Driving track generation method, device and electronic equipment based on passable domain constraint

Abstract

The application provides a driving track generation method, device and electronic equipment based on a passable domain constraint. The method comprises the following steps: acquiring a passable domain map, parameter information, a starting point and an ending point of a mobile platform, determining a skeleton path of the mobile platform by using a path search algorithm according to the passable domain map, the starting point and the ending point; determining a passable corridor of the mobile platform according to the parameter information and the skeleton path of the mobile platform, and determining a mobile candidate in the passable corridor, wherein the mobile candidate comprises a center keeping candidate, a left offset candidate and a right offset candidate of the passable corridor; calculating a driving cost of each mobile candidate respectively, and determining a continuous driving track of the mobile platform according to the driving cost and the passable corridor. Thus, the problem that the existing scheme depends on lane semantics in road behavior decision and cannot meet the driving track generation demand in a lane point cloud environment is solved.

Description

Technical Field

[0001] This application relates to the field of driving trajectory generation technology based on traversable domain constraints, and more specifically, to a driving trajectory generation method, apparatus, computer-readable storage medium, and electronic device based on traversable domain constraints. Background Technology

[0002] When ground mobile robots, unmanned inspection vehicles, and low-speed unmanned driving platforms operate in park roads, squares, building interiors, tunnels, ramps, construction sites, and multi-layered point cloud structures, they often cannot reliably obtain lane lines, lane center lines, traffic rule topology, or high-precision lane semantic maps. In such cases, the more common input to the system is a point cloud map, a navigable grid, a hierarchical cost map, or other forms of high-dimensional navigable domains.

[0003] In this type of environment, A Dijkstra, Hybrid A Hierarchical graph search can solve the problem of "whether the origin and destination are geometrically connected", but the search results are usually a discrete skeleton path or geometric center line. For practical robot navigation, a skeleton path alone is not enough: the robot also needs to determine within the passable area whether to proceed along the center, detour to the left, detour to the right, or stop when blocked ahead;

[0004] In summary, existing solutions lack a technical approach for generating driving trajectories in laneless point cloud environments. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, computer-readable storage medium, and electronic device for generating driving trajectories based on traversable domain constraints, so as to at least solve the problem that existing solutions rely on lane semantics for road behavior decisions and have difficulty meeting the requirements for generating driving trajectories in laneless point cloud environments.

[0006] To achieve the above objectives, according to one aspect of this application, a method for generating a driving trajectory based on traversable domain constraints is provided, comprising: acquiring a traversable domain map, parameter information, a start point, and a destination point of a mobile platform; determining a skeleton path of the mobile platform using a path search algorithm based on the traversable domain map, the start point, and the destination point; determining a traversable corridor of the mobile platform based on the parameter information and the skeleton path, and determining movement candidates within the traversable corridor, wherein the movement candidates include center-keeping candidates, left-off candidates, and right-off candidates of the traversable corridor; calculating the driving cost of each movement candidate; and determining a continuous driving trajectory of the mobile platform based on the driving cost and the traversable corridor.

[0007] Optionally, determining the passable corridor of the mobile platform based on the parameter information of the mobile platform and the skeleton path includes: determining the horizontal normal vector at each path point in the skeleton path, scanning the passable domain to the left and right respectively in the direction of the horizontal normal vector to obtain the left and right passable distances of the path points; determining the passable corridor of the mobile platform based on the left and right passable distances and the parameter information, wherein the parameter information includes the width and wheelbase of the mobile platform.

[0008] Optionally, the driving cost of each of the moving candidates is calculated separately, including: determining the safety cost, efficiency cost, deviation cost, and comfort cost of each of the moving candidates; and performing a weighted summation calculation on the safety cost, the efficiency cost, the deviation cost, and the comfort cost to obtain the driving cost of each of the moving candidates.

[0009] Optionally, determining the safety cost, efficiency cost, bias cost, and comfort cost of each of the proposed movement candidates includes: according to a first formula: Determine the security cost of the moving candidate, wherein, For the aforementioned security cost, Indicates target movement candidate The number of collisions with obstacles Indicates the desired safe width. , The penalty coefficient is... This represents the positive part of the function that takes the value of zero when it is less than zero; according to the second formula: Determine the efficiency cost of the proposed move, wherein, For the efficiency cost, Indicates target movement candidate Let B be the length of the centerline, and B be the set of movement candidates within this action decision cycle. This represents any one of the movement candidates in the set of movement candidates; according to the third formula: Determine the deviation cost of the moving candidate, wherein, The cost of the deviation, Indicates target movement candidate The number of path points, The center remains the candidate in the first place. The corresponding centerline point of each path point Describing the Euclidean norm, Indicates target movement candidate In the The centerline point at each path point; according to the fourth formula: Determine the comfort cost of the mobility candidate, wherein, For the aforementioned comfort, This indicates the comfort or behavior switching penalty corresponding to the movement candidate type. This indicates that the center remains a candidate. Indicates the left offset candidate, This indicates the right offset candidate. This indicates that the candidate should be suspended.

[0010] Optionally, determining the continuous driving trajectory of the mobile platform based on the driving cost and the passable corridor includes: determining the candidate with the minimum driving cost as the target candidate, and determining the target trajectory of the mobile platform based on the target candidate and the passable corridor; obtaining the centerline, left and right boundary information, and obstacle information corresponding to the target trajectory, and optimizing the target trajectory based on the centerline, left and right boundary information, and obstacle information to obtain the continuous driving trajectory.

[0011] Optionally, the continuous driving trajectory is obtained by optimizing the target trajectory based on the centerline, the left and right boundary information, and the obstacle information, including: optimizing the target trajectory using the AL-iLQR trajectory optimization algorithm based on the centerline, the left and right boundary information, and the obstacle information to obtain the continuous driving trajectory.

[0012] Optionally, after determining the continuous driving trajectory of the mobile platform based on the driving cost and the passable corridor, the method further includes: projecting the continuous driving trajectory onto the passable corridor to perform corridor projection detection on the continuous driving trajectory, and simultaneously performing heading continuity detection on the continuous driving trajectory to obtain a detection result; if the detection result indicates that the path is passed, controlling the mobile platform to move according to the continuous driving trajectory.

[0013] According to another aspect of this application, a driving trajectory generation device based on traversable domain constraints is provided, comprising: a first determining unit, configured to acquire a traversable domain map, parameter information, start point, and end point of a mobile platform, and determine a skeleton path of the mobile platform using a path search algorithm based on the traversable domain map, the start point, and the end point; a second determining unit, configured to determine a traversable corridor of the mobile platform based on the parameter information and the skeleton path of the mobile platform, and determine movement candidates within the traversable corridor, wherein the movement candidates include center-keeping candidates, left-off candidates, and right-off candidates of the traversable corridor; and a third determining unit, configured to calculate the driving cost of each of the movement candidates respectively, and determine a continuous driving trajectory of the mobile platform based on the driving cost and the traversable corridor.

