Method for path planning for dynamic connection between memory parking cruise and automatic parking and related equipment

Abstract

This application provides a path planning method and related equipment for dynamically connecting memory-based parking cruise and automatic parking. The method includes: acquiring a target memory path, which includes a cruise trajectory generated during the learning phase and a parking trajectory corresponding to the target parking space; generating a real-time driving trajectory during cruise replication based on the cruise trajectory, and determining the degree of overlap between the real-time driving trajectory and the cruise trajectory; based on the degree of overlap, determining, or re-generating, a target connection trajectory that connects with the parking trajectory and satisfies kinematic continuity constraints, from the real-time driving trajectory; and controlling the vehicle to enter the automatic parking phase from the cruise phase based on the target connection trajectory and the parking trajectory to complete the vehicle parking in the parking space. This application eliminates the physical dependence of pause actions or fixed switching points on the transition from the cruise phase to the automatic parking phase, avoids the parking process introduced by phase switching, and improves the continuity of parking task path connection and parking execution efficiency.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a path planning method and related equipment for dynamically connecting memory parking cruise and automatic parking. Background Technology

[0002] With the development of intelligent driving technology, automatic parking systems and memory parking systems have been gradually applied to various intelligent vehicles. Memory parking technology typically records the vehicle's driving route in low-speed scenarios such as parking lots. In subsequent uses, the vehicle automatically cruises along the recorded path and ultimately completes the parking maneuver, thus providing the driver with a more convenient parking experience.

[0003] Existing memory parking systems typically divide the entire parking process into two independent phases: first, the vehicle cruises according to a pre-recorded route; when the vehicle approaches the target parking space, the system switches to automatic parking, generating a parking path using an automatic parking algorithm and controlling the vehicle to complete the parking maneuver. In other words, the cruise phase and the automatic parking phase are usually implemented by different path planning modules. However, in the above-mentioned technical solutions, due to the obvious function switching process between the cruise and automatic parking phases, the vehicle usually needs to be stopped at the transition point between the two phases before the automatic parking planning process is restarted, resulting in a noticeable pause at the transition point. Furthermore, because the automatic parking phase requires replanning the parking trajectory, it often increases the number of parking steps and gear shifts, thereby reducing overall parking efficiency.

[0004] Therefore, how to achieve a more continuous and efficient connection between the memory parking cruise phase and the automatic parking phase has become a technical problem that needs to be solved in this field. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application provides a path planning method and related equipment for dynamically connecting memory parking cruise and automatic parking, which at least solves the problem that the memory parking cruise stage and the automatic parking stage require stopping to switch, resulting in discontinuous connection and low parking efficiency.

[0006] To achieve the above objectives and other advantages, some embodiments of this application provide the following aspects:

[0007] In a first aspect, some embodiments of this application provide a path planning method for dynamically connecting memory parking cruise and automatic parking, including:

[0008] Obtain the target memory path, which includes the cruise trajectory generated during the learning phase and the parking trajectory corresponding to the target parking space;

[0009] During the cruise replication process based on the cruise trajectory, a real-time driving trajectory is generated, and the degree of overlap between the real-time driving trajectory and the cruise trajectory is determined.

[0010] Based on the degree of overlap, a target connection trajectory that connects with the parking trajectory and satisfies the kinematic continuous driving constraint is determined from the real-time driving trajectory or generated by replanning.

[0011] Based on the target connection trajectory and the parking trajectory, the vehicle is controlled to enter the automatic parking phase from the cruise phase to complete the parking of the vehicle into the target parking space.

[0012] Secondly, some embodiments of this application also provide an electronic device, the electronic device comprising:

[0013] One or more processors; and a memory storing computer program instructions that, when executed, cause the processors to perform the path planning method as described above.

[0014] Thirdly, some embodiments of this application also provide a computer-readable storage medium having a computer program and / or instructions stored thereon, which, when executed by a processor, implement the path planning method as described above.

[0015] Fourthly, some embodiments of this application also provide a computer program product, including a computer program and / or instructions that, when executed by a processor, implement the path planning method as described above.

[0016] Compared with existing technologies, the solution provided in this application determines the degree of overlap between the real-time driving trajectory and the cruise trajectory during the cruise replication process, and adaptively determines or re-plans the target connecting trajectory accordingly. This eliminates the physical dependence on pauses or fixed switching points for the transition from the cruise phase to the automatic parking phase, achieving flexible phase switching in continuous driving. Simultaneously, by applying kinematic continuous driving constraints to the target connecting trajectory, a high degree of consistency between the connecting segment and the parking trajectory in terms of geometry and dynamics is ensured. This effectively avoids the parking process introduced by phase switching while maintaining path continuity, improving the continuity of parking task path connection and parking execution efficiency. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other implementation methods can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is one of the flowcharts illustrating a path planning method for dynamically connecting memory parking cruise and automatic parking, as provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the process for generating candidate connection trajectories based on dynamic programming, provided in an embodiment of this application.

[0020] Figure 3 This is a schematic diagram of the system structure provided in the embodiments of this application for realizing the dynamic connection between memory parking cruise and automatic parking;

[0021] Figure 4 This is a second schematic flowchart of a path planning method for dynamically connecting memory parking cruise and automatic parking, provided in an embodiment of this application.

[0022] Figure 5 This is a schematic diagram illustrating the construction of the target memory path in an embodiment of this application;

[0023] Figure 6 This is a schematic diagram of dynamic connection based on the degree of trajectory overlap in the embodiments of this application;

[0024] Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] Some embodiments of this application relate to a path planning method for dynamically connecting memory-based parking cruise and automatic parking, which can be applied to intelligent driving systems with multi-source perception and fusion capabilities. This system acquires information about the vehicle's surrounding environment and its own motion state, and combines this with a pre-learned target memory path to achieve a continuous transition and dynamic connection between the cruise phase and the automatic parking phase. (Refer to...) Figure 1 As shown, the method may include the following steps:

[0027] Step S1: Obtain the target memory path, which includes the cruise trajectory generated during the learning phase and the parking trajectory corresponding to the target parking space.

[0028] During path learning, the vehicle travels along the target path to the target parking space in either manual or assisted driving mode. Real-time data collection and recording of this journey generates a target memory path. In this process, the system acquires vehicle motion state and environmental information from multiple sensors and continuously records the vehicle's pose information at each sampling moment to form the vehicle's trajectory data. The pose information includes at least the vehicle's three-dimensional coordinates and heading. In some implementations, it may also include the vehicle's speed, acceleration, gear position, and cumulative distance from the starting point, thus constituting multi-dimensional attribute data for the trajectory points. The pose information can be determined by fusing real-time differential positioning data, inertial measurement data, and dead reckoning results to improve the accuracy and continuity of trajectory recording.

[0029] Based on the changes in the vehicle's spatial position during travel and its relative position to the target parking space, the driving trajectory is segmented. The trajectory before the vehicle enters the parking area is determined as the cruising trajectory, and the trajectory during the parking maneuver is determined as the parking trajectory corresponding to the target parking space. The parking trajectory represents the path information of the vehicle from its initial parking position to its final parking position in the target parking space.

[0030] Both the cruising trajectory and the parking trajectory can be represented in terms of data structure as a sequence of multiple trajectory points arranged in chronological order. Each trajectory point is indexed by a sequential subscript to reflect the vehicle's travel order along the path. In some implementations, type identification information can also be set for different trajectory segments to distinguish between the cruising phase and the parking phase.

[0031] After extracting the cruise trajectory and berthing trajectory, the connection points between the two trajectories are marked, and the corresponding end pose and start pose information are recorded to provide basic data support for subsequent path connection and continuity constraint processing. The cruise trajectory, berthing trajectory, and connection point information are then stored together to form a target memory path.

[0032] Step S2: During the cruise replication process based on the cruise trajectory, a real-time driving trajectory is generated, and the degree of overlap between the real-time driving trajectory and the cruise trajectory is determined.

[0033] In an optional embodiment, step S2, generating a real-time driving trajectory during cruise replication based on the cruise trajectory, specifically includes:

[0034] Step S201: Using the cruise trajectory as a reference path, perform trajectory sampling along the lateral offset direction and the forward driving direction within the neighborhood of the cruise trajectory to generate multiple target sampling points.

[0035] In this step, the cruise trajectory is parameterized. Specifically, the cruise trajectory can be represented as a discrete sequence of trajectory points distributed along the path arc length parameter, and a local coordinate system (such as the Frenet coordinate system) related to the path is established, where the vertical coordinate represents the cumulative distance along the path direction, and the horizontal coordinate represents the offset relative to the path centerline.

