Cruise scene reverse trajectory planning method and system
By generating passable paths, constructing virtual lanes, and adapting them to the planned starting point, the problem of the disconnect between reversing planning and forward planning is solved by using a forward planning scheme, thus achieving high efficiency, safety, and smoothness of reversing planning in complex scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- VOYAH AUTOMOBILE TECH CO LTD
- Filing Date
- 2026-04-13
- Publication Date
- 2026-06-12
Smart Images
Figure CN122186146A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of intelligent driving technology, specifically to a method and system for planning reversing trajectory in a cruise scenario. Background Technology
[0002] With the rapid development of intelligent driving technology, the environmental perception, decision-making, planning, and control execution capabilities of autonomous vehicles have been significantly improved. Trajectory planning, as one of the core modules of an intelligent driving system, is primarily responsible for generating a safe, comfortable, and executable driving trajectory while satisfying vehicle kinematic constraints, dynamic constraints, and traffic rules. In real-world driving scenarios, in addition to conventional forward driving, reversing scenarios are also widespread, such as parking lot navigation, U-turns in dead ends, yielding to oncoming traffic on narrow roads, and parallel parking. These scenarios often require vehicles to possess precise reversing planning capabilities to ensure that the vehicle can safely and efficiently complete specific tasks.
[0003] Currently, trajectory planning systems for intelligent driving vehicles are typically deeply optimized for forward driving, while reversing planning functions are relatively independent and underdeveloped. Existing reversing functions are mainly applied to Automatic Parking Assistance (APA) scenarios, while reversing decision-making and planning technologies for parking lot cruising and public road scenarios are still immature. Most solutions tend to develop forward planning and reversing planning as two separate modules. This not only increases the development and maintenance costs of the system but also prevents mature obstacle avoidance, smoothing, and compliance checks from being reused in reversing scenarios.
[0004] Although some solutions for vehicle trajectory planning and reversing control exist in existing technologies, the following limitations still exist in practical applications: For example, Chinese patent application CN117008615A discloses a method and system for trajectory planning of unmanned vehicles with strategy switching. This scheme attempts to encompass the advantages of structured roads and OpenSpace scenarios, switching strategies in real time through scene recognition and classification. However, this scheme directly uses the OpenSpace algorithm for trajectory planning in reversing planning. Although the OpenSpace algorithm can handle unstructured road scenarios, when faced with complex dynamic scenarios (such as yielding to oncoming traffic in both directions on narrow roads, dynamic obstacle avoidance, etc.), the paths it generates often lack sufficient flexibility and are difficult to effectively cope with complex interactive scenarios, resulting in a reduced planning success rate or an uneven trajectory.
[0005] For example, Chinese patent application CN116022147A discloses a control method and device for reversing a vehicle. This method performs lateral planning based on historical trajectories according to boundaries, and then performs longitudinal planning based on the lateral planning results and the positional relationship of obstacles. A significant drawback of this method is its excessive reliance on historical trajectories. In dynamically changing traffic environments, historical trajectories may no longer be applicable (e.g., new obstacles appear ahead or traffic rules change). If planning is forcibly based on historical trajectories, the vehicle may be unable to respond promptly to unexpected situations, posing a safety hazard. Furthermore, this method is difficult to effectively handle complex dynamic scenarios and has poor generalization ability.
[0006] Existing reversing planning solutions typically require the development of a separate planning module, disconnected from the forward decision-making and planning system. This means that proven capabilities in forward planning, such as spatiotemporal joint planning and constraint optimization, cannot be directly extended to backward planning scenarios. This redundant development not only increases computational consumption and system complexity but also leads to inconsistencies in forward and backward planning behavior, impacting passenger comfort. Furthermore, the lack of a unified planning framework makes it difficult to flexibly expand reversing functionality without affecting the original forward decision-making and planning scheme.