[0014] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute any of the described driving trajectory generation methods based on traversable domain constraints.

[0015] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any of the described driving trajectory generation methods based on traversable domain constraints.

[0016] By applying the technical solution of this application, and introducing a skeleton path as a geometric connectivity reference instead of direct output, the traditional A... Isograph search algorithms can only address point connectivity and cannot express local obstacle avoidance intentions. Secondly, by constructing passable corridors on the skeleton path and generating laneless semantic behavior candidates such as center-keeping, left offset, and right offset, this approach overcomes the reliance of existing technologies on lane-line semantic maps, improving the adaptability of mobile platforms in unstructured environments lacking clear road markings, such as parks, squares, and tunnels. Finally, by calculating driving costs and determining continuous trajectories, discrete behavior selection is combined with continuous trajectory optimization, ensuring that the final output trajectory is not only geometrically feasible but also semantically consistent with the local decision-making logic of safety and efficiency. This significantly enhances the robustness and interpretability of the navigation system in complex dynamic environments. This solves the problem that existing solutions rely on lane semantics for road behavior decisions, making it difficult to meet the needs of driving trajectory generation in laneless point cloud environments. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a mobile terminal that performs a driving trajectory generation method based on traversable domain constraints, according to an embodiment of this application, is shown.

[0019] Figure 2 A flowchart illustrating a driving trajectory generation method based on traversable domain constraints according to an embodiment of this application is shown.

[0020] Figure 3 A flowchart illustrating a specific method for generating a driving trajectory based on traversable domain constraints according to an embodiment of this application is shown.

[0021] Figure 4 A structural block diagram of a driving trajectory generation device based on traversable domain constraints provided according to an embodiment of this application is shown. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0026] High-dimensional traversable domain: A traversable space generated from point clouds, raster maps, layered tomographic maps, or other environmental representations, containing at least one or more of the following information: location, ground elevation, clearance, layer index, obstacle distance, or continuous traversal cost.

[0027] Skeleton path: A path on the traversable domain. Geometric connected paths obtained by Dijkstra's algorithm, hierarchical graph search, or other graph search methods.

[0028] Passable corridor: A local constraint region obtained by extending along the skeleton path in the left and right normal directions or in the hierarchical neighborhood. It usually includes a center reference line, left boundary, right boundary, minimum width and average width.

[0029] Laneless semantic behavior candidates: In scenarios without lane lines, lane center lines, or Lanelet-type semantic maps, local behavior candidates such as centering, left offset, right offset, and yielding are synthesized from passable corridors.

[0030] Candidate Behavior Generation and Scoring Mechanism: This application refers to a decision-making mechanism that generates multiple candidate behaviors based on passable corridors and scores them according to their feasibility, safety, traffic efficiency, trajectory smoothness, and constraint satisfaction, thereby selecting the target behavior. This mechanism draws on the multi-candidate behavior generation and evaluation ideas of the Efficient Uncertainty-aware Decision-making (EUDM) method, but is not limited to the complete EUDM framework in road traffic scenarios, nor does it require the inclusion of traffic participant interaction modeling, intent branching, or uncertainty inference.

[0031] AL-iLQR trajectory optimization method: AL-iLQR is short for Augmented Lagrangian iterative linear quadratic regulator. It is used to iteratively optimize trajectory state variables and control variables under the constraints of vehicle nonlinear kinematics, traversable corridor boundary constraints, obstacle avoidance constraints and trajectory smoothness requirements, taking candidate behavior or reference path as input, to obtain the target trajectory that meets the constraints.

[0032] Kinematic trajectory: A continuous or discrete trajectory containing state variables such as position, heading, velocity, acceleration, rotation angle or curvature, which can be tracked and executed by the control module of a ground mobile robot.

[0033] As described in the background section, existing solutions rely on lane semantics for road behavior decision-making, which makes it difficult to meet the needs of driving trajectory generation in laneless point cloud environments. To address the problem that existing solutions rely on lane semantics for road behavior decision-making and thus cannot meet the needs of driving trajectory generation in laneless point cloud environments, embodiments of this application provide a driving trajectory generation method, apparatus, computer-readable storage medium, and electronic device based on traversable domain constraints.

[0034] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0035] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal based on a method for generating driving trajectories based on traversable domain constraints, according to an embodiment of the present invention. Figure 1As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0036] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the driving trajectory generation method based on traversable domain constraints in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0037] This embodiment provides a method for generating a driving trajectory based on traversable domain constraints, which runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] Figure 2This is a flowchart of a driving trajectory generation method based on traversable domain constraints according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0039] Step S201: Obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform. Based on the accessible domain map, the starting point and the ending point, use a path search algorithm to determine the skeleton path of the mobile platform.

[0040] Specifically, mobile platforms include ground mobile robots, unmanned inspection vehicles, and low-speed unmanned driving platforms.

[0041] Step S202: Based on the above parameter information of the mobile platform and the above skeleton path, determine the passable corridor of the mobile platform and determine the mobile candidates within the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor.

[0042] Step S203: Calculate the travel cost of each of the above-mentioned mobile candidates, and determine the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor.

[0043] In this embodiment, by applying steps S201, S202, and S203 above, and introducing a skeleton path as a geometric connectivity reference instead of direct output, the problem of traditional A is solved. Isograph search algorithms can only address point connectivity and cannot express local obstacle avoidance intentions. Secondly, by constructing passable corridors on the skeleton path and generating laneless semantic behavior candidates such as center-keeping, left offset, and right offset, this approach overcomes the reliance of existing technologies on lane-line semantic maps, improving the adaptability of mobile platforms in unstructured environments lacking clear road markings, such as parks, squares, and tunnels. Finally, by calculating driving costs and determining continuous trajectories, discrete behavior selection is combined with continuous trajectory optimization, ensuring that the final output trajectory is not only geometrically feasible but also semantically consistent with the local decision-making logic of safety and efficiency. This significantly enhances the robustness and interpretability of the navigation system in complex dynamic environments. This solves the problem that existing solutions rely on lane semantics for road behavior decisions, making it difficult to meet the needs of driving trajectory generation in laneless point cloud environments.

