segment splitting refers to dividing a single straightening process into two or more straightening intervals; short forward correction segments refer to inserting attitude correction path segments in the forward direction during reversing parking; and secondary reversing segments refer to regenerating the reversing parking path segment after the short forward correction segment.

[0126] When a phase conflict segment is caused by the alternating occurrence of the front and rear corner points as latched clearance phases, a fixed curvature segment is used to split the segment or a low curvature transition segment is inserted. When a phase conflict segment is caused by frequent changes in the nearest boundary corresponding to the same corner point, a straightening segment is used to split the segment or a low curvature transition segment is inserted. When the current reversing path segment cannot eliminate the phase conflict segment, a short forward correction segment or a secondary reversing segment is generated.

[0127] After path segment reconstruction, the vehicle's critical corner point correction clearance, latched clearance phase, corner-boundary risk pairing relationship, low clearance trigger state, and phase conflict state are recalculated. If the reconstructed path still has hard safety points or still does not meet vehicle execution constraints, the path is not output.

[0128] S503, Execution path output constraint verification.

[0129] Constraint validation is performed on the updated or reconstructed parking path. The constraints include corner clearance constraints, phase switching constraints, boundary association constraints, vehicle kinematic constraints, vehicle execution constraints, and target parking attitude constraints.

[0130] Corner clearance constraints must include at least the corrected clearance at the vehicle's critical corners being no less than the hard safety clearance threshold. In the formula, Indicates the first element in the path to be output. The key corner points of the vehicle at the first Corrected clearance at each sampling point This indicates the hard safety clearance threshold.

[0131] For sampling points below the minimum clearance threshold but not below the hard safety clearance threshold, these sampling points should belong to the low clearance non-trigger zone and not be located in phase conflict segments or execution-sensitive path segments. Phase switching constraints are used to confirm that phase switching points in the output path meet the phase switching safety threshold. Boundary association constraints are used to confirm that there are no unprocessed boundary association abrupt changes in corner-boundary risk pairs in the output path. Vehicle kinematic constraints and vehicle execution constraints are used to confirm that the output path meets the requirements for steering angle, steering angle change rate, vehicle direction of motion, and path segment continuity. Target parking attitude constraints are used to confirm that the deviations between the vehicle reference point coordinates and heading angle at the end of the path and the vehicle's target parking attitude do not exceed the allowable deviations.

[0132] For the generation of vehicle chassis control commands, steering control, braking control and drive control, those skilled in the art can complete them using existing low-speed parking control methods based on the vehicle controller interface. The conversion of control commands is a well-known technology in the field and will not be described in detail here.

[0133] S504, Output automatic parking path.

[0134] When the updated or reconstructed parking path satisfies the constraints in S503, the parking path is output as the automatic parking path. The output automatic parking path includes the vehicle reference point trajectory, path segment type, vehicle direction of movement, target speed, target steering angle, vehicle critical corner point correction clearance, latched clearance phase, and corner point-boundary risk pairing relationship.

[0135] The vehicle controller executes parking control according to the output automatic parking path and retains the latched clearance phase and corner-boundary risk pairing relationship for deviation judgment during the execution process.

[0136] S505, Closed-loop update during execution.

[0137] When the vehicle executes the automatic parking path, it continuously acquires information on vehicle pose, heading angle, wheel speed, steering angle and environmental boundary, and updates the vehicle's critical corner point correction clearance, latched clearance phase and corner point-boundary risk pairing relationship.

[0138] When the deviation between the actual vehicle state and the planned vehicle state exceeds the path correction threshold, and the actual corrected clearance is not lower than the hard safety clearance threshold, local path parameter correction is performed according to the path parameter correction authority and path parameter sensitivity in S400.

[0139] When a new phase conflict segment is formed during execution, and the local path parameter correction cannot eliminate the phase conflict segment, path segment reconstruction is performed on the phase conflict segment.