[0007] In summary, there is a lack of a reversing planning solution in the existing technology that can be compatible with forward planning capabilities, adapt to complex dynamic scenarios, and does not require the development of a separate independent module. Summary of the Invention
[0008] This application provides a method and system for reversing trajectory planning in a cruise scenario, which can solve the technical problems in the prior art where reversing planning and forward planning are separated, resulting in the inability to reuse mature forward planning capabilities, as well as the insufficient adaptability and poor trajectory smoothness of existing reversing algorithms in complex dynamic scenarios.
[0009] In a first aspect, embodiments of this application provide a method for planning a reversing trajectory in a cruise scenario, the method comprising: Generate a passable path based on a path search algorithm; Construct a virtual lane based on the passable path; The starting point of the plan is adapted during the gear shift cycle; Based on the virtual lane and the adapted planning starting point, a forward planning scheme is used for decision-making and planning, and the reversing planning is processed into forward planning.
[0010] In conjunction with the first aspect, in one implementation, the generation of a passable path based on the path search algorithm includes a scenario decision-making step: The scenario requiring reversing is determined based on a preset rule scheme; When the sum of the width of the moving obstacle ahead and the width of the vehicle is greater than the width of the road, set an entry sign and begin path search and planning; When the vehicle reaches the vicinity of the destination, set an exit flag to exit the route search and planning.
[0011] In conjunction with the first aspect, in one implementation, the generation of a feasible path based on the path search algorithm includes a boundary decision step: Use the end point of the previous cycle trajectory as the starting point for planning; Find a location on the map behind the vehicle where the lane width is greater than the sum of the vehicle width and the width of the moving obstacle, and set the planned endpoint as the target location near the search boundary within that area; Determine the search boundary, which includes solid lines, physical boundaries, and static obstacles.
[0012] In conjunction with the first aspect, in one implementation, the generation of a passable path based on the path search algorithm includes a path search step: Search for feasible paths based on an asynchronous path search scheme; Calculate the path from the current node to the destination using a curve; Based on the cost function, the node with the lowest cost is selected from the candidate nodes for expansion.
[0013] In conjunction with the first aspect, in one implementation, the generation of a passable path based on the path search algorithm includes a path segmentation step: The search path is segmented according to the direction of movement; Based on whether the vehicle's current position has reached the end of the current segment path, determine whether to set the current segment path as the next segment path.
[0014] In conjunction with the first aspect, in one implementation, the path segmentation step further includes a boundary generation step: The current segment path is offset to the left and right by a preset distance to obtain the left and right boundary lines.
[0015] In conjunction with the first aspect, in one implementation, the boundary generation step further includes boundary constraints: Constrain the Frenet coordinates of the left and right boundary lines so that the Frenet coordinates of the left and right boundary lines do not exceed the Frenet coordinates of the physical boundary at the corresponding positions.
[0016] In conjunction with the first aspect, in one implementation, constructing a virtual lane based on the passable path includes: Set the centerline according to the current segment path; Set the boundary lines according to the left and right boundary lines obtained by offset; Set the boundary line type to solid line.
[0017] In conjunction with the first aspect, in one implementation, adapting the planning starting point during the gear shifting cycle includes: When switching from R to D, the planning starting point coordinates are set to the vehicle's rear axle center positioning coordinates.
[0018] In conjunction with the first aspect, in one implementation, adapting the planning starting point during the gear shifting cycle includes: When switching from D mode to R mode, an adaptation calculation is performed on the planning starting point; The adaptation calculation includes: The planned starting point heading angle is obtained by rotating the rear axle center heading angle by a preset angle offset. Based on the coordinates of the rear axle center, the direction vector corresponding to the rear axle center heading angle, and the virtual distance, the coordinates of the planning starting point are calculated. The virtual distance is calculated based on the vehicle length and the distance from the rear bumper to the center of the rear axle.
[0019] In conjunction with the first aspect, in one implementation, the decision-making planning using a forward planning scheme includes: Set a stop line constraint at the end of the virtual lane; Limit the speed at the end of the planned trajectory to zero.