[0044] In the specific implementation process, based on the aforementioned parameter information and skeleton path of the mobile platform, the passable corridor of the mobile platform is determined, including: determining the horizontal normal vector at each path point in the skeleton path, scanning the passable domain to the left and right respectively in the direction of the horizontal normal vector to obtain the left and right passable distances of the path points; and determining the passable corridor of the mobile platform based on the left and right passable distances and the aforementioned parameter information, wherein the aforementioned parameter information includes the width and wheelbase of the mobile platform.

[0045] In this embodiment, a specific generation mechanism for passable corridors is defined, achieving high-precision, dynamically adaptive local constraint modeling. By calculating the horizontal normal vector at each point along the skeleton path and performing bidirectional scanning, the actual passable space relative to the robot body at that location can be accurately obtained, rather than relying on a preset fixed width. This method based on real-time passable domain query can adaptively reflect changes in road width, obstacle intrusion, and terrain undulations. Combined with the width parameter of the mobile platform, the system can dynamically calculate the boundary of the available corridor that meets the vehicle size safety margin. This not only avoids the risk of trajectory boundary collisions caused by overestimation of the corridor but also prevents path deadlock caused by underestimation. In addition, this method is compatible with layered tomographic maps or multi-layered point cloud structures, enabling corridor generation to retain high-dimensional information such as height and layer index, providing a precise geometric constraint basis for subsequent processing of smooth transitions in complex three-dimensional spaces such as spiral ramps and multi-layered buildings.

[0046] Specifically, the driving cost of each of the aforementioned moving candidates is calculated, including: determining the safety cost, efficiency cost, deviation cost, and comfort cost of each of the aforementioned moving candidates; and performing a weighted summation calculation on the aforementioned safety cost, efficiency cost, deviation cost, and comfort cost to obtain the aforementioned driving cost of each of the aforementioned moving candidates.

[0047] In this embodiment, a multi-dimensional weighted candidate scoring mechanism is established, achieving a balance between safety, efficiency, and comfort. Traditional navigation methods often use only path length or distance to a single obstacle as evaluation criteria, which can easily lead to the robot being overly aggressive in narrow passages or wandering in open areas. This embodiment constructs a cost function that includes four dimensions: safety, efficiency, deviation, and comfort, making behavior selection more comprehensive. The safety cost ensures that the trajectory stays away from obstacles and maintains the minimum safe distance; the efficiency cost encourages choosing shorter paths to improve passage efficiency; the deviation cost tends to maintain center-of-motion to reduce lateral jitter; and the comfort cost ensures a comfortable riding experience by penalizing frequent behavior switching (such as frequent left and right deviations). This multi-objective weighted summation approach allows the system to automatically weigh various indicators in different scenarios, thereby selecting the optimal local behavior, avoiding the side effects of optimizing a single indicator, and significantly improving the intelligence level of navigation decision-making.

[0048] More specifically, the safety cost, efficiency cost, bias cost, and comfort cost of each of the aforementioned moving candidates are determined, including according to the first formula: Determine the security costs of the aforementioned moving candidates, wherein, For the aforementioned security costs, Indicates target movement candidate The number of collisions with obstacles Indicates the desired safe width. , The penalty coefficient is... This represents the positive part of the function that takes the value of zero when it is less than zero; according to the second formula: Determine the efficiency cost of the aforementioned moving candidates, wherein, For the aforementioned efficiency cost, Indicates target movement candidate B is the length of the centerline; B is the set of candidate moves within this decision-making cycle. This represents any one of the move candidates in the move candidate set. Indicates a moving candidate The corresponding centerline length; according to the third formula: Determine the aforementioned bias cost of the aforementioned moving candidate, wherein, The cost of the aforementioned deviation, Indicates target movement candidate The number of path points, The center remains the candidate in the first place. The corresponding centerline point of each path point Describing the Euclidean norm, Indicates target movement candidate In the The centerline point at each path point; according to the fourth formula: Determine the aforementioned comfort costs of the aforementioned mobility candidates, wherein, For the aforementioned cost of comfort, This indicates the comfort or behavior switching penalty corresponding to the movement candidate type. This indicates that the aforementioned centers remain candidates. This indicates the above left-off offset candidates. This indicates the above right-offset candidates. This indicates that the candidate should be suspended.

[0049] In this embodiment, the cost calculation model for each dimension is specified, providing a quantitative and physically meaningful basis for decision-making. The safety cost, by penalizing obstacle conflicts and narrow passages, intuitively reflects collision risk. The introduction of the positive part function ensures that penalties are only incurred when constraints are violated, which is logical and computationally efficient. The efficiency cost, by comparing the path lengths of different candidates, quantifies the time cost of detours, guiding the system to prioritize direct paths. The deviation cost, by calculating the average lateral distance between the candidate centerline and the centerline, quantifies the degree of deviation from the centerline, suppressing unnecessary lateral swaying and ensuring the stability of straight-line travel. The comfort cost, by pre-setting penalty values ​​for different behavior types, clarifies the setting logic of centering being superior to deviation and moving being superior to stopping. These specific mathematical formulas not only make the scoring mechanism mathematically rigorous and easy to solve in optimization algorithms, but also make the behavior selection process transparent and interpretable.

[0050] Further, determining the continuous driving trajectory of the mobile platform based on the aforementioned driving cost and the aforementioned passable corridor includes: determining the aforementioned candidate with the minimum driving cost as the target candidate, and determining the target trajectory of the mobile platform based on the aforementioned target candidate and the aforementioned passable corridor; obtaining the centerline, left and right boundary information, and obstacle information corresponding to the aforementioned target trajectory, and optimizing the aforementioned target trajectory based on the aforementioned centerline, the aforementioned left and right boundary information, and the aforementioned obstacle information to obtain the aforementioned continuous driving trajectory.

[0051] This embodiment describes the seamless integration of behavior selection and trajectory generation, achieving a seamless connection between discrete decision-making and continuous optimization. First, by selecting the candidate target movement with the lowest overall cost, the dimensionality of subsequent trajectory optimization is reduced. Second, the centerline, left and right boundaries, and obstacle information of the selected candidate are used as constraint inputs to the trajectory optimization module, ensuring that the optimization process always takes place within a compliant physical space. It preserves the interpretability of high-level behavior decisions (e.g., knowing explicitly that the robot has chosen a right offset) while utilizing the ability of low-level trajectory optimization algorithms (such as AL-iLQR) to handle nonlinear kinematic constraints. This method effectively avoids the potential for boundary violations or unevenness that may result from directly smoothing the skeleton path, ensuring that the final trajectory strictly conforms to the geometric boundaries of the passable corridor while satisfying vehicle dynamics constraints, thus improving the feasibility and execution accuracy of the trajectory.