[0036] Based on this, using the projection point of the current vehicle on the cruise trajectory as the reference starting point, the longitudinal position is discretely sampled within a preset forward distance range, and the lateral position is discretely sampled within a preset lateral offset range, thereby constructing a two-dimensional sampling grid in the neighborhood of the cruise trajectory. By combining the longitudinal sampling points and the lateral offset, multiple sampling points can be obtained in the local coordinate system, represented as follows:

[0037]

[0038] in, Indicates the first The arc length position of each sampling point on the cruise trajectory This indicates the lateral offset relative to the centerline of the cruise trajectory.

[0039] To enable the sampling points to be used for subsequent path generation and vehicle control, the sampling points in the local coordinate system need to be mapped to the global coordinate system. The mapping relationship can be expressed as follows:

[0040]

[0041]

[0042]

[0043] in, For the cruise trajectory at the arc length position The coordinates of the reference point at that location This is the corresponding reference heading angle.

[0044] Therefore, the target sampling point set can be obtained:

[0045]

[0046] in, For the target sampling point set, These are the position coordinates of the sampling point in the global coordinate system. This is the corresponding reference heading angle.

[0047] Step S202: Based on the vehicle's current pose, generate multiple candidate paths connecting to each target sampling point.

[0048] In this step, based on the vehicle's current pose, multiple candidate paths connecting to each target sampling point are generated. This can be achieved by path fitting and constraint solving between the vehicle's current state and the target sampling points. Specifically, the vehicle's current pose information is first obtained and represented as:

[0049]

[0050] in, The coordinates of the vehicle's current position. The vehicle's current heading angle. This represents the current curvature.

[0051] For each target sampling point generated in step S201:

[0052]

[0053] Based on the current position of the vehicle As the starting point of the path, with the target sampling point As the endpoint of the path, construct candidate paths that satisfy the boundary constraints.

[0054] In the path modeling process, polynomial curves can be used to parametrically represent the path, for example, constructing parameters related to the path arc length. Polynomial functions:

[0055]

[0056] Among them, coefficient Determined by the following boundary conditions:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] in, This is the path length parameter corresponding to the target sampling point. The reference curvature at the target sampling point.

[0064] By solving the polynomial coefficients above, a continuous path curve connecting the vehicle's current pose and the target sampling point can be obtained. Discrete sampling of this path curve yields a series of trajectory points along the path:

[0065]

[0066] in, This represents the number of discrete sampling points along the path.

[0067] By performing the above path generation process on all target sampling points, a set of multiple candidate paths can be obtained, which is represented as follows:

[0068]

[0069] This completes the path connection modeling from the vehicle's current pose to each target sampling point in the neighborhood of the cruise trajectory, providing a basic path set for subsequent speed planning and trajectory evaluation.

[0070] Step S203: Based on the vehicle's current motion state information, match the corresponding speed curve for each candidate path to form multiple candidate driving trajectories.

[0071] In this step, each candidate path Both can be expressed as parameters related to the path arc length. Space curves:

[0072]

[0073] For each candidate path Construct the corresponding velocity function based on the vehicle's current motion state information. The vehicle's current motion status information includes its current speed. and current acceleration .

[0074] The velocity function satisfies the following constraints:

[0075]

[0076]

[0077]

[0078] Based on this, a mapping relationship between path arc length and time can be constructed:

[0079]

[0080] in, Indicates the vehicle at time Time along the first The driving distance of each candidate path.

[0081] By coupling the path function with the velocity function, the corresponding candidate driving trajectory can be obtained, and the first... Candidate driving trajectories:

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] The resulting set of candidate driving trajectories can be represented as:

[0088]

[0089] in, Indicates the first Candidate driving trajectories; Indicates the first Candidate driving trajectories at time The position coordinates of the corresponding trajectory point in the global coordinate system; This represents the heading angle of the corresponding trajectory point; and These represent the velocity and acceleration of the corresponding trajectory points, respectively. The time parameter represents the trajectory.

[0090] Step S204: Evaluate each candidate driving trajectory based on a preset first cost function to determine the real-time driving trajectory.

[0091] In an optional embodiment, step S204 specifically includes:

[0092] Step S2041: Obtain the first trajectory feature parameters of each candidate driving trajectory. The first trajectory feature parameters include at least: the lateral offset of the trajectory relative to the center line of the cruise trajectory, the spatial distance between the trajectory and environmental obstacles, the curvature change information of the trajectory, the trajectory continuity parameter, and the overlap information between the trajectory and the cruise trajectory.

[0093] Step S2042: Calculate the first centering cost, the first obstacle avoidance cost, the first smoothing cost, the first continuity cost, and the trajectory overlap cost based on the first trajectory feature parameters;

[0094] Step S2043: Using the first cost function, the first centering cost, the first obstacle avoidance cost, the first smoothing cost, the first continuous cost, and the trajectory overlap cost are weighted and summed to obtain the first comprehensive evaluation value corresponding to each candidate driving trajectory.

[0095] Step S2044: From multiple candidate driving trajectories, determine the trajectory with the best first comprehensive evaluation value as the real-time driving trajectory.

[0096] Specifically, regarding the first obtained in step S203 Candidate driving trajectories:

[0097]

[0098] in, For the first The duration of each candidate driving trajectory.

[0099] Map each trajectory point in the candidate driving trajectory to the reference position corresponding to the cruise trajectory, and calculate its lateral offset relative to the centerline of the cruise trajectory:

[0100] Map each sampling point in the candidate driving trajectory to the reference position corresponding to the cruise trajectory, and calculate its lateral offset:

[0101]

[0102] By statistically analyzing the lateral offset of the entire trajectory, we can obtain:

[0103]

[0104] in, Indicates the candidate driving trajectory at time... Lateral offset relative to the centerline of the cruise trajectory; Indicates the center line of the cruise trajectory; This represents the lateral distance calculation function, used to calculate the shortest lateral distance from a trajectory point to the centerline of the cruise trajectory; Indicates the first The average lateral offset of each candidate driving trajectory over the entire time range.

[0105] For each position point in the trajectory at any given time, calculate the minimum distance between it and the set of obstacles in the environment:

[0106]

[0107] in, Represents the set of obstacles in the environment. This represents a spatial distance calculation function used to calculate the distance between a trajectory point and an obstacle; Indicates the first The minimum distance between a candidate driving trajectory and an obstacle.

[0108] Calculate the curvature of the trajectory based on its geometry:

[0109]

[0110] The rate of change of curvature with respect to the path is statistically analyzed to obtain information on curvature changes that reflect the smoothness of the trajectory.

[0111]

[0112] in, Indicates the first Candidate driving trajectories at time The curvature; This represents the rate of change of curvature with respect to time; A statistic representing the degree of change in trajectory curvature, used to characterize the smoothness of the trajectory.

[0113] By comparing the differences between the heading angle and curvature of the candidate driving trajectory starting point and the vehicle's current actual motion state, trajectory continuity parameters characterizing the continuity of the trajectory start are obtained:

[0114]

[0115]

[0116]

[0117] in, and These represent the vehicle's heading angle and curvature at the current moment, respectively. Indicates the first The continuity parameter of the candidate driving trajectory is used to characterize the degree of consistency between the initial state of the trajectory and the current state of the vehicle.

[0118] The candidate driving trajectory is matched with the cruise trajectory within the corresponding path interval, and the spatial deviation or path overlap ratio between the two is calculated to obtain overlap information that characterizes the degree of trajectory following:

[0119]

[0120]

[0121] in, This indicates the coordinates of a reference point on the cruise trajectory at the position corresponding to the time or arc length parameter. This indicates the spatial deviation between the candidate driving trajectory and the cruise trajectory at the corresponding position; This represents the cumulative value of the spatial deviation over the entire trajectory range, used to characterize the degree of overlap between the trajectory and the cruise trajectory.

[0122] For the The candidate driving trajectories have the following first trajectory feature parameters:

[0123]

[0124] Based on the above feature parameters, the following cost functions can be constructed respectively.

[0125] Based on the lateral offset Calculate the cost of the first central tendency:

[0126]

[0127] It is used to characterize the degree of deviation of the candidate driving trajectory from the center line of the cruise trajectory.