[0020] Secondly, embodiments of this application provide a reversing trajectory planning system for cruising scenarios, the system comprising: The path generation module is used to generate passable paths based on path search algorithms. A lane construction module is used to construct virtual lanes based on the passable path; The starting point adaptation module is used to adapt the planned starting point during the gear shifting cycle. The decision planning module is used to make decisions and plans based on the virtual lane and the adapted planning starting point, using a forward planning scheme.
[0021] In conjunction with the second aspect, in one implementation, the path generation module includes a scene decision unit, a boundary decision unit, a path search unit, and a path segmentation unit. The scenario decision unit is used to determine the scenario that requires reversing based on the rule scheme, and to enter the path search and planning when the triggering conditions are met; The boundary decision unit is used to determine the planning start point, end point, and search boundary; The path search unit is used to search for a passable path based on an asynchronous path search scheme; The path segmentation unit is used to segment the search path according to the direction of movement.
[0022] Thirdly, embodiments of this application provide a vehicle including the aforementioned reversing trajectory planning system for cruise scenarios.
[0023] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, characterized in that the program, when executed by a processor, implements the reversing trajectory planning method for the cruise scenario.
[0024] The beneficial effects of the technical solutions provided in this application include: By generating passable paths based on the OpenSpace scheme, constructing virtual lanes based on the passable paths, adapting the planning starting point during gear shifting cycles, and using a forward planning scheme to make decision-making based on the virtual lanes and the adapted planning starting point, reversing planning is processed into forward planning. This solves the problem that reversing planning in related technologies is difficult to cope with complex scenarios and cannot effectively reuse mature forward planning capabilities. Thus, without affecting the original forward decision-making planning scheme, the forward planning capability is extended to the reversing planning scenario, improving the reversing performance in parking lot cruising and public road scenarios. Attached Figure Description
[0025] Figure 1 This is a schematic diagram of the functional modules of an embodiment of the reversing trajectory planning system for cruise scenarios in this application; Figure 2 This is a flowchart illustrating an embodiment of the reversing trajectory planning method for cruise scenarios in this application; Figure 3 This is a schematic diagram of a scenario where a vehicle reverses to yield in a narrow lane, as provided in an embodiment of the present invention. Detailed Implementation
[0026] 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 are within the scope of protection of the present application.
[0027] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0028] In one aspect, embodiments of this application provide a reversing trajectory planning system for cruise scenarios.
[0029] In one embodiment, reference is made to Figure 1 , Figure 1 This is a functional module diagram of an embodiment of the reversing trajectory planning system for cruise scenarios according to this application. Figure 1 As shown, the system can be applied to the on-board controller of intelligent driving vehicles, and mainly includes a path generation module 101, a lane construction module 102, a starting point adaptation module 103, and a decision planning module 104.
[0030] The path generation module 101 is used to generate a passable path based on a path search algorithm. This module further includes the following units: The scenario decision unit is used to determine the scenarios in which reversing is required based on preset decision rules. Specifically, this unit receives obstacle information and road information from the perception module. When the sum of the width of the moving obstacle in front and the width of the vehicle is greater than the width of the road, an entry flag is set and the vehicle enters path search and planning; when the vehicle reaches the vicinity of the destination, an exit flag is set and the vehicle exits path search and planning.
[0031] The boundary decision unit is used to determine the planning start point, end point, and search boundary. This unit uses the end point of the previous cycle's trajectory as the planning start point, searches the map behind the vehicle for a target area where the lane width is greater than the sum of the vehicle's width and the width of the moving obstacle, and sets the planning end point as the target location within that target area, close to the search boundary. The search boundary includes solid lines, physical boundaries, and static obstacles.
[0032] The path search unit is used to search for traversable paths based on an asynchronous path search scheme. This unit calculates the path from the current node to the destination using Reed-Shepp curves, expands nodes based on vehicle kinematics, and selects the node with the lowest cost from the candidate nodes for expansion based on a cost function. The asynchronous search scheme ensures that path search does not affect the main program's execution time.