[0052] Furthermore, the target trajectory is optimized based on the centerline, the left and right boundary information, and the obstacle information to obtain the continuous driving trajectory, including: using the AL-iLQR trajectory optimization algorithm to optimize the target trajectory based on the centerline, the left and right boundary information, and the obstacle information to obtain the continuous driving trajectory.

[0053] In this embodiment, the AL-iLQR algorithm is used for trajectory optimization, improving the real-time performance and kinematic feasibility of trajectory generation. The AL-iLQR (Augmented Lagrange Iterative Linear Quadratic Adjustment) method exhibits excellent convergence speed and solution accuracy in handling nonlinear system models and inequality constraints. By applying it to trajectory optimization under corridor constraints, the system can iteratively solve for the optimal control sequence that satisfies strict kinematic constraints such as vehicle acceleration and steering angle online. Compared to traditional QP or simple smoothing algorithms, AL-iLQR can better handle constraint violations by dynamically adjusting the penalty intensity through the augmented Lagrange term, ensuring that the trajectory is both smooth and adheres closely to the boundaries. This optimization method is particularly suitable for mobile platforms with high-speed or high-precision requirements, maximizing traffic efficiency while effectively suppressing trajectory oscillations and achieving smooth and stable motion control, significantly outperforming traditional simple smoothing methods based on spline interpolation.

[0054] Specifically, after determining the continuous driving trajectory of the mobile platform based on the aforementioned driving cost and the aforementioned passable corridor, the method further includes: projecting the continuous driving trajectory onto the aforementioned passable corridor to perform corridor projection detection on the continuous driving trajectory, and simultaneously performing heading continuity detection on the continuous driving trajectory to obtain a detection result; if the detection result indicates that the test is passed, controlling the mobile platform to move according to the aforementioned continuous driving trajectory.

[0055] In this embodiment, trajectory post-processing and closed-loop control are added to construct a complete safety verification and execution link, ensuring zero-error transmission from planning to execution. Corridor projection detection and heading continuity checks are performed before trajectory output to promptly detect and correct minor boundary violations or heading abrupt changes that may occur during optimization, preventing collisions or control instability caused by numerical errors. Corridor projection detection ensures that trajectory points are strictly within the passable domain, serving as the final fallback verification of corridor constraints; heading continuity checks guarantee a smooth transition in vehicle steering, avoiding large actuator jumps. Platform movement is only controlled if the detection results are satisfactory. This mechanism not only improves the system's adaptability to dynamic obstacles but also enhances the robustness of the entire navigation stack, enabling the robot to operate reliably and safely in complex and changing environments, meeting the high reliability requirements of industrial applications.

[0056] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the driving trajectory generation method based on traversable domain constraints will be described in detail below with reference to specific embodiments.

[0057] This embodiment relates to a specific method for generating driving trajectories based on traversable domain constraints, addressing how to generate trajectories A under high-dimensional traversable domain constraints. The geometric skeleton path obtained from other graph searches is improved into a candidate set of laneless semantic behaviors, and the selected behavior is transformed into a robot-executable trajectory that satisfies the traversable domain, corridor boundaries, and kinematic constraints. Figure 3 As shown, this embodiment of the application takes a high-dimensional traversable domain, a starting point, an ending point, robot dimensions, and obstacle information as input. First, in the high-dimensional traversable domain, A... The system first obtains a skeleton path using Dijkstra's algorithm, hierarchical graph search, or other search methods; this skeleton path serves only as a geometric connectivity reference. Next, the tangential and normal directions are estimated along the skeleton path, and a traversable corridor is generated by combining the traversable domain query results. Then, three movement candidates—center-maintaining, left-offsetting, and right-offsetting—are synthesized within the corridor; if all movement candidates are blocked by obstacles, stopping candidates are added based on the nearest obstacle location and a preset stopping distance. The system then selects the optimal behavior based on safety, efficiency, center deviation, and comfort costs. Finally, the corridor centerline, left and right boundaries, obstacles, and robot state corresponding to the selected behavior are input into the AL-iLQR or constrained smoothing trajectory module to obtain a trajectory that satisfies the traversable domain, corridor boundaries, and kinematic constraints.

[0058] Therefore, this embodiment is similar to A. The relationship for other graph search is: A This embodiment only addresses geometric connectivity; in A... Alternatively, a behavior candidate layer and a kinematic trajectory optimization layer can be added on top of the graph search results. The graph search algorithm itself is not the innovation point limited to this embodiment. The innovation point of this embodiment focuses on "the generation of laneless semantic behavior candidates, candidate scoring selection, and kinematic trajectory optimization within the selected corridor under the constraint of high-dimensional traversable domain".

[0059] like Figure 3 As shown, this embodiment is divided into three groups: "input and geometric connectivity", "laneless semantic behavior candidates", and "corridor-constrained trajectory optimization and output". The first group receives the high-dimensional traversable domain and obtains the skeleton path and corridor; the second group generates and selects semantic behavior candidates within the corridor; and the third group transforms the selected behavior into a kinematic trajectory that satisfies the constraints.

[0060] 1) Input and Geometric Connectivity Group: This includes input data, skeleton path search, and traversable corridor generation modules. The input data module receives traversable domains, start and end points, robot dimensions, and obstacles; the skeleton path search module is only responsible for obtaining geometrically connected paths; the traversable corridor generation module scans the left and right traversable boundaries along the skeleton path to form local corridors with width attributes.

[0061] 2) Laneless Semantic Behavior Candidate Group: This includes candidate synthesis, candidate verification, and behavior selection modules. The candidate synthesis module generates candidates for center-keeping, left-shifting, right-shifting, and yielding / stopping when necessary; the candidate verification module checks whether candidates cross boundaries, are too narrow, or conflict with obstacles; the behavior selection module selects candidates based on safety, efficiency, center deviation, and comfort costs.

[0062] 3) Corridor-constrained trajectory optimization and output group: including AL-iLQR or constrained smoothing, trajectory post-processing, and output modules. The optimization module generates a reference path and trajectory within the selected corridor; the post-processing module projects or clamps the results into the corridor and checks heading continuity; the output module provides trajectory status such as position, altitude, heading, velocity, acceleration, and turning angle.

[0063] The specific method for generating driving trajectories based on traversable domain constraints in this embodiment includes the following steps:

[0064] S1: Read the high-dimensional traversable region, start point, end point, robot width, wheelbase, speed limit, and obstacle information. The high-dimensional traversable region can be derived from point cloud traversable regions, hierarchical tomographic maps, 2D cost maps, topology maps, or other environmental representations.