[0140] When the actual clearance at critical corner points of the vehicle falls below the hard safety clearance threshold, or when changes in the environmental boundary cause the original automatic parking path to no longer meet the constraints, the execution of the original automatic parking path is stopped, and replanning or parking control begins. Parking control includes setting the target vehicle speed to zero, maintaining braking, stopping path tracking, maintaining the vehicle's current gear or shifting to a safe gear, and outputting a parking stop status to the vehicle control system.

[0141] After processing by the S500, the automatic parking path completes path parameter updates or path segment reconstruction before output, and performs closed-loop updates during execution based on vehicle status, environmental boundaries, and key corner points of the vehicle to correct clearance.

[0142] The above formulas are all dimensionless calculations, and the preset parameters in the formulas should be set by those skilled in the art according to the actual situation.

[0143] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0144] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0145] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An automatic parking path planning method for autonomous vehicles, characterized in that, Includes the following steps: Acquire parking environment information, establish a local parking environment model, and determine the target parking posture of the vehicle; Candidate parking paths are generated based on vehicle kinematic constraints and path parameters. The candidate parking paths are sampled, and the corrected clearance of the vehicle's key corner points at each sampling point is calculated. Based on the corrected clearance, a corner clearance phase table is established, and clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification are performed. The path parameter correction authority is determined based on the latch clearance phase, the nearest boundary type, and the current path segment type. The path parameter sensitivity is calculated, and candidate path parameter correction schemes are screened based on the phase consequence prediction table. Update the path parameters based on the filtering results or reconstruct the path segment for phase conflict segments, and output the automatic parking path.

2. The automatic parking path planning method for autonomous vehicles according to claim 1, characterized in that, The local parking environment model includes a set of environment boundaries, and each environment boundary in the set of environment boundaries is configured with a boundary number, boundary geometry, boundary type, boundary error, boundary valid state, and boundary update time; The boundary types include at least one of the following: parking space boundary, adjacent vehicle boundary, wall boundary, pillar boundary, curb boundary, limiter boundary, and general obstacle boundary. When the confidence level of boundary type identification is lower than the preset condition, or when the boundary category is inconsistent within a continuous planning period, the corresponding environmental boundary will be used as a general obstacle boundary in the correction clearance calculation of the vehicle's key corner points.

3. The automatic parking path planning method for autonomous vehicles according to claim 1, characterized in that, The key corner points of the vehicle include at least the front left corner point, the front right corner point, the rear left corner point, and the rear right corner point; The calculation of the corrected clearance of the vehicle's key corner points at each sampling point includes: determining the coordinates of the vehicle's key corner points in the vehicle's body coordinate system based on the vehicle's external dimensions; The positions of the vehicle's key corner points in the parking local coordinate system are determined based on the vehicle reference point coordinates and the vehicle heading angle at the sampling points. Calculate the distance from the vehicle's key corner points to each environmental boundary, take the minimum distance as the original clearance, and record the nearest boundary number and the nearest boundary type; calculate the corrected clearance based on the original clearance and the clearance correction amount.

4. The automatic parking path planning method for autonomous vehicles according to claim 1, characterized in that, The clearance correction is determined based on the boundary error of the nearest boundary, the vehicle's planar positioning error, the vehicle's heading angle error, the steering angle error, and the steering actuator response hysteresis. Among them, the impact of vehicle heading angle error on clearance is determined based on the position of the vehicle's key corner points relative to the vehicle's reference point; the impact of steering angle error on clearance is determined based on the corrected clearance changes before and after steering angle disturbance; and the impact of steering actuator response lag on clearance is determined based on changes in vehicle speed, response lag time, and clearance along the path direction.

5. The automatic parking path planning method for autonomous vehicles according to claim 1, characterized in that, The corner clearance phase table records at least the sampling point number, vehicle status, path segment type, corrected clearance of each vehicle's key corner point, candidate dominant corner point, nearest boundary number corresponding to the candidate dominant corner point, and nearest boundary type corresponding to the candidate dominant corner point. The candidate dominant corner point is the vehicle's critical corner point with the smallest corrected clearance at the same sampling point; The corner clearance phase table is used to determine the clearance phase status distributed along the candidate parking path and serves as input for clearance phase latching and release, corner-boundary risk pairing, low clearance non-trigger zone filtering, and phase conflict segment identification.