[0033] The path segmentation unit is used to segment the search path according to the direction of movement. This unit determines whether to set the current segment path as the next segment path based on whether the vehicle's current position has reached the end point of the current segment path.
[0034] The lane construction module 102 is used to construct virtual lanes based on the passable path. This module is used to offset the current segmented path to the left and right by preset distances to obtain left and right boundary lines; constrain the Frenet coordinates of the left and right boundary lines so that the Frenet coordinates of the left and right boundary lines do not exceed the Frenet coordinates of the physical boundary at the corresponding position; set the center line according to the current segmented path, set the boundary lines according to the offset left and right boundary lines, and set the boundary line type to solid line type.
[0035] The starting point adaptation module 103 is used to adapt the planned starting point during gear shifting cycles. This module sets the planned starting point coordinates to the rear axle center positioning coordinates when the vehicle shifts from R to D gear; when shifting from D to R gear, it performs adaptation calculations on the planned starting point, rotating the rear axle center heading angle by a preset angle offset to obtain the planned starting point heading angle. Based on the rear axle center coordinates, the direction vector corresponding to the rear axle center heading angle, and a virtual distance, the planned starting point coordinates are calculated, where the virtual distance is determined based on the vehicle length and the distance from the rear bumper to the rear axle center.
[0036] The decision planning module 104 is used to perform decision planning based on the virtual lane and the adapted planning starting point using a forward planning scheme. This module is used to set a stop constraint at the end of the virtual lane, limiting the speed at the end of the planned trajectory to zero, ensuring that the rear axle center of the vehicle can stop at the end position of the virtual lane. This module can reuse existing forward decision planning schemes without modifying the original decision planning architecture.
[0037] The system may also include a sensor interface module for receiving environmental perception data from sensors such as lidar, cameras, ultrasonic radar, and millimeter-wave radar, as well as vehicle status data (such as gear position, vehicle speed, and steering angle) from the vehicle's CAN bus. The sensor interface module transmits the processed data to the path generation module 101 and the starting point adaptation module 103.
[0038] The system may also include a control execution interface module, which converts the trajectory planning results output by the decision planning module 104 into vehicle control commands, including steering control commands, throttle control commands and brake control commands, and sends them to the corresponding actuators via the vehicle CAN bus.
[0039] Secondly, this application also provides a method for planning reversing trajectory in a cruise scenario.
[0040] In one embodiment, reference is made to Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the reversing trajectory planning method for a cruise scenario according to this application. Figure 2 As shown, the reversing trajectory planning method for cruise scenarios includes: Step S201: Generate a passable path based on the path search algorithm.
[0041] This step aims to solve the reachability problem of the initial reversing path in complex scenarios. In this embodiment, the path search algorithm can be a search algorithm based on the OpenSpace framework (such as the HybridA* algorithm), or other search algorithms capable of generating collision-free paths such as HybridA* and RRT*. The following detailed explanation uses the OpenSpace scheme as an example. This step specifically includes four sub-steps: scene decision, boundary decision, path search, and path segmentation.
[0042] When making scenario decisions, the system determines the scenario requiring reversing based on preset decision rules. These preset decision rules can be state machine judgments based on if-then logic, running in the decision planning thread. Specifically, the system receives obstacle and road information transmitted in real time from the perception module. When the sum of the width of a moving obstacle ahead and the width of the vehicle is greater than the current road width, it is determined to be a narrow road meeting scenario. At this time, the system sets the entry flag flg_open_space_enable=true, triggering entry into path search and planning mode. When the vehicle reaches near the destination (e.g., less than a preset threshold, such as 0.5 meters), the reversing task is determined to be complete, the exit flag flg_open_space_enable=false is set, exiting path search and planning mode and resuming normal driving planning. By setting the flag mechanism, a smooth switch between reversing planning mode and normal driving mode can be achieved, avoiding state jumps.