[0065] S2: Execute A on a high-dimensional walkable domain The algorithm uses Dijkstra's algorithm, hierarchical graph search, or other search methods to obtain the skeleton path connecting the start and end points. This step only guarantees geometric connectivity and does not directly output the final behavior or trajectory.

[0066] S3: Resample and smooth the skeleton path, calculate the tangent and normal at each path point, and query the passable domain along the normal to obtain the left and right boundaries and available width, forming a passable corridor.

[0067] S4: Generate center-preserving candidates, left-offset candidates, and right-offset candidates within the corridor. The left and right offset candidates use a head-and-tail offset transition method to ensure that the starting and ending points return to the vicinity of the original center reference.

[0068] S5: Perform passable domain boundary verification, minimum width verification, and obstacle conflict verification on each moving candidate, and eliminate unusable candidates or impose safety penalties on candidates.

[0069] S6: If all moving candidates are blocked by obstacles, then based on the nearest position of the obstacle on the center line of the corridor and the preset parking distance, cut off the center candidate and add a stop candidate.

[0070] S7: Calculate the safety cost, efficiency cost, center deviation cost, and comfort cost for each candidate, and select the candidate with the minimum overall cost.

[0071] S8: Input the centerline, left and right boundaries, obstacles and the current state of the robot corresponding to the selected candidate into the AL-iLQR or constrained smooth trajectory module.

[0072] S9: Apply constraints such as corridor boundaries, vehicle size, speed, acceleration, turning angle, or curvature during the optimization process; if the external AL-iLQR is not available, use a constrained smooth trajectory as a deterministic alternative.

[0073] S10: Perform corridor projection, boundary clamping, and heading continuity checks on the optimization results, and output a trajectory that satisfies the traversable domain, corridor boundary, and kinematic constraints.

[0074] In this embodiment, the graph search result of S2 is the geometric input; S4 to S7 constitute the laneless semantic behavior candidate layer of this embodiment; and S8 to S10 constitute the corridor constraint kinematic trajectory optimization layer of this embodiment.

[0075] 1. The specific schemes for generating the skeleton path and passable corridor in S2 and S3 of this embodiment are as follows:

[0076] Let the high-dimensional passable region be... It can provide query results such as whether the location is passable, local cost, ground elevation, layer index, or clearance. The start and end points are respectively... and .exist The skeleton path is obtained by searching the execution graph above: In the formula, Indicates the skeleton path; Indicates the first 1 path point; This represents the number of path points. Graph search can be A... Dijkstra, Layer A Or other equivalent search methods.

[0077] After resampling the skeleton path, calculate the first... Unit tangent vector at each path point and horizontal normal vector Scanning the passable region left and right along the normal direction yields the passable distances to the left and right:

[0078] ;

[0079] ;

[0080] In the formula, and These represent the passable distances on the left and right sides, respectively. Indicates the scanning distance; Indicates the maximum horizontal scan distance; This represents the intermediate distance during the scanning process. If a layered tomographic map is used, the query can also include the layer index, ground elevation, and clearance conditions.

[0081] Considering robot width and safety margin , No. The available corridor width at each path point is:

[0082] In the formula, Indicates the available corridor width. If If the distance is less than the robot's minimum passage width, the candidate at that location can be eliminated or given a higher safety penalty.

[0083] 2. The specific scheme for generating laneless semantic behavior candidates in S4 of this embodiment is as follows:

[0084] This embodiment does not rely on road lane semantics, but synthesizes semantic behavior candidates within a passable corridor. Let the candidate set be: In the formula, This indicates that the center remains a candidate; Indicates a left offset candidate; Indicates a candidate for right offset; This represents the set of candidate moves.

[0085] For each candidate Let the lateral target offset be... The center remains a candidate Left offset candidate Right-off candidate To avoid abrupt changes at the start and end points, the cumulative distance along the path is defined. and offset transition distance The offset coefficient is:

[0086] In the formula, Indicates the first The offset ratio of each path point; Indicates the total length of the skeleton path. Candidate The center line point can be written as:

[0087] In the formula, Indicates candidate In the The centerline point at each path point; This indicates the lateral offset of the candidate. After generation, the system... The passable region is queried point by point, and the minimum width of the candidate is calculated by combining the left and right boundaries. If the candidate goes beyond the corridor, or there are continuous impassable sections, or If the value is less than the threshold, the candidate can be eliminated.

[0088] When all moving candidates are blocked by obstacles, this embodiment adds a stopping candidate. Let the set of obstacles be... If for any If both the candidate and the obstacle conflict, then the index of the central candidate that is closest to the obstacle is taken. And based on parking distance Select truncated index:

[0089] In the formula, This indicates the index of the last path point that allowed the candidate to stop; Indicates the first The path length calculated along the center candidate direction between each path point and the nearest path point to the obstacle.

[0090] 3. The specific scheme for candidate scoring and behavior selection in S7 of this embodiment is as follows:

[0091] For candidate set Each candidate in This invention is scored based on a comprehensive consideration of safety, efficiency, center deviation, and comfort costs. ;

[0092] In the formula, Indicates candidate The overall cost; This indicates the cost of security; Indicates the cost of efficiency; This represents the cost of deviation relative to the central reference. This indicates a trade-off in comfort; , , , These represent the corresponding weights.

[0093] The safety costs can be comprised of both the number of obstacle conflicts and insufficient corridor width:

[0094] In the formula, Indicates candidate The number of collisions with obstacles; Indicates the desired safety width; and This is the penalty coefficient; This represents the positive part of the function that takes the value of zero when the value is less than zero.

[0095] The efficiency cost can be represented by the increase in candidate length relative to the shortest candidate length:

[0096] In the formula, Indicates candidate B is the length of the centerline; B is the set of candidate moves within this decision-making cycle. It represents any one of the move candidates in the move candidate set.

[0097] The cost of center deviation can be represented by the average lateral distance between the candidate centerline and the center that keeps the candidate in place:

[0098] In the formula, Indicates candidate The number of path points; This indicates the corresponding centerline point that remains the center of the candidate; This represents the Euclidean norm. This term ensures that the system preferentially maintains a smaller offset when safety conditions are similar.

[0099] Comfort costs can be set according to behavior type; for example, keeping the center has the lowest cost, shifting left or right has a moderate cost, and stopping has a higher cost.

[0100] In the formula, This represents the comfort level or behavior switching penalty corresponding to the candidate behavior type. The candidate with the lowest overall cost is ultimately selected.

[0101] In the formula, This indicates the final selected behavior candidate.