[0128] Based on the minimum distance to the obstacle Calculate the cost of the first obstacle avoidance:

[0129]

[0130] in, To prevent tiny positive numbers with a denominator of zero. This cost is used to characterize the safety of the trajectory; the smaller the distance, the greater the cost.

[0131] Based on curvature change statistics Calculate the first smoothing cost:

[0132]

[0133] Used to characterize the smoothness of a trajectory; the more drastic the curvature change, the greater the cost.

[0134] Based on trajectory continuity parameters Calculate the first continuous cost:

[0135]

[0136] It is used to characterize the degree of connection between the candidate trajectory and the current state of the vehicle.

[0137] Based on the cumulative deviation between the trajectory and the cruise trajectory Cost of calculating trajectory overlap:

[0138]

[0139] Used to characterize the degree to which a candidate driving trajectory follows the cruise trajectory.

[0140] After obtaining the above costs, a first cost function can be constructed to comprehensively evaluate each candidate driving trajectory:

[0141]

[0142] in, Indicates the first The first comprehensive evaluation value corresponding to each candidate driving trajectory; These are the weighting coefficients for the corresponding cost items, used to adjust the impact of different evaluation indicators on the comprehensive evaluation results. They can be preset according to actual application needs or determined through empirical parameter adjustment.

[0143] Because different candidate driving trajectories differ in terms of path following performance, safety, smoothness, and continuity with the current vehicle posture, a single evaluation index cannot comprehensively reflect the overall executability of the trajectory. Therefore, a multi-dimensional evaluation system is constructed, including centering cost, obstacle avoidance cost, smoothness cost, continuity cost, and trajectory overlap cost. This system uniformly quantifies the performance of each candidate driving trajectory across multiple performance dimensions and obtains a comprehensive evaluation value through weighted fusion, enabling the candidate driving trajectories to be evaluated comparably under a unified standard. Based on this, the candidate driving trajectory with the optimal first comprehensive evaluation value is selected. This ensures trajectory safety while allowing the vehicle to closely follow the cruising trajectory, and also takes into account driving smoothness and continuity with the current motion state, thereby improving the executability and stability of the trajectory.

[0144] In the solution provided in this application embodiment, steps S2041 to S2044 involve multi-dimensional feature extraction and unified evaluation processing of candidate driving trajectories. Evaluation indicators with different physical meanings are mapped to uniformly measurable costs. A comprehensive evaluation mechanism is constructed through weighted fusion, enabling unified quantitative evaluation of each candidate driving trajectory across multiple dimensions, including path following, safety, smoothness, and motion continuity. This transforms the trajectory selection process from being driven by a single indicator to a holistic optimization process under multi-objective constraints. This method can ensure obstacle avoidance safety while allowing the vehicle to closely follow the cruising trajectory, taking into account both trajectory smoothness and kinematic continuity, thereby effectively avoiding trajectory instability caused by single-indicator decisions or local optimal choices.

[0145] In an optional embodiment, step S2, determining the degree of overlap between the real-time driving trajectory and the cruise trajectory, specifically includes:

[0146] Step S205: Calculate the lateral distance deviation and heading angle deviation of each trajectory point in the real-time driving trajectory relative to the cruise trajectory.

[0147] The system acquires the positional information and corresponding heading information of each trajectory point in the real-time driving trajectory, and maps each trajectory point spatially to the reference path corresponding to the cruise trajectory to determine the cruise trajectory reference point corresponding to each trajectory point. After obtaining the corresponding reference point, the lateral distance deviation of each trajectory point relative to the centerline of the cruise trajectory can be calculated based on the relative positional relationship between the trajectory point and the reference point, thus reflecting the degree of deviation of the real-time driving trajectory from the cruise trajectory in spatial position. Simultaneously, based on the difference between the heading angle of each trajectory point in the real-time driving trajectory and the trajectory tangent direction of the corresponding reference point, the heading angle deviation of each trajectory point is calculated to represent the degree of consistency between the real-time driving direction and the cruise trajectory direction.

[0148] The aforementioned lateral distance deviation and heading angle deviation are statistically processed over the entire trajectory range. For example, by calculating the average value, maximum value, or weighted cumulative value, statistical parameters characterizing the overall deviation of the real-time driving trajectory are obtained. Among them, the lateral distance deviation and heading angle deviation are used to characterize the overall deviation of the real-time driving trajectory throughout the entire driving process.

[0149] Step S206: Obtain the end pose of the real-time driving trajectory and calculate the pose deviation of the end pose relative to the starting pose of the parking trajectory.

[0150] First, the end-point trajectory is extracted from the real-time driving trajectory as the end-point pose, which includes at least the end-point position coordinates, heading angle, and the vehicle's kinematic state parameters. Simultaneously, the starting pose information of the parking trajectory corresponding to the target parking space is obtained. This starting pose also includes the position coordinates, heading angle, and corresponding kinematic state parameters. The kinematic state parameters characterize the vehicle's steering and motion state, including but not limited to parameters such as curvature, steering angle, and angular velocity.

[0151] Based on this, for the end pose and the starting pose, the deviations between the two in terms of end position, heading angle, and kinematic state parameters are calculated. Among them, the position deviation is used to characterize the difference in spatial distance between the end position and the parking starting point, the heading angle deviation is used to characterize the consistency of the vehicle's orientation, and the kinematic state parameter deviation is used to characterize the matching degree of the vehicle in terms of steering and motion continuity.

[0152] Furthermore, the aforementioned deviations can be uniformly measured and processed. For example, through weighted fusion or normalization, the position deviation, heading deviation, and kinematic state parameter deviation can be integrated to obtain the pose deviation, which characterizes the end-of-course connection capability. The smaller the pose deviation, the higher the consistency between the end of the real-time driving trajectory and the beginning of the parking trajectory, and the more conducive it is to achieving a smooth transition; conversely, it indicates that there is a large difference between the two, which needs to be adjusted through subsequent path planning.

[0153] Step S207: Determine the degree of overlap based on the lateral distance deviation, heading angle deviation, and pose deviation.

[0154] The statistical parameters obtained in step S205 that characterize the overall deviation of the real-time driving trajectory include the statistics of lateral distance deviation and heading angle deviation, and the pose deviation that characterizes the end-connection capability obtained in step S206. These parameters describe the real-time driving trajectory from two aspects: the overall trajectory following situation and the end-matching situation.

[0155] Based on this, a unified measurement process is applied to lateral distance deviation, heading angle deviation, and pose deviation. For example, a weighted fusion method is used to integrate the various deviation parameters to obtain an overlap evaluation value that characterizes the degree of matching between the real-time driving trajectory and the cruise trajectory. The weights of each deviation parameter can be set according to the actual application scenario to adjust the influence of different factors on the overlap determination.

[0156] Furthermore, the comparison between the overlap evaluation value and a preset judgment threshold can be used to determine whether the current real-time driving trajectory meets the preset overlap conditions. In one implementation, when the overlap evaluation value is less than or equal to the preset threshold, the overlap between the real-time driving trajectory and the cruise trajectory is determined to meet the preset conditions; when the overlap evaluation value is greater than the preset threshold, the overlap is determined not to meet the preset conditions. In another implementation, the overlap evaluation value can be converted into a normalized index or a similarity index, and compared with the preset threshold based on this index to determine the degree of overlap.

[0157] In the solution provided in this application embodiment, through steps S201 to S207, a neighborhood sampling space is constructed based on the cruise trajectory and multiple candidate driving trajectories are generated. On this basis, through multi-dimensional trajectory feature extraction and unified cost evaluation, a comprehensive trade-off is achieved in trajectory following performance, safety, smoothness, and kinematic continuity, thereby selecting the optimal real-time driving trajectory. At the same time, through statistical evaluation of the overall lateral deviation and heading deviation of the trajectory, and joint analysis of the difference between the end pose and the starting pose of the parking trajectory, an accurate determination mechanism for the degree of trajectory overlap is formed. This enables the system to adaptively predict whether to directly connect or trigger replanning based on the current driving state, thereby improving the accuracy and real-time performance of path connection decisions and enhancing the system's adaptive adjustment capability under environmental disturbances or trajectory deviation.

[0158] In an optional embodiment, to achieve a smooth transition between the cruise phase and the automatic parking phase, transition decision triggering conditions can be pre-set when the vehicle approaches the target parking space, thereby forming a pre-planning window. This allows the system to complete trajectory transition strategy planning before entering the parking area. Even in the event of obstacle avoidance or trajectory deviation, it can still adaptively select or generate a transition path, avoiding the response lag problem caused by switching paths only when approaching the parking position.