[0043] When making boundary decisions, the endpoint of the previous cycle's trajectory is used as the planning starting point. A target area is located on the map behind the vehicle where the lane width is greater than the sum of the vehicle's width and the width of the moving obstacle. The planning endpoint is set as the target position within this target area, close to the search boundary. It should be noted that the target position close to the search boundary here refers to a location within the target area with a large lateral offset, allowing more space for obstacles to pass, such as a position maintaining a safe distance of 0.5 meters from the search boundary. When determining the search boundary, the search boundary includes solid road lines, physical boundaries (such as curbs and walls), and static obstacles (such as cones and water-filled barriers). The path generated during the search process must not exceed the search boundary.
[0044] An asynchronous approach is used during path search to address the issue that path search algorithms (such as HybridA*) are computationally expensive and cannot meet the high-frequency operation requirements of decision planning. Asynchronous search runs in an independent thread or low-priority task, ensuring that the path search process does not affect the main program's execution time and real-time performance. The specific search process includes calculating the path from the current node to the destination using Reed-Shepp curves or other obstacle-free curves as a heuristic cost; expanding nodes based on vehicle kinematics models (such as the Ackerman steering model) to ensure the generated path conforms to vehicle motion constraints; and selecting the node with the lowest cost from candidate nodes based on the cost function for further expansion.
[0045] The cost function typically includes a weighted sum of multiple factors, for example: Cost = w1\Distance + w2\Curvature + w3\Obstacle_Risk. Here, Distance is the path length, used to optimize path efficiency; Curvature is the path curvature, used to optimize path smoothness and reduce sharp steering wheel movements; Obstacle_Risk is the obstacle risk cost, used to ensure safety; and w1, w2, and w3 are weighting coefficients that can be adjusted according to scenario requirements.
[0046] When segmenting the path, the search path is divided into segments based on the direction of movement. Specifically, since the path search results may contain multiple segments of forward and backward movement (e.g., moving forward to adjust the angle, then reversing to enter the parking space), it is necessary to determine whether to set the current segment as the next segment based on whether the vehicle's current position has reached the end point of the current segment. Through segmentation, the gear requirement (D or R) for each stage can be clearly defined, providing a basis for subsequent gear shifting and starting point adaptation.
[0047] Step S202: Construct a virtual lane based on the passable path.
[0048] This step aims to convert the reversing path into lane information that the forward planner can recognize, thereby reusing the forward planning capability.
[0049] In this embodiment, the Frenet coordinates of the left and right boundary lines are constrained so that the Frenet coordinates of the left and right boundary lines do not exceed the Frenet coordinates of the physical boundary at the corresponding positions.
[0050] Specifically, the generated left and right boundary lines are projected onto the reference line, and their lateral offset (L value) is calculated. If this offset exceeds the lateral offset of the physical boundary at the same longitudinal position (S value), the coordinates of the boundary line are clipped to the physical boundary. This constraint ensures that the constructed virtual lane's left and right boundary lines do not exceed the physical boundary, guaranteeing the lane's legality.
[0051] A centerline is set based on the current segmented path, and boundary lines are set based on the offset left and right boundary lines. The boundary line type is set to solid line, thereby constructing a virtual lane that meets the requirements of the forward planning input. For example, the offset left and right boundary lines are obtained by offsetting 1.8 meters to the left and right of the current segmented path, respectively.
[0052] Setting the boundary line type to solid line is to inform the forward planner that the lane cannot be changed arbitrarily and that the vehicle must stay within the lane, which conforms to the constraint characteristics of the reversing scenario.
[0053] Step S203: Adapt the planned starting point during the gear shifting cycle.
[0054] In this embodiment, trajectory planning typically uses trajectory splicing to ensure the smoothness of the output trajectory. This invention adapts the planning starting point during gear shifting cycles, ensuring that both forward and reverse planning meet the requirements of forward planning.