6. The automatic parking path planning method for autonomous vehicles according to claim 5, characterized in that, The net clearance phase latching and release includes: when the same candidate dominant corner point appears consecutively in the latching judgment window, and the current path segment belongs to the allowed latching path segment, or the current minimum corrected net clearance is lower than the boundary and close to the threshold, the candidate dominant corner point is determined as the latched net clearance phase. When new candidate dominant corner points appear consecutively within the release judgment window, the current path segment belongs to the allowed phase switching path segment, and both the original latched corner point and the new candidate dominant corner point meet the phase switching safety threshold at the switching position, the latch clearance phase is switched to the new candidate dominant corner point. If the phase switching conditions are not met, the original latch net phase is maintained.

7. The automatic parking path planning method for autonomous vehicles according to claim 6, characterized in that, The corner-boundary risk pairing includes the latched net space phase, the nearest boundary number corresponding to the latched net space phase, the nearest boundary type corresponding to the latched net space phase, and the current path segment type; When there is a latching clearance phase at adjacent sampling points, the boundary association change marker is determined based on whether the dominant corner point changes and whether the nearest boundary number changes. When no latching clear phase is formed between any of the adjacent sampling points, the nearest boundary number is not used to determine the boundary association change.

8. The automatic parking path planning method for autonomous vehicles according to claim 7, characterized in that, The low headroom non-triggered zone filtering includes: determining sampling points where the minimum corrected headroom is lower than the minimum headroom threshold as low headroom points, and determining sampling points where the minimum corrected headroom is lower than the hard safety headroom threshold as hard safety points. The hard safety point does not enter the low-headroom non-triggered zone filter and serves as the trigger for replanning, path segment reconstruction, or parking control. For low clearance points that are not hard safety points, whether to trigger path correction is determined based on phase continuity, boundary correlation continuity, phase switching neighborhood, and execution sensitivity. Low clearance points that do not trigger path correction are classified into the low clearance non-triggered zone and continue to participate in phase conflict segment identification.

9. The automatic parking path planning method for autonomous vehicles according to claim 8, characterized in that, The phase conflict segment identification includes: counting the non-zero latch clearance phase, the nearest boundary number, the number of phase switching, and the number of boundary association changes within the path interval to be judged; When at least two non-zero latch clearance phases appear in the path interval to be judged, and at least one of the following conditions is met: the number of nearest boundary numbers is not less than two, the number of phase switching is greater than the upper limit of the number of phase switching, and the number of boundary association changes is greater than the upper limit of the number of boundary association changes, the path interval to be judged is determined as a phase conflict segment. The number of phase switching and the number of boundary correlation changes are both based on the statistics of sampling points corresponding to the non-zero latch net phase.

10. The automatic parking path planning method for autonomous vehicles according to claim 1, characterized in that, The path parameter correction permission is determined by the path parameter correction permission table based on the latch clearance phase, the nearest boundary type, and the current path segment type. When multiple corner-boundary risk pairs exist within the path correction trigger area, the trigger sampling points are grouped according to the latch clearance phase, the type of the nearest boundary, and the type of the current path segment. The same group uses the same path parameter correction authority, and different groups participate in the path parameter sensitivity calculation. The path parameter sensitivity is determined by applying a perturbation to the path parameters that are allowed to be corrected and comparing the changes in corrected clearance at the vehicle's key corner points before and after the perturbation. The phase consequence prediction table is used to record the prediction correction clearance, prediction latch clearance phase, prediction nearest boundary number, prediction boundary association change marker, prediction phase conflict state, prediction terminal attitude state, and prediction execution constraint state within the prediction window, and is used to filter candidate path parameter correction schemes that meet the constraints.

Images