[0159] Specifically, the decision to enter the connection decision zone can be determined based on the relative spatial relationship between the vehicle and the target parking space. A vehicle coordinate system is established with the vehicle's current center point as the origin, where the direction of the parking space opening is defined as the positive Y-axis, and the direction perpendicular to it is defined as the X-axis. The lateral and longitudinal distances of the target parking space center relative to the vehicle are calculated. When the lateral and longitudinal distances meet preset range conditions, the vehicle is determined to have entered the connection decision zone. When the vehicle enters the connection decision zone, a trajectory connection decision process is triggered to select a subsequent trajectory connection strategy based on the degree of overlap between the real-time driving trajectory and the cruise trajectory.

[0160] Step S3: Based on the degree of overlap, determine the target connection trajectory from the real-time driving trajectory or generate it through replanning, which connects with the parking trajectory and satisfies the kinematic continuous driving constraints.

[0161] In an optional embodiment, step S3 specifically includes:

[0162] Step S301: When the degree of overlap meets the preset conditions, select the trajectory segment from the real-time driving trajectory whose end pose and the starting pose of the parking trajectory meet the preset kinematic continuity constraints, and determine the trajectory segment as the target connecting trajectory.

[0163] When the overlap meets preset conditions, it indicates that the current real-time driving trajectory is within acceptable limits in terms of both overall path following and end-point connection capability. The preset conditions for meeting the overlap requirements include: the statistical values ​​of the lateral distance deviation and heading angle deviation of the real-time driving trajectory on the cruise trajectory are both less than the corresponding threshold ranges; and the positional deviation, heading deviation, and curvature deviation of the end pose of the real-time driving trajectory relative to the initial pose of the berthing trajectory are all less than the corresponding threshold ranges. If the above conditions are met, it means that the current real-time driving trajectory can not only stably follow the cruise path, but its end state also has the geometric and kinematic basis for direct connection with the berthing trajectory. Furthermore, the trajectory has satisfied obstacle avoidance constraints during its generation process, meaning the minimum distance between the trajectory and environmental obstacles is greater than a preset safe distance threshold. This indicates that the current trajectory has not deviated significantly due to obstacle avoidance and has the basis for direct connection.

[0164] In an optional embodiment, step S301 may include the following steps:

[0165] Step S3011: Traverse each trajectory point in the real-time driving trajectory along the driving direction of the real-time driving trajectory.

[0166] The real-time driving trajectory can be represented as a sequence of trajectory points arranged in chronological or path arc length order, with each trajectory point distributed sequentially according to the vehicle's actual driving direction. In this step, the traversal can be performed according to the time sequence or the increasing order of the path arc length to ensure that the traversal direction is consistent with the vehicle's current driving direction. In some implementations, the traversal can also be performed in reverse from the end closest to the starting position of the parking trajectory to prioritize searching for trajectory points closer to the parking trajectory, thereby improving the efficiency of determining the connection point.

[0167] Step S3012: For each trajectory point, obtain the pose information of the trajectory point and calculate the deviation parameter between the trajectory point and the starting pose of the berthing trajectory.

[0168] Obtaining the pose information of a trajectory point includes at least its spatial coordinates, heading angle, and curvature or steering state parameters. Based on this, the positional deviation, heading deviation, and curvature deviation between the trajectory point and the initial pose of the parking trajectory can be calculated. The positional deviation characterizes the consistency of spatial position, the heading deviation characterizes the consistency of driving direction, and the curvature deviation characterizes the consistency of vehicle steering state, thus comprehensively reflecting the kinematic matching degree of the trajectory point as a candidate connection point.

[0169] Step S3013: Determine whether the position deviation, heading deviation and curvature deviation meet the corresponding preset error ranges.

[0170] The position deviation, heading deviation, and curvature deviation are compared with their respective preset error ranges to determine whether the current trajectory point meets the preset kinematic continuity constraints. Kinematic continuity constraints include at least position continuity constraints, heading continuity constraints, and curvature continuity constraints. Position continuity constraints ensure that the end position of the target connecting trajectory is spatially consistent with the starting position of the parking trajectory, or that its deviation is less than a preset position threshold, thus avoiding spatial jumps in the trajectory. Heading continuity constraints ensure that the difference between the heading angle at the end of the target connecting trajectory and the starting heading angle of the parking trajectory is less than a preset heading threshold, allowing for a smooth transition in vehicle direction. Curvature continuity constraints ensure that the difference between the curvature or steering state at the end of the target connecting trajectory and the starting curvature of the parking trajectory is less than a preset curvature threshold, thus avoiding abrupt changes in steering control.

[0171] When the position deviation is less than a preset position threshold, the heading deviation is less than a preset heading threshold, and the curvature deviation is less than a preset curvature threshold, the trajectory point is determined to satisfy the kinematic continuity constraint. In some implementations, the minimum distance between the trajectory point and obstacles in the environment can be further combined to constrain its safety, so as to avoid selecting trajectory points with large deviations due to obstacle avoidance as connection points, thereby ensuring that the selected trajectory has safety and executability.

[0172] Step S3014: When there is a target trajectory point that satisfies the preset kinematic continuity constraint, the target trajectory point is used as the connection point, and the trajectory segment between the starting point and the connection point in the real-time driving trajectory is extracted as the target connection trajectory.

[0173] The connection point can be selected as the first trajectory point to satisfy the constraints along the traversal direction, or the trajectory point with the smallest deviation from the initial pose of the parking trajectory can be selected as the optimal connection point from all trajectory points that satisfy the conditions. After determining the connection point, the trajectory segment from the current vehicle's starting position to the connection point in the real-time driving trajectory is extracted as the target connection trajectory, and the target connection trajectory is spliced ​​with the parking trajectory to ensure that the splicing point is continuous and consistent in position, heading, and curvature. If necessary, the splicing area can also be locally smoothed to further eliminate minor discontinuities, thereby ensuring the consistency of the trajectory in terms of geometry and vehicle kinematics, so that the vehicle can drive continuously along the target connection trajectory and smoothly enter the parking trajectory.

[0174] In the solution provided in this application embodiment, kinematic continuity driving constraints are introduced in steps S3011 to S3014 to constrain and screen candidate trajectory points. This accurately determines the target connection points that satisfy geometric continuity and vehicle kinematic executability from the current real-time driving trajectory. Based on the target connection points, the trajectory is truncated to directly construct a target connection trajectory that matches the parking trajectory. The above connection process, without triggering additional path replanning, completes the transition between the cruising trajectory and the parking trajectory in the absence of significant trajectory deviation caused by environmental obstacles, thus effectively reducing the number of parking steps and the number of parking operation steps. Furthermore, since this method generates the path based on local trajectory determination and truncation, the overall calculation process is efficient, enabling a smooth and efficient dynamic connection between the cruising phase and the automatic parking phase under continuous driving conditions.

[0175] Step S302: When the degree of overlap does not meet the preset conditions, multiple candidate planning trajectories are generated based on the vehicle's current motion state information and environmental information. Based on the preset second cost function, the trajectory with the best comprehensive evaluation value is determined from the multiple candidate planning trajectories as the target connecting trajectory. The second cost function includes a constraint term used to characterize the kinematic continuity between the trajectory and the parking trajectory, so that the target connecting trajectory satisfies the kinematic continuity driving constraint.

[0176] In practical applications, two typical scenarios may arise: First, the parking trajectory generated during the memory phase may not be the globally optimal trajectory, and its initial pose may deviate from the cruise trajectory, making direct connection during replication difficult. Second, during cruise replication, dynamic or static obstacles in the environment may cause the vehicle to perform obstacle avoidance maneuvers, resulting in a significant deviation of the real-time driving trajectory from the original cruise trajectory. In these situations, attempting to connect according to the original memory trajectory would not only fail to meet the continuity constraints of position, heading, and curvature but may also introduce unnecessary path detours or control abrupt changes, thereby reducing path execution efficiency and driving stability. Therefore, when the overlap does not meet the preset conditions, it can be determined that the current real-time driving trajectory is not suitable for direct connection with the parking trajectory, and a replanning process needs to be triggered to generate a target connection trajectory that matches the current vehicle pose and environmental information.