[0055] When switching from R to D, the coordinates of the trajectory point from the previous cycle are behind the vehicle, with the heading angle pointing backward, which does not meet the requirements of the current cycle's forward-oriented planning. In this case, the planning starting point coordinates are set to the vehicle's rear axle center positioning coordinates.
[0056] When switching from D to R, the coordinates of the trajectory point from the previous cycle are in front of the vehicle, with the heading angle pointing forward, which does not meet the requirement of planning towards the rear of the vehicle in the current cycle. At this time, an adaptation calculation is performed on the planning starting point, mapping the planning starting point to a virtual position behind the vehicle.
[0057] The heading angle at the planning starting point is calculated using the following formula (1): Formula (1).
[0058] The coordinates of the planning starting point are calculated using the following formula (2): Formula (2).
[0059] The virtual distance is calculated using the following formula (3): Formula (3).
[0060] in, Indicates the heading angle at the starting point of the plan; Indicates the coordinates of the starting point of the plan; Indicates the heading angle at the center of the rear axle; Indicates the coordinates of the rear axle center; This represents the direction vector corresponding to the heading angle at the rear axle center; This represents the distance from the virtual rear axle center to the actual rear axle center; Indicates the length of the vehicle; This indicates the distance from the rear bumper to the center of the rear axle; Step S204: Based on the virtual lane and the adapted planning starting point, make a decision-making plan using the forward planning scheme.
[0061] This step aims to leverage mature forward planning capabilities to generate the final reversing trajectory.
[0062] In this embodiment, regardless of whether the decision-making and planning scheme is based on spatiotemporal separation or spatiotemporal joint decision-making and planning, the overall scheme does not need to be modified. This means that the method has good compatibility and there is no need to redevelop a decision-making and planning module 104 specifically for reversing.
[0063] All that's needed is to set a stop constraint at the end of the virtual lane. Specifically, the speed at the end of the planned trajectory is limited to zero to ensure that the rear axle center of the vehicle stops at the end of the virtual lane.
[0064] This stopping constraint serves as a boundary condition for the optimization problem, ensuring that the planned trajectory smoothly reduces its speed to zero at the endpoint, thus avoiding abrupt stops.
[0065] This step extends all the processing capabilities of forward planning (such as obstacle avoidance, smoothing, and constraint handling) to the reversing planning scenario, significantly improving the performance and stability of reversing planning.
[0066] Figure 3 This is a schematic diagram illustrating a scenario where a vehicle reverses to yield in a narrow lane, as provided in an embodiment of the present invention. Figure 3 As shown, a vehicle is traveling on a narrow road and encounters an oncoming vehicle or obstacle ahead, where the road width is insufficient to support two vehicles driving side-by-side. The specific execution flow of this embodiment is as follows: During the scenario triggering phase, the system continuously monitors the sum of the width of the obstacle ahead and the width of the vehicle. When the sum is greater than the current road width, it is determined to be a narrow road meeting scenario, triggering the reversing planning process.
[0067] During the path planning phase, the system generates a collision-free path that includes a reversing segment based on the OpenSpace algorithm. This path ensures that the vehicle does not collide with surrounding obstacles during the reversing process.
[0068] During the lane construction phase, virtual lanes are constructed based on the generated collision-free paths. In this process, the boundaries of the virtual lanes are constrained within the physical road edges to ensure the legality and safety of the lanes.
[0069] During the starting point adaptation phase, when the vehicle switches from forward to reverse, the system calculates the adapted planned starting point using the adaptation formula from the aforementioned embodiment. This step ensures that the reverse planned starting point is consistent with the forward planned coordinate system, guaranteeing smooth trajectory stitching.
[0070] During the trajectory execution phase, the decision planning module 104 reuses the forward planning capability to control the vehicle to reverse along the virtual lane to the target location and stop. Through the above process, the vehicle makes way for oncoming vehicles, completing the reversing yielding operation.
[0071] As can be seen from this embodiment, the solution of this application can effectively cope with complex scenarios such as narrow roads and achieve smooth and safe reversing planning.