[0102] 4. The specific schemes for optimizing the kinematic trajectory under corridor constraints in S8 and S9 of this embodiment are as follows:

[0103] Select behavior Subsequently, in this embodiment, the centerline, left and right boundaries, and obstacle information are input into the trajectory optimization module. Let the discrete trajectory state be:

[0104] In the formula, Indicates the first The state of each trajectory node; Indicates planar position; Indicates the heading angle; Indicates speed; This represents the control quantity of the k-th trajectory node; Indicates acceleration; Indicates the steering angle.

[0105] Taking the bicycle model as an example, the kinematic constraints can be written as:

[0106] ;

[0107] In the formula, Represents a discrete kinematic model; Indicates the time step; Indicates wheelbase.

[0108] Corridor boundary constraints are implemented by projecting trajectory points onto the local coordinates of a selected corridor. Let the first... The projection center of each trajectory point on the corridor is: , normal direction is The left and right horizontal boundaries are respectively and The vehicle's lateral safety margin is Then the trajectory points must satisfy:

[0109] In the formula, the left-hand inequality ensures that the trajectory does not exceed the right boundary, and the right-hand inequality ensures that the trajectory does not exceed the left boundary. For multi-level passable regions, the layer index, ground height, or clearance of the trajectory points can also be constrained simultaneously.

[0110] The trajectory optimization objective can be written as:

[0111] ;

[0112] In the formula, Indicate the trajectory optimization objective; Indicates the total number of trajectory nodes; Indicates the position of the trajectory point; Indicates the selected candidate reference centerline point; Indicates the positional deviation weight. express Weighted norm; Indicates the desired speed; , , , These represent the penalty weights for speed deviation, acceleration, steering angle, and control changes, respectively. This is the control variable for the (k-1)th trajectory node.

[0113] When using AL-iLQR, constraints such as corridor boundaries, velocity, acceleration, turning angle, and obstacle distance can be denoted as... And construct enhanced Lagrange targets:

[0114] In the formula, This indicates an enhancement of the Lagrange objective; Representing constraints The multiplier; Indicates the penalty parameter; This indicates the result of the constraint violation calculated using the positive part function. If the AL-iLQR backend is unavailable, the constrained smooth trajectory module can also be used: first, resample the selected candidate centerlines, perform quick connection, spline or elastic smoothing, then project or clamp all points into the corridor boundary, and refresh the heading and speed.

[0115] 5. The specific implementation scheme of the output data structure and downstream interface in S10 of this embodiment is as follows:

[0116] This embodiment outputs two types of results. The first type is the behavior selection result, which includes at least the candidate name, behavior type, centerline, left and right boundaries, minimum width, average width, path length, and overall score. The second type is the optimized trajectory, which includes at least the position, altitude, heading, speed, acceleration, and turning angle or curvature. The system can provide a sequence of trajectory points to the downstream controller, and can also provide all candidates and selected corridor boundaries to the visualization module.

[0117] The interface of this embodiment does not require the input traversable region to be generated by a specific algorithm. As long as the input environment representation can query location traversability, ground elevation, layer index, clearance, or cost, this embodiment can generate candidate and optimized trajectories based on it. Therefore, this embodiment can be combined with 2.5D traversable grids, multi-layer tomographic maps, voxel traversable maps, weighted topology maps, occupied grids, or other traversable region representations.

[0118] Compared to existing solutions, the embodiments of this application specifically achieve the following technical effects:

[0119] 1) This embodiment does not include A Instead of directly using other graph search paths as the final trajectory, it adds a layer of behavior candidates and behavior selection on top of geometrically connected paths, enabling interpretable decisions between center-keeping, left and right offsets, and stopping.

[0120] 2) This embodiment does not rely on lane lines, lanelets, or road topology, but generates candidates from high-dimensional traversable domains and corridor boundaries, making it suitable for laneless scenarios such as parks, squares, ramps, and multi-story buildings.

[0121] 3) In this embodiment, candidate selection is integrated with trajectory optimization. The selected behavior enters the AL-iLQR or constrained smoothing module, and finally outputs an executable trajectory that includes states such as position, heading, speed, acceleration and turning angle.

[0122] 4) This embodiment explicitly uses corridor boundaries, vehicle dimensions and kinematic constraints in trajectory optimization, and can retain high-dimensional information such as layer index, ground height and clearance, taking into account the connectivity of multi-layer structures and the smoothness of flat straight roads.

[0123] In addition, this embodiment also includes the following alternatives:

[0124] 1) The input can be replaced with a point cloud traversable region, a 2D occupancy grid, a layered cost map, a voxel map, a topology map, an ESDF distance field, or a multi-sensor fusion map; the skeleton path can be derived from A Dijkstra, Theta Hybrid A Generate RRT, PRM, or hierarchical graph search results.

[0125] 2) Corridors can be obtained through normal scanning, distance field contour lines, morphological processing, convex decomposition, sampling boundaries, passable patches, or topological neighborhood expansion; behavioral candidates can be expanded to include obstacle avoidance, deceleration following, narrow passage yielding, layer switching deceleration, local U-turn, or waiting for recovery.

[0126] 3) Left and right offsets can be generated using linear transition, polynomial, spline, cosine function, or lookup table methods; candidate scores can be generated using a combination of linear weighting, piecewise functions, logical functions, learning scorers, rule priority, or multi-objective ranking methods.

[0127] 4) Obstacle conflicts can be judged by circular envelope, rectangular envelope, polygonal envelope, distance field threshold, spatiotemporal occupancy, predicted trajectory, or risk probability; trajectory optimization can be replaced by AL-iLQR, CILQR, MPC, SQP, QP, spline smoothing, elastic band, CHOMP, or STOMP.

[0128] 5) The kinematic model can be a differential, Ackerman, bicycle, tracked, omnidirectional chassis or ground robot model with height variation; the output can be discrete trajectory points, time parameterized trajectory, ROS path, control command sequence or interface for downstream controller to query.

[0129] This application also provides a driving trajectory generation device based on traversable domain constraints. It should be noted that the driving trajectory generation device based on traversable domain constraints in this application can be used to execute the driving trajectory generation method based on traversable domain constraints provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0130] The following describes the driving trajectory generation device based on traversable domain constraints provided in the embodiments of this application.

[0131] Figure 4 This is a schematic diagram of a driving trajectory generation device based on traversable domain constraints according to an embodiment of this application. Figure 4As shown, the device includes:

[0132] The first determining unit 41 is used to obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform, and to determine the skeleton path of the mobile platform using a path search algorithm based on the accessible domain map, the starting point and the ending point.

[0133] The second determining unit 42 is used to determine the passable corridor of the mobile platform based on the parameter information of the mobile platform and the skeleton path, and to determine the mobile candidates in the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate and the right-off candidate of the passable corridor.