[0177] In an optional embodiment, refer to Figure 2 As shown, in step S302, multiple candidate trajectories are generated based on the vehicle's current motion state information and environmental information, specifically including:

[0178] Step S3021: Construct a feasible velocity space for the vehicle within a preset time window based on the vehicle's current pose, linear velocity, and angular velocity, and perform velocity sampling within the feasible velocity space to obtain multiple sets of candidate motion control parameters.

[0179] The motion state information of the vehicle at the current moment is obtained, which includes at least the vehicle's current pose, linear velocity, and angular velocity; at the same time, the kinematic constraint parameters of the vehicle body are combined, including maximum linear velocity, minimum linear velocity, maximum angular velocity, minimum angular velocity, as well as acceleration constraints and angular acceleration constraints.

[0180] Within a preset time window, a feasible speed space for the vehicle is constructed based on its current speed state and dynamic constraints. The candidate linear velocity range can be represented as:

[0181]

[0182] in, Indicates the candidate linear velocity; Indicates the vehicle's current linear speed; Indicates the vehicle's maximum acceleration; Indicates the preset time window; , These represent the minimum and maximum permissible linear speeds of the vehicle, respectively.

[0183] The candidate angular velocity range can be expressed as:

[0184]

[0185] in, Indicates the candidate angular velocity; Indicates the vehicle's current angular velocity; Indicates angular acceleration constraints; , These represent the minimum and maximum values ​​of angular velocity, respectively.

[0186] Subsequently, within the feasible velocity space, candidate linear velocities and candidate angular velocities are discretely sampled. For example, the linear velocity interval can be sampled according to a preset sampling interval. Perform uniform sampling, and perform sampling at preset intervals for the angular velocity range. Uniform sampling is performed to obtain multiple discrete velocity combinations.

[0187] By combining candidate linear velocities and candidate angular velocities, multiple sets of candidate motion control parameters are formed. Each set of candidate motion control parameters can be expressed as:

[0188]

[0189] in, Indicates the first Group of candidate motion control parameters Indicates the first Group candidate linear velocity, Indicates the first Group of candidate angular velocities.

[0190] Step S3022: Based on each group of candidate motion control parameters, predict the motion state of the vehicle within a preset time period in the future, and generate the corresponding short-range predicted trajectory.

[0191] During trajectory prediction, based on the candidate motion control parameters obtained in step S3021 and combined with the vehicle kinematics model, the vehicle pose state is updated discretely over time. At this location, the vehicle's position and orientation status is updated to:

[0192]

[0193] Used to update the vehicle's lateral position in the global coordinate system.

[0194]

[0195] Used to update the vehicle's longitudinal position in the global coordinate system.

[0196]

[0197] Used to update the vehicle's heading angle.

[0198] in, , , They represent the first The trajectory corresponding to the group of candidate control parameters at time 1 The pose state. This represents the discrete time step.

[0199] By analyzing time intervals By iteratively calculating at each discrete time point, the result corresponding to the first... The short-range predicted trajectory of the candidate motion control parameters can be expressed as:

[0200]

[0201] in, Indicates the first A short-range predicted trajectory.

[0202] During the prediction process, environmental obstacle information can be combined to filter the effectiveness of the trajectory. For example, when the distance between any trajectory point in the predicted trajectory and an obstacle is less than a preset safety threshold, the corresponding trajectory can be marked as an infeasible trajectory and removed.

[0203] Step S3023: Select at least one target point from each short-range predicted trajectory as the planning starting point. Based on the vehicle kinematics model and environmental obstacle distribution information, perform heuristic path search with the initial pose of the parking trajectory as the target to generate candidate driving paths connecting the target point and the initial pose.

[0204] Based on the short-range predicted trajectories generated in step S3022, a target point is selected from the short-range predicted trajectories as the starting point for path planning. For example, if the endpoint of the trajectory is selected, the target point can be represented as:

[0205]

[0206] in, Indicates the first The predicted termination time of a short-range predicted trajectory.

[0207] In other implementations, multiple key points can be selected from the short-range predicted trajectory as candidate target points based on changes in trajectory curvature, heading, or time intervals to improve the flexibility of path planning.

[0208] The initial pose of the parking trajectory is obtained as the target state for path planning. The initial pose includes the position coordinates, heading angle, and vehicle kinematic parameters, which can be represented as:

[0209]

[0210] in, , Indicates the coordinates of the starting position of the berthing trajectory. This indicates the corresponding heading angle.

[0211] Based on this, a path search space is constructed. This is based on the vehicle's pose state. As a search node, and combined with environmental perception results, obstacles in the environment are modeled as impassable areas. To ensure path safety, obstacle boundaries can be expanded to form obstacle-occupied areas with safety distance constraints, thereby avoiding collisions or close contact between the generated path and obstacles. Vehicle kinematics model constraints are introduced during path search to limit path expansion. Path expansion must meet constraints such as minimum turning radius, curvature continuity, and steering rate of change to ensure the generated path is executable and provides a smooth driving experience.

[0212] Under the premise of satisfying the above constraints, a heuristic path search is performed. The heuristic path search can employ path planning methods guided by heuristic functions, such as the A* algorithm, a hybrid A* algorithm, or a sampling-based path search method. During the path search process, a heuristic evaluation function used to assess the priority of node expansion guides the path search direction, prioritizing the search process towards the initial pose direction of the parking trajectory. Simultaneously, at each node expansion, the newly generated node undergoes a validity check, including: whether it satisfies vehicle kinematic constraints, whether it enters an obstacle area, and whether it satisfies the minimum safe distance constraint. If a node does not meet any of the above conditions, the corresponding path branch is determined to be an infeasible path and pruned.

[0213] As the search process progresses, when the search path reaches or approaches the initial pose of the parking trajectory and meets the preset pose error threshold (including position error and heading error), the search is considered successful, thus obtaining a candidate driving path from the target point to the initial pose of the parking trajectory. The above path search process is performed on the target point corresponding to each short-range predicted trajectory, generating multiple candidate driving paths, thus forming a candidate driving path set. The candidate driving paths, while satisfying obstacle avoidance constraints and vehicle kinematic constraints, can achieve an effective transition from the current driving state to the parking trajectory.

[0214] Step S3024: Concatenate each short-range predicted trajectory with the corresponding candidate driving path to generate multiple candidate planning trajectories.

[0215] The boundary pose information of each short-range predicted trajectory and its corresponding candidate driving path is obtained separately. Before stitching, pose consistency detection is performed on the two to determine the feasibility of connecting the end pose of the short-range predicted trajectory and the starting pose of the candidate driving path. When the pose deviation meets a preset threshold, the short-range predicted trajectory and the candidate driving path can be directly stitched together according to time or path order to obtain a spatially continuous candidate planning trajectory.

[0216] In another implementation, when there is a certain deviation between the end pose of the short-range predicted trajectory and the starting pose of the candidate driving path, but it is still within an adjustable range, a transition trajectory can be inserted between the two to achieve a smooth connection. The transition trajectory can be generated based on a polynomial curve or a spline curve, so that the transition segment meets the continuity constraints in terms of position, heading angle, and curvature, thereby avoiding control instability problems caused by abrupt trajectory changes.

[0217] After completing the trajectory stitching, the stitched trajectory can be parameterized based on the path arc length. Combined with the vehicle's current speed, maximum acceleration, maximum deceleration, and steering constraints, time parameters can be redistributed to each discrete point on the trajectory to generate a time-series trajectory that satisfies the vehicle's kinematic and dynamic constraints.

[0218] Through the above processing, a corresponding candidate planning trajectory can be generated for each short-range predicted trajectory, thereby realizing a continuous transition connection from the current vehicle pose state to the starting pose of the parking trajectory via the short-range predicted trajectory.

[0219] In the solution provided in this application embodiment, through steps S3021 to S3024, a feasible speed space is constructed and sampled based on the vehicle's current motion state to generate multiple short-range predicted trajectories. Combining the vehicle's kinematic model and environmental obstacle distribution information, a candidate driving path connected to the initial pose of the parking trajectory is generated through heuristic path search. Based on this, multiple candidate planned trajectories are formed through trajectory splicing and continuity constraint processing. Therefore, during cruise replication, when environmental obstacles exist or the actual driving trajectory deviates from the memorized cruise trajectory due to obstacle avoidance, the system can still dynamically generate a transition path from the cruise phase to the parking trajectory based on the current state, without relying on the direct reuse of the original memorized trajectory.