[0072] In summary, the beneficial effects of the technical solution provided in this application include: This application employs a path search-based algorithm (such as the OpenSpace scheme) to generate initial passable paths. Unlike traditional planning methods based on fixed lane lines, this algorithm comprehensively considers vehicle kinematic constraints and complex obstacle distributions. Therefore, it can solve the initial path passability problem in complex scenarios such as narrow spaces and areas without standard lane lines with high accuracy. This feature is not only applicable to reversing planning scenarios but can also be extended to special forward planning scenarios beyond the scope of standard lane models (such as obstacle avoidance and driving on unstructured roads), significantly enhancing the system's scenario adaptability.
[0073] This application cleverly transforms the reversing planning problem into a forward planning problem by constructing a virtual lane based on a passable path and adapting the planning starting point during gear shifting cycles. The principle lies in constructing a virtual lane structure (centerline, boundary lines) that meets the input requirements of the forward planner and a planning starting point with a consistent coordinate system. Therefore, it can effectively adapt to forward decision-making and planning schemes based on rule-based roads with minimal modifications, eliminating the need to redevelop a separate reversing decision-making and planning module 104, thus reducing system development costs and maintenance complexity.
[0074] This application successfully extends the capabilities of forward decision-making and planning schemes to reverse planning scenarios. Since forward planning schemes typically possess mature capabilities in trajectory optimization, constraint handling, and rule adherence, reversing planning can inherit the advantages of forward planning in terms of trajectory smoothness, comfort, and safety by reusing these capabilities. This significantly improves reversing performance in parking lot cruising and public road scenarios, solving the problem of insufficient reversing planning capabilities in complex scenarios in existing technologies.
[0075] This application employs an asynchronous search scheme for path generation, decoupling time-consuming path search calculations (such as HybridA* node expansion) from the main decision-making and planning process. The principle is that the search is performed in an independent thread or low-priority task, while the main program only needs to read the search results. Therefore, this ensures the running frequency and real-time performance of the planning system, avoids increased main program execution time due to complex path searches, and guarantees the stability and response speed of vehicle control.
[0076] In one specific embodiment, the left and right boundary lines are obtained by offsetting the path to the left and right by a preset distance, respectively. The preset distance is not a fixed value, but a dynamic safety offset that is dynamically adjusted according to the surrounding environment.
[0077] Specifically, the system acquires real-time distance information of static obstacles around the vehicle. For each discrete point on the path, the lateral distance from that point to the nearest static obstacle is calculated. The dynamic safety offset is determined by taking the larger of the base safety offset and the calculated value. The base safety offset is half the vehicle width plus a safety margin; the calculated value is the obstacle distance multiplied by a distance sensitivity coefficient plus a minimum offset compensation value. The distance sensitivity coefficient is a value between zero and one, and the minimum offset compensation value is a preset fixed length.
[0078] This dynamic offset method automatically shrinks the virtual lane boundary when the vehicle is in a narrow space and close to an obstacle to avoid conflict with the physical boundary, while ensuring sufficient passage width. When the vehicle is in an open space, the virtual lane boundary widens appropriately, providing more optimization space for the forward planning module, thereby generating a smoother trajectory. Furthermore, if the calculated dynamic safety offset causes the left and right boundary lines to intersect or overlap, the system will automatically mark that segment of the virtual lane as impassable and trigger replanning.
[0079] In another specific embodiment, to solve the data synchronization problem between the asynchronous search thread and the main planning thread, this embodiment adopts a double buffer mechanism for communication.
[0080] Specifically, two storage areas, designated Buffer A and Buffer B, are allocated in system memory to store candidate path data generated by the path search unit. After completing a full path search, the path search thread writes the newly generated path data to the currently free buffer and updates the version number flag. At the beginning of each planning cycle, the main planning thread reads the version number flag. If the version number has changed, the main planning thread switches to the corresponding buffer to read the latest path data; if the version number has not changed, it continues to use the data from the buffer of the previous cycle.