[0134] The third determining unit 43 is used to calculate the driving cost of each of the above-mentioned moving candidates, and determine the continuous driving trajectory of the above-mentioned moving platform based on the driving cost and the above-mentioned passable corridor.

[0135] In this embodiment, the first determining unit is used to acquire the traversable domain map, parameter information, start point, and end point of the mobile platform, and determine the skeleton path of the mobile platform using a path search algorithm based on the traversable domain map, start point, and end point. The second determining unit is used to determine the traversable corridor of the mobile platform based on the parameter information and skeleton path, and determine the movement candidates within the traversable corridor, wherein the movement candidates include center-keeping candidates, left-off candidates, and right-off candidates of the traversable corridor. The third determining unit is used to calculate the driving cost of each movement candidate, and determine the continuous driving trajectory of the mobile platform based on the driving cost and the traversable corridor. By introducing the skeleton path as a geometric connectivity reference, instead of direct output, the traditional A Isograph search algorithms can only address point connectivity and cannot express local obstacle avoidance intentions. Secondly, by constructing passable corridors on the skeleton path and generating laneless semantic behavior candidates such as center-keeping, left offset, and right offset, this approach overcomes the reliance of existing technologies on lane-line semantic maps, improving the adaptability of mobile platforms in unstructured environments lacking clear road markings, such as parks, squares, and tunnels. Finally, by calculating driving costs and determining continuous trajectories, discrete behavior selection is combined with continuous trajectory optimization, ensuring that the final output trajectory is not only geometrically feasible but also semantically consistent with the local decision-making logic of safety and efficiency. This significantly enhances the robustness and interpretability of the navigation system in complex dynamic environments. This solves the problem that existing solutions rely on lane semantics for road behavior decisions, making it difficult to meet the needs of driving trajectory generation in laneless point cloud environments.

[0136] As an optional solution, the second determining unit includes a first determining module and a second determining module; the first determining module is used to determine the horizontal normal vector at each path point in the skeleton path, and scan the passable domain to the left and right respectively in the direction of the horizontal normal vector to obtain the left and right passable distances of the path points; the second determining module is used to determine the passable corridor of the mobile platform according to the left and right passable distances and the parameter information, wherein the parameter information includes the width and wheelbase of the mobile platform.

[0137] In one optional scheme, the third determining unit includes a third determining module and a calculation processing module; the third determining module is used to determine the safety cost, efficiency cost, deviation cost and comfort cost of each of the above-mentioned moving candidates; the calculation processing module is used to perform weighted summation calculation on the above-mentioned safety cost, efficiency cost, deviation cost and comfort cost to obtain the above-mentioned driving cost of each of the above-mentioned moving candidates.

[0138] In one optional scheme, the third determining module includes a first determining submodule, a second determining submodule, a third determining submodule, and a fourth determining submodule; the first determining submodule is used to determine the formula according to the first formula: Determine the security costs of the aforementioned moving candidates, wherein, For the aforementioned security costs, Indicates a moving candidate The number of collisions with obstacles Indicates the desired safe width. , The penalty coefficient is... The function represents the positive part that takes the value of zero when it is less than zero; the second determining submodule is used to determine the positive part based on the second formula: Determine the efficiency cost of the aforementioned moving candidates, wherein, For the aforementioned efficiency cost, Indicates target movement candidate Let B be the length of the centerline, and B be the set of movement candidates within this action decision cycle. This represents any one of the movement candidates in the aforementioned movement candidate set. Indicates a moving candidate The centerline length; the third determining submodule is used according to the third formula: Determine the aforementioned bias cost of the aforementioned moving candidate, wherein, The cost of the aforementioned deviation, Indicates a moving candidate The number of path points, The center remains the candidate in the first place. The corresponding centerline point of each path point Describing the Euclidean norm, Indicates a moving candidate In the The centerline point at each path point; the fourth determination submodule is used to determine the fourth formula: Determine the aforementioned comfort costs of the aforementioned mobility candidates, wherein, For the aforementioned cost of comfort, This indicates the comfort or behavior switching penalty corresponding to the movement candidate type. This indicates that the aforementioned centers remain candidates. This indicates the above left-off offset candidates. This indicates the above right-offset candidates. This indicates that the candidate should be suspended.

[0139] In one optional scheme, the third determining unit further includes a fourth determining module and an optimization processing module; the fourth determining module is used to determine the moving candidate with the lowest driving cost as the target moving candidate, and to determine the target trajectory of the moving platform based on the target moving candidate and the passable corridor; the optimization processing module is used to obtain the centerline, left and right boundary information and obstacle information corresponding to the target trajectory, and to optimize the target trajectory based on the centerline, left and right boundary information and obstacle information to obtain the continuous driving trajectory.

[0140] In one optional scheme, the optimization processing module includes an optimization processing submodule, which is used to optimize the target trajectory using the AL-iLQR trajectory optimization algorithm based on the centerline, the left and right boundary information and the obstacle information, to obtain the continuous driving trajectory.

[0141] In one alternative embodiment, the apparatus further includes a projection unit and a control unit; the projection unit is used to project the continuous driving trajectory of the mobile platform onto the accessible corridor after determining the continuous driving trajectory based on the driving cost and the accessible corridor, so as to perform corridor projection detection on the continuous driving trajectory and simultaneously perform heading continuity detection on the continuous driving trajectory to obtain a detection result; the control unit is used to control the mobile platform to move according to the continuous driving trajectory when the detection result indicates that the test is passed.

[0142] The aforementioned trajectory generation device based on traversable domain constraints includes a processor and a memory. The first determining unit, the second determining unit, the third determining unit, etc., are all stored as program units in the memory, and the processor executes the program units stored in the memory to achieve the corresponding functions. All of the above modules are located in the same processor; alternatively, the modules may be located in different processors in any combination.

[0143] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured. By adjusting kernel parameters, the problem that existing solutions rely on lane semantics for road behavior decisions and struggle to meet the requirements for generating driving trajectories in laneless point cloud environments can be addressed.

[0144] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0145] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to execute the driving trajectory generation method based on traversable domain constraints.

[0146] Specifically, the driving trajectory generation method based on traversable domain constraints includes:

[0147] Step S201: Obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform. Based on the accessible domain map, the starting point and the ending point, use a path search algorithm to determine the skeleton path of the mobile platform.

[0148] Step S202: Based on the above parameter information of the mobile platform and the above skeleton path, determine the passable corridor of the mobile platform and determine the mobile candidates within the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor.

[0149] Step S203: Calculate the travel cost of each of the above-mentioned mobile candidates, and determine the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor.