[0220] In an optional embodiment, step S302, determining the trajectory with the optimal comprehensive evaluation value from multiple candidate planning trajectories based on a preset second cost function as the target connection trajectory, specifically includes:

[0221] Step S3025: Obtain the second trajectory feature parameters of the short-range predicted trajectory, and calculate the cost values ​​of each item in the first sub-evaluation dimension based on the second trajectory feature parameters. The cost values ​​of each item in the first sub-evaluation dimension include at least: distance cost, second obstacle avoidance cost, gear cost, cornering cost, and second continuity cost.

[0222] Step S3026: Obtain the third trajectory feature parameters of the candidate driving path, and calculate the cost values ​​of each item in the second sub-evaluation dimension based on the third trajectory feature parameters. The cost values ​​of each item in the second sub-evaluation dimension include at least: the second centering cost, the third obstacle avoidance cost, the second smoothing cost, the third continuity cost, and the path distance cost.

[0223] Step S3027: Use the second cost function to weight and fuse the cost values ​​of each item in the first sub-evaluation dimension and the cost values ​​of each item in the second sub-evaluation dimension to obtain the second comprehensive evaluation value of each candidate planning trajectory, and select the candidate planning trajectory with the best second comprehensive evaluation value as the target connection trajectory.

[0224] According to the Trajectory length of a short-range predicted trajectory Calculate the distance cost:

[0225]

[0226] The path length is used to characterize the short-range predicted trajectory; the longer the path, the greater the cost.

[0227] Based on the minimum distance between the short-range predicted trajectory and the obstacle Calculate the cost of the second obstacle avoidance:

[0228]

[0229] in, To prevent tiny positive numbers with a denominator of zero, this cost is used to characterize the safety of short-range predicted trajectories; the smaller the distance, the greater the cost.

[0230] Based on the number of gear changes during the short-range predicted trajectory execution process Calculate the cost of shifting gears:

[0231]

[0232] It is used to characterize the number of gear shifts during the execution of short-range predicted trajectory; the more gear shifts, the greater the cost.

[0233] Based on the statistical values ​​of the angle change of the short-range predicted trajectory Calculate the corner cost:

[0234]

[0235] It is used to characterize the degree of change in the turning angle of a short-range predicted trajectory; the more drastic the change in turning angle, the greater the cost.

[0236] Based on the continuity parameters of the short-range predicted trajectory Calculate the second continuous cost:

[0237]

[0238] It is used to characterize the degree of connection between the short-range predicted trajectory and the current state of the vehicle; the greater the difference, the higher the value.

[0239] After obtaining the above costs, a cost function for the first sub-evaluation dimension can be constructed to evaluate the short-range predicted trajectory:

[0240]

[0241] in, Indicates the first The first sub-evaluation value corresponding to the short-range predicted trajectory; ~ These are the weighting coefficients corresponding to each cost item, used to adjust the influence of different evaluation indicators in the comprehensive evaluation. They can be preset according to actual application needs or determined through empirical parameter tuning.

[0242] Based on the lateral offset of the candidate driving path relative to the centerline of the parking trajectory Calculate the cost of the second centralization:

[0243]

[0244] Used to characterize the degree of deviation of the candidate driving path from the center line of the parking trajectory.

[0245] Based on the minimum distance between the candidate driving path and the obstacle Calculate the cost of the third obstacle avoidance:

[0246]

[0247] in, To prevent tiny positive numbers with a denominator of zero, this cost is used to characterize the safety of candidate driving paths; the smaller the distance, the greater the cost.

[0248] Based on the curvature change statistics of the candidate driving path Calculate the second smoothing cost:

[0249]

[0250] Used to characterize the smoothness of candidate driving paths, the more drastic the curvature change, the greater the cost.

[0251] Based on the continuity parameters of candidate driving paths Calculate the third continuous cost:

[0252]

[0253] It is used to characterize the degree of connection between the candidate driving path and the end of the short-range predicted trajectory.

[0254] Based on the path length of the candidate driving path Calculate the path distance cost:

[0255]

[0256] It is used to characterize the overall length of the candidate driving path; the longer the path, the greater the cost.

[0257] After obtaining the above costs, a cost function for the second sub-evaluation dimension can be constructed to evaluate the candidate driving paths:

[0258]

[0259] in, Indicates the first The second sub-evaluation value corresponding to each candidate driving path; ~ These are the weighting coefficients corresponding to each cost item, used to adjust the influence of different evaluation indicators in the comprehensive evaluation. They can be preset according to actual application needs or determined through empirical parameter tuning.

[0260] After obtaining the first and second sub-evaluation values, a second cost function can be constructed to comprehensively evaluate the candidate planning trajectories.

[0261]

[0262] in, Indicates the first The second comprehensive evaluation value corresponding to each candidate planning trajectory.

[0263] Ultimately, a plan was selected from multiple candidate trajectories. The optimal candidate planning trajectory is used as the target connection trajectory.

[0264] In the solution provided in this application embodiment, through steps S3025 to S3027, a comprehensive balance can be achieved among multiple factors such as path length, obstacle avoidance safety, steering smoothness, shift constraints, and kinematic continuity. This avoids trajectory instability or unexecutability caused by a single evaluation index or local optimal decision. At the same time, even in the presence of obstacle avoidance interference or path deviation, it can still stably select the optimal connecting path that meets vehicle dynamics constraints and has good environmental adaptability, thereby improving the smoothness of the connection between the cruise phase and the automatic parking phase, the execution reliability, and the overall parking efficiency.

[0265] Step S4: Based on the target connection trajectory and parking trajectory, control the vehicle to enter the automatic parking phase from the cruise phase to complete the parking of the vehicle into the target parking space.

[0266] In actual execution, the target connecting trajectory and the parking trajectory are first associated. The end of the target connecting trajectory is set as the target state region corresponding to the initial pose of the parking trajectory. The initial pose of the parking trajectory includes at least the initial position coordinates, the initial heading angle, and the corresponding kinematic state parameters. These kinematic state parameters characterize the vehicle's steering and motion state at the starting point, including information such as curvature, steering angle, or angular velocity. Through these settings, the target connecting trajectory gradually approximates the initial conditions of the parking trajectory in terms of geometry and kinematic state, thereby constructing a transition path from cruising to parking.

[0267] During the vehicle's journey along the cruise trajectory, when the vehicle's current position enters the starting area of ​​the target connecting trajectory and meets the preset trajectory switching conditions, the vehicle's trajectory tracking target is switched from the cruise trajectory to the target connecting trajectory. In this phase, the control system calculates steering and longitudinal control quantities in real time based on the deviation between the vehicle's current position and the target connecting trajectory reference point, causing the vehicle to gradually converge towards and travel along the target connecting trajectory. As the vehicle continues to advance along the target connecting trajectory, its spatial position gradually approaches the starting position of the parking trajectory, its heading angle gradually adjusts to match the starting direction of the parking trajectory, and its curvature or steering state gradually converges to a range matching the starting state of the parking trajectory, thus achieving a gradual transition in the vehicle's motion state.

[0268] It should be noted that, since the target connecting trajectory already satisfies the kinematic continuity constraints during its generation, its end state remains consistent with the initial pose of the parking trajectory in terms of position, heading, and curvature. Therefore, no geometric abrupt changes or kinematic discontinuities occur during trajectory switching. In other words, the end of the target connecting trajectory and the initial pose of the parking trajectory are equivalently aligned within a preset error range, allowing the vehicle to naturally enter the parking trajectory without stopping or additional attitude adjustments.

[0269] In one specific embodiment, such as Figure 3 The vehicle's intelligent driving system shown performs the following: Figure 4 The illustrated path planning process for the dynamic connection between memory parking cruise and automatic parking is presented. This vehicle's intelligent driving system includes a multi-source sensor unit, a domain controller, and an execution and interaction unit. The various functional modules collaborate under the unified scheduling of the domain controller, and the processing flow is as follows:

[0270] The multi-source sensor unit acquires real-time information about the vehicle's surrounding environment and its own status. Among them, the vision sensor and surround-view camera collect image data around the vehicle, and the vehicle data provides motion status information such as the vehicle's current speed and steering wheel angle. Sensors such as ultrasonic, millimeter-wave, inertial measurement units and real-time differential positioning are used to acquire obstacle information and high-precision vehicle pose information.