[0150] This invention provides a processor for running a program, wherein the program executes the driving trajectory generation method based on traversable domain constraints.

[0151] Specifically, the driving trajectory generation method based on traversable domain constraints includes:

[0152] Step S201: Obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform. Based on the accessible domain map, the starting point and the ending point, use a path search algorithm to determine the skeleton path of the mobile platform.

[0153] Step S202: Based on the above parameter information of the mobile platform and the above skeleton path, determine the passable corridor of the mobile platform and determine the mobile candidates within the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor.

[0154] Step S203: Calculate the travel cost of each of the above-mentioned mobile candidates, and determine the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor.

[0155] This invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0156] Step S201: Obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform. Based on the accessible domain map, the starting point and the ending point, use a path search algorithm to determine the skeleton path of the mobile platform.

[0157] Step S202: Based on the above parameter information of the mobile platform and the above skeleton path, determine the passable corridor of the mobile platform and determine the mobile candidates within the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor.

[0158] Step S203: Calculate the travel cost of each of the above-mentioned mobile candidates, and determine the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor.

[0159] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0160] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0161] Step S201: Obtain the accessible domain map, parameter information, starting point and ending point of the mobile platform. Based on the accessible domain map, the starting point and the ending point, use a path search algorithm to determine the skeleton path of the mobile platform.

[0162] Step S202: Based on the above parameter information of the mobile platform and the above skeleton path, determine the passable corridor of the mobile platform and determine the mobile candidates within the passable corridor, wherein the mobile candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor.

[0163] Step S203: Calculate the travel cost of each of the above-mentioned mobile candidates, and determine the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor.

[0164] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0165] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0166] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0167] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0168] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0169] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0170] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0171] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0172] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0173] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0174] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for generating driving trajectories based on traversable domain constraints, characterized in that, include: Obtain the traversable domain map, parameter information, start point and end point of the mobile platform, and determine the skeleton path of the mobile platform using a path search algorithm based on the traversable domain map, the start point and the end point; Based on the parameter information of the mobile platform and the skeleton path, the passable corridor of the mobile platform is determined, and the movement candidates within the passable corridor are determined, wherein the movement candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor; The driving cost of each of the mobile candidates is calculated, and the continuous driving trajectory of the mobile platform is determined based on the driving cost and the passable corridor.

2. The method according to claim 1, characterized in that, Based on the parameter information of the mobile platform and the skeleton path, determine the passable corridor of the mobile platform, including: Determine the horizontal normal vector at each path point in the skeleton path, and scan the passable domain to the left and right respectively in the direction of the horizontal normal vector to obtain the left and right passable distances of the path points; Based on the left and right passable distances and the parameter information, the passable corridor of the mobile platform is determined, wherein the parameter information includes the width and wheelbase of the mobile platform.

3. The method according to claim 1, characterized in that, Calculate the travel cost for each of the aforementioned moving candidates, including: Determine the safety cost, efficiency cost, bias cost, and comfort cost of each of the aforementioned mobile candidates; The safety cost, efficiency cost, deviation cost, and comfort cost are weighted and summed to obtain the driving cost for each of the moving candidates.

4. The method according to claim 3, characterized in that, Determine the safety cost, efficiency cost, bias cost, and comfort cost of each of the aforementioned motion candidates, including: According to the first formula: Determine the security cost of the moving candidate, wherein, For the security cost, Indicates target movement candidate The number of collisions with obstacles Indicates the desired safe width. , The penalty coefficient is... This represents the positive part of the function that takes the value of zero when the value is less than zero. According to the second formula: Determine the efficiency cost of the proposed move, wherein, For the efficiency cost mentioned above, Indicates target movement candidate Let B be the length of the centerline, and B be the set of movement candidates within this action decision cycle. This represents any one of the movement candidates in the set of movement candidates. Indicates a moving candidate The length of the centerline; According to the third formula: Determine the deviation cost of the moving candidate, wherein, The cost of the deviation, Indicates target movement candidate The number of path points, The center remains the candidate in the first place. The corresponding centerline point of each path point Denotes the Euclidean norm. Indicates target movement candidate In the The centerline point at each path point; According to the fourth formula: Determine the comfort cost of the mobility candidate, wherein, For the aforementioned comfort, This indicates the comfort or behavior switching penalty corresponding to the movement candidate type. This indicates that the center remains a candidate. Indicates the left offset candidate, This indicates the right offset candidate. This indicates that the candidate should be suspended.

5. The method according to claim 1, characterized in that, Determining the continuous travel trajectory of the mobile platform based on the travel cost and the passable corridor includes: The candidate with the lowest travel cost is identified as the target candidate, and the target trajectory of the mobile platform is determined based on the target candidate and the passable corridor. The centerline, left and right boundary information, and obstacle information corresponding to the target trajectory are obtained, and the target trajectory is optimized based on the centerline, left and right boundary information, and obstacle information to obtain the continuous driving trajectory.

6. The method according to claim 5, characterized in that, The target trajectory is optimized based on the centerline, the left and right boundary information, and the obstacle information to obtain the continuous driving trajectory, including: Based on the centerline, the left and right boundary information, and the obstacle information, the AL-iLQR trajectory optimization algorithm is used to optimize the target trajectory to obtain the continuous driving trajectory.

7. The method according to claim 1, characterized in that, After determining the continuous travel trajectory of the mobile platform based on the travel cost and the accessible corridor, the method further includes: The continuous driving trajectory is projected onto the passable corridor to perform corridor projection detection on the continuous driving trajectory, and at the same time, the heading continuity of the continuous driving trajectory is detected to obtain the detection result; If the detection result indicates that the motion is successful, the mobile platform is controlled to move along the continuous driving trajectory.

8. A driving trajectory generation device based on traversable domain constraints, characterized in that, include: The first determining unit is used to obtain the traversable domain map, parameter information, starting point and ending point of the mobile platform, and determine the skeleton path of the mobile platform using a path search algorithm based on the traversable domain map, the starting point and the ending point. The second determining unit is used to determine the passable corridor of the mobile platform based on the parameter information of the mobile platform and the skeleton path, and to determine the movement candidates within the passable corridor, wherein the movement candidates include the center-keeping candidate, the left-off candidate, and the right-off candidate of the passable corridor; The third determining unit is used to calculate the driving cost of each of the mobile candidates respectively, and determine the continuous driving trajectory of the mobile platform based on the driving cost and the passable corridor.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the driving trajectory generation method based on traversable domain constraints as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing the driving trajectory generation method based on drivable domain constraints as described in any one of claims 1 to 7.

Images