[0271] Data collected by the multi-source sensor unit is input to the domain controller for processing. The bird's-eye view perception and fusion module spatially maps and fuses the multi-channel visual data to generate a bird's-eye view representation of the vehicle's surrounding environment. Based on vehicle data and information such as inertial measurement and real-time differential positioning, dead reckoning, collision detection, parking space detection, and perception fusion processing are performed in the domain controller to obtain the vehicle's current precise pose, the distribution of environmental obstacles, and target parking space information. The synchronous localization and mapping module fuses the environmental perception information and vehicle pose information obtained from the bird's-eye view perception and fusion module and the dead reckoning and perception fusion module to achieve vehicle localization and local map construction within the environment.

[0272] In the planning and decision-making module, based on the target memory path and the vehicle's current pose information, a cruise replication process is first executed to generate a real-time driving trajectory during the cruise phase. During the cruise, it is determined whether a trajectory connection decision is triggered. If not, the vehicle continues to replicate along the cruise trajectory. If a connection decision is triggered, the degree of overlap between the real-time driving trajectory and the memory cruise trajectory is assessed. If the degree of overlap meets preset conditions, a trajectory segment satisfying the kinematic continuous driving constraint is directly selected from the real-time driving trajectory as the target connection trajectory, and the vehicle enters the parking trajectory along this target connection trajectory to perform parking. If the degree of overlap does not meet preset conditions, a short-range predicted trajectory is generated based on the current vehicle pose information, and a target point is selected from the short-range predicted trajectory. A path search is then performed, combining the vehicle's kinematic model and environmental obstacle information, to generate candidate driving paths connecting the target point and the starting pose of the parking trajectory. Subsequently, the short-range predicted trajectory and the candidate driving paths are concatenated to form multiple candidate planned trajectories, and the optimal candidate planned trajectory is selected as the target connection trajectory through cost function evaluation.

[0273] After obtaining the target connecting trajectory, the planning and decision-making module combines the target connecting trajectory with the parking trajectory to form a complete target parking path, which is then sent to the control module. Based on the pose information of each trajectory point in the target parking path and the corresponding speed, steering, and gear control strategies, the control module generates vehicle control commands, including steering control commands, drive control commands, and braking control commands, and sends them to the vehicle actuators to control the vehicle to complete path tracking and parking operations. During execution, the vehicle transitions along the target connecting trajectory, gradually converging its spatial position, heading angle, and steering state to the initial pose of the parking trajectory. When the vehicle's current state meets the parking trajectory access conditions, it smoothly switches to the parking trajectory, enters the automatic parking stage, and finally completes the parking of the vehicle in the target parking space.

[0274] Meanwhile, the scene reconstruction module, based on the environmental information and vehicle pose information obtained by multi-source perception fusion and combined with the trajectory planning results output by the planning and decision module, performs structured reconstruction of the environment around the vehicle, forming scene expression data containing environmental elements and planned trajectories. The scene expression data is then output to the vehicle display terminal to display the vehicle's surrounding environment and parking process in real time, thereby improving the system's visualization and interaction capabilities.

[0275] By combining global memory path replication with local dynamic programming, a continuous transition mechanism is constructed between the cruising phase and the automatic parking phase, ensuring that there are no obvious pauses or discontinuities in the entire memory parking process. Compared with traditional solutions, this effectively reduces the number of stops and gear shifts during parking, achieving zero-gear shift parking when space conditions permit, thereby significantly improving the continuity of path connection and overall parking efficiency.

[0276] In one specific embodiment, such as Figure 5 As shown, during the route memorization process for parking, the system jointly records the vehicle's driving trajectory during the cruising phase and its parking trajectory upon entering the automatic parking phase, thus forming a target memory path. This target memory path includes the cruising trajectory (shown in blue in the figure) and the parking trajectory (shown in yellow in the figure). The cruising trajectory represents the path information of the vehicle from its starting position to the vicinity of the target parking space, while the parking trajectory represents the automatic parking execution path of the vehicle entering the parking space. Constructing the target memory path in this way provides basic trajectory data support for trajectory overlap determination and dynamic connection strategy selection during subsequent replication driving.

[0277] After the target memory path is constructed, the vehicle enters the path replication stage. For example... Figure 6 As shown, the vehicle replicates the memorized cruise trajectory, thus forming a real-time driving trajectory (shown in green in the figure). Unlike the pre-determined cruise endpoint in the memorization phase, the replicated cruise endpoint is not a fixed position. Instead, it refers to the dynamically determined end position of the trajectory after the vehicle enters the connection decision area along the real-time driving trajectory, based on the current vehicle motion state and the degree of overlap between the real-time driving trajectory and the memorized cruise trajectory. This position serves as a candidate connection point, used to determine whether it can be continuously connected to the starting pose of the parking trajectory.

[0278] When the vehicle approaches the target parking space and enters the connection decision area, the system evaluates the degree of overlap between the real-time driving trajectory and the remembered cruise trajectory. If the real-time driving trajectory and the remembered cruise trajectory have a high degree of overlap, it indicates that the vehicle did not experience significant obstacle avoidance or path deviation during the replication process. At this point, the system selects a trajectory point from the real-time driving trajectory that satisfies preset kinematic continuous driving constraints as a connection point, and uses this connection point as the starting position to directly connect to the remembered parking trajectory (shown in yellow in the figure), thus forming a continuous target connection trajectory. In this way, the vehicle can smoothly transition from the cruise phase to the automatic parking phase without braking or shifting gears, achieving a seamless connection between cruise and parking.

[0279] Furthermore, when environmental obstacles or traffic interference cause the vehicle to circumvent obstacles during the replicated cruise, resulting in a significant deviation between the real-time driving trajectory and the memorized cruise trajectory, and thus the trajectory overlap does not meet the preset conditions, the system no longer uses a fixed connection method but triggers a dynamic replanning mechanism. A short-range predicted trajectory is generated based on the current vehicle pose and motion state, and a target point is selected from the short-range predicted trajectory. Combined with the initial pose of the parking trajectory, a candidate driving path is generated through heuristic path search. Subsequently, the short-range predicted trajectory and the candidate driving path are concatenated, and each candidate planned trajectory is comprehensively evaluated through a cost function. The trajectory that satisfies the kinematic continuity constraint and has the best comprehensive evaluation is selected as the target connection trajectory, thereby achieving adaptive correction for trajectory deviation scenarios.

[0280] Through the above processing method, this embodiment introduces a dynamic connection mechanism based on the degree of trajectory overlap in the path replication stage, which changes the trajectory connection from being driven by a fixed endpoint to being driven by a state. This allows for flexible selection of connection strategies based on real-time driving status, achieving a continuous transition between the cruise stage and the automatic parking stage while ensuring driving safety and kinematic feasibility. This effectively reduces the number of stops and parking operation steps, and improves overall parking efficiency and system robustness.

[0281] In summary, the path planning method for dynamically connecting memory-based parking cruise and automatic parking provided in this application determines the degree of overlap between the real-time driving trajectory and the cruise trajectory during the cruise replication process, and adaptively determines or re-plans the target connecting trajectory accordingly. This eliminates the physical dependence on pauses or fixed switching points for the transition from the cruise phase to the automatic parking phase, achieving flexible phase switching in continuous driving conditions. Simultaneously, by applying kinematic continuous driving constraints to the target connecting trajectory, a high degree of consistency between the connecting segment and the parking trajectory in terms of geometry and dynamics is ensured. This effectively avoids the parking process introduced by phase switching while maintaining path continuity, improving the continuity of parking task path connections and parking execution efficiency.

[0282] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but without changing the core design of the algorithm and process, are also within the scope of protection of this application.

[0283] Furthermore, some embodiments of this application also provide an electronic device. The electronic device can be various forms of digital computer, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device can also be various forms of mobile devices, such as cellular phones, smartphones, wearable devices, and other similar computing devices.

[0284] The electronic device includes: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform a path planning method for dynamically connecting memory parking cruise and automatic parking as provided in any one or more of the above embodiments. Figure 7 An exemplary structural diagram of the electronic device is disclosed. The electronic device includes one or more processors 1101, a memory 1102, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components are interconnected via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple electronic devices can be connected, each providing some of the necessary operations. The components, their connections and relationships, and their functions shown herein are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0285] The electronic device may further include an input device 1103 and an output device 1104. The processor 1101, memory 1102, input device 1103, and output device 1104 may be connected via a bus or other means. Figure 7 Taking the example of a connection between China and Israel via a bus.

[0286] Input device 1103 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the electronic device, such as a touch screen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 1104 may include a display device, auxiliary lighting device (e.g., LED), and haptic feedback device (e.g., vibration motor). The display device may include, but is not limited to, a liquid crystal display, a light-emitting diode display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0287] To provide interaction with the user, the electronic device can be a computer. The computer has: a display device (e.g., a cathode ray tube or LCD monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback); and input from the user can be received in any form (e.g., voice input or tactile input).

[0288] In this embodiment, a computer-readable medium stores a computer program / instruction, which, when executed by a processor, implements a path planning method for dynamically connecting memory-based parking cruise and automatic parking, as provided in any one or more of the above embodiments. This computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist independently and not be assembled into that device. The aforementioned computer-readable medium carries one or more computer-readable instructions.

[0289] The memory 1102 can serve as a non-transitory computer-readable storage medium, used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions, and modules stored in the memory 1102, thereby implementing the program instructions / modules corresponding to the methods provided in any one or more of the embodiments described above in this application.

[0290] The memory 1102 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 1102 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 1102 may optionally include memory remotely located relative to the processor 1101, and these remote memories can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0291] It should be noted that the computer-readable medium described in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections having one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0292] Computer-readable media include permanent and non-permanent, removable and non-removable media, which can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory, static random access memory, dynamic random access memory, other types of random access memory, read-only memory, electrically erasable programmable read-only memory, flash memory or other memory technologies, read-only optical discs, digital versatile optical discs or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0293] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including local area networks (LANs) or wide area networks (WANs), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0294] In the above embodiments, all or part of the implementation can be achieved through software, hardware, firmware, or any combination thereof. For example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of this application can be executed by a processor to implement the above steps or functions. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium, such as RAM memory, magnetic or optical drives, floppy disks, and similar devices. In addition, some steps or functions of this application can be implemented in hardware, for example, as circuitry that cooperates with a processor to perform the various steps or functions.

[0295] The computer program product provided in this application includes one or more computer programs / instructions. When executed by a processor, these computer programs / instructions generate, in whole or in part, the processes or functions described in this application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0296] The flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-specific system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0297] The scope of this application is defined by the appended claims rather than the foregoing description, and is therefore intended to encompass all variations falling within the meaning and scope of equivalents of the claims. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. Terms such as "first," "second," etc., are used only to distinguish descriptions and do not indicate any particular order, nor should they be construed as indicating or implying relative importance.

[0298] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily made 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. Therefore, the scope of protection of this application should be determined by the scope of the claims, and the above embodiments should be regarded as exemplary and non-limiting.

Claims

1. A path planning method for dynamically connecting memory-based parking cruise and automatic parking, characterized in that, include: Obtain the target memory path, which includes the cruise trajectory generated during the learning phase and the parking trajectory corresponding to the target parking space; During the cruise replication process based on the cruise trajectory, a real-time driving trajectory is generated, and the degree of overlap between the real-time driving trajectory and the cruise trajectory is determined. Based on the degree of overlap, a target connecting trajectory is determined from the real-time driving trajectory or generated through replanning, which connects with the parking trajectory and satisfies the kinematic continuity driving constraint, including: When the degree of overlap meets the preset conditions, a trajectory segment that satisfies the preset kinematic continuous driving constraint between the end pose and the starting pose of the parking trajectory is selected from the real-time driving trajectory, and the trajectory segment is determined as the target connecting trajectory. When the degree of overlap does not meet the preset conditions, multiple candidate planning trajectories are generated based on the vehicle's current motion state information and environmental information. The trajectory with the best comprehensive evaluation value is determined from the multiple candidate planning trajectories based on the preset second cost function as the target connecting trajectory. The second cost function includes a constraint term for characterizing the kinematic continuity between the trajectory and the parking trajectory, so that the target connecting trajectory satisfies the kinematic continuity driving constraint. Based on the target connection trajectory and the parking trajectory, the vehicle is controlled to enter the automatic parking phase from the cruise phase to complete the parking of the vehicle into the target parking space.

2. The path planning method according to claim 1, characterized in that, The process of generating a real-time driving trajectory during cruise replication based on the cruise trajectory includes: Using the cruise trajectory as a reference path, trajectory sampling is performed in the neighborhood of the cruise trajectory along the lateral offset direction and the forward driving direction to generate multiple target sampling points; Based on the vehicle's current pose, multiple candidate paths connecting to each of the target sampling points are generated; Based on the vehicle's current motion state information, a corresponding speed curve is matched for each of the candidate paths to form multiple candidate driving trajectories; The candidate driving trajectories are evaluated based on a preset first cost function to determine the real-time driving trajectory.

3. The path planning method according to claim 2, characterized in that, The evaluation of each candidate driving trajectory based on a preset first cost function to determine the real-time driving trajectory includes: The first trajectory feature parameters of each candidate driving trajectory are obtained. The first trajectory feature parameters include at least: the lateral offset of the trajectory relative to the center line of the cruise trajectory, the spatial distance between the trajectory and environmental obstacles, the curvature change information of the trajectory, the trajectory continuity parameter, and the overlap information between the trajectory and the cruise trajectory. Based on the first trajectory feature parameters, calculate the first centering cost, the first obstacle avoidance cost, the first smoothing cost, the first continuity cost, and the trajectory overlap cost respectively. Using the first cost function, the first centering cost, the first obstacle avoidance cost, the first smoothing cost, the first continuous cost, and the trajectory overlap cost are weighted and summed to obtain the first comprehensive evaluation value corresponding to each candidate driving trajectory. From the multiple candidate driving trajectories, the trajectory with the best first comprehensive evaluation value is determined as the real-time driving trajectory.

4. The path planning method according to claim 1, characterized in that, Determining the degree of overlap between the real-time driving trajectory and the cruise trajectory includes: Calculate the lateral distance deviation and heading angle deviation of each trajectory point in the real-time driving trajectory relative to the cruise trajectory; Obtain the end pose of the real-time driving trajectory and calculate the pose deviation of the end pose relative to the starting pose of the parking trajectory. The degree of overlap is determined based on the lateral distance deviation, the heading angle deviation, and the pose deviation.

5. The path planning method according to claim 1, characterized in that, The process of generating multiple candidate trajectories based on the vehicle's current motion state information and environmental information includes: Based on the vehicle's current pose, linear velocity, and angular velocity, a feasible velocity space for the vehicle within a preset time window is constructed, and velocity sampling is performed within the feasible velocity space to obtain multiple sets of candidate motion control parameters. Based on the candidate motion control parameters described in each group, the motion state of the vehicle within a preset time period in the future is predicted, and a corresponding short-range predicted trajectory is generated. At least one target point is selected from each of the short-range predicted trajectories as the planning starting point. Based on the vehicle kinematics model and environmental obstacle distribution information, a heuristic path search is performed with the starting pose of the parking trajectory as the target to generate a candidate driving path connecting the target point and the starting pose. Each short-range predicted trajectory is concatenated with its corresponding candidate driving path to generate multiple candidate planned trajectories.

6. The path planning method according to claim 5, characterized in that, The step of determining the trajectory with the optimal comprehensive evaluation value from the multiple candidate planning trajectories based on a preset second cost function as the target connection trajectory includes: The second trajectory feature parameters of the short-range predicted trajectory are obtained, and the cost values ​​of each item of the first sub-evaluation dimension are calculated based on the second trajectory feature parameters. The cost values ​​of each item of the first sub-evaluation dimension include at least: distance cost, second obstacle avoidance cost, gear cost, cornering cost, and second continuity cost. Obtain the third trajectory feature parameters of the candidate driving path, and calculate the cost values ​​of each item of the second sub-evaluation dimension based on the third trajectory feature parameters. The cost values ​​of each item of the second sub-evaluation dimension include at least: second centering cost, third obstacle avoidance cost, second smoothing cost, third continuity cost, and path distance cost. The second cost function is used to weight and fuse the cost values ​​of each item in the first sub-evaluation dimension and the cost values ​​of each item in the second sub-evaluation dimension to obtain the second comprehensive evaluation value of each candidate planning trajectory, and the candidate planning trajectory with the best second comprehensive evaluation value is selected as the target connection trajectory.

7. An electronic device, characterized in that, The electronic device includes: One or more processors; and a memory storing computer program instructions that, when executed, cause the processors to perform the path planning method as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program and / or instructions stored thereon, characterized in that, When the computer program and / or instructions are executed by the processor, they implement the path planning method as described in any one of claims 1-6.

9. A computer program product, comprising a computer program and / or instructions, characterized in that, When the computer program and / or instructions are executed by the processor, they implement the path planning method as described in any one of claims 1-6.

Images