[0081] Furthermore, to prevent path data from being overwritten by write operations during the reading process, this embodiment introduces a read-write lock mechanism. When the path search thread is writing to the buffer, a write lock is set, and the main planning thread waits for the write lock to be released before reading; when the main planning thread is reading from the buffer, a read lock is set, and the path search thread waits for the read lock to be released before writing. Through this communication buffer mechanism, the asynchronous nature of path search is ensured to not affect the real-time performance of the main thread, while also ensuring data consistency and integrity, avoiding trajectory jumps caused by data contention.
[0082] In another specific embodiment, the forward planning scheme is used for decision planning. Although this scheme reuses the forward planning scheme, considering the special characteristics of the reversing scenario, this embodiment adds differentiated constraints during the reuse process.
[0083] Specifically, when constructing the optimization problem, the decision planning module 104 identifies the gear information corresponding to the current planning segment. If the current segment is a reversing segment, then restrictions such as speed limit constraints, acceleration constraints, and rate of curvature change constraints are added to the constraints of the optimization problem. The speed limit constraint means setting the maximum allowable speed of the reversing segment to be lower than the speed limit of the forward segment, for example, setting it to no more than five kilometers per hour.
[0084] Acceleration constraints refer to limiting the absolute value of longitudinal acceleration during the reversing segment to a low range, such as no more than 1.5 meters per second squared, to prevent excessively abrupt starts or stops while reversing.
[0085] The curvature change rate constraint refers to increasing the penalty weight on the curvature change rate of the path. Since the steering wheels are at the rear when reversing, violent steering wheel rotation can easily lead to uncontrollable rear-end sweep range. Therefore, increasing the weight coefficient of the curvature change rate makes the generated trajectory smoother.
[0086] If the current segment is a forward segment, then the normal speed and acceleration constraints are restored. Through this differentiated constraint, the system ensures the safety and comfort of the reversing process while reusing the forward planning architecture.
[0087] In another specific embodiment, after the decision planning module 104 outputs the final trajectory, the control execution interface module performs post-processing verification on the trajectory before sending the command. The system simulates the vehicle's movement along the trajectory based on its actual kinematic model. Verification includes steering angle limit verification, steering speed limit verification, and collision re-checking. Steering angle limit verification checks whether the required steering angle for all points on the trajectory exceeds the vehicle's maximum physical steering angle limit. Steering speed limit verification checks whether the change in steering angle between adjacent trajectory points divided by the time step exceeds the maximum angular velocity limit of the steering motor. Collision re-checking involves discretizing the vehicle contour along the trajectory to reconfirm whether there is a spatiotemporal conflict with dynamic obstacles. If the verification passes, the trajectory is sent to the execution mechanism; if the verification fails, the current trajectory is discarded, and a replanning request is sent to the path generation module 101, or an emergency braking command is triggered.
[0088] Thirdly, embodiments of this application provide a vehicle, including a reversing trajectory planning system for cruise scenarios.
[0089] In this embodiment, during vehicle operation, when encountering scenarios requiring reversing, such as passing on narrow roads or cruising in parking lots, the reversing trajectory planning system for cruising scenarios acquires information about obstacles ahead and road width information through a perception system. The path generation module 101 generates a passable path based on a path search algorithm. The lane construction module 102 constructs a virtual lane based on the passable path. The starting point adaptation module 103 adapts the planned starting point during gear shifting cycles. The decision planning module 104, based on the virtual lane and the adapted planned starting point, performs decision planning using a forward planning scheme to generate a target trajectory. The target trajectory is transmitted to the vehicle controller, which controls the steering, power, and braking actuators to enable the vehicle to complete the reversing action along the target trajectory.
[0090] Specifically, during vehicle operation, when encountering scenarios requiring reversing, such as passing oncoming traffic on narrow roads, cruising in parking lots, making a U-turn at a dead end, or parking on the side of the road, the reversing trajectory planning system for cruising scenarios executes the following detailed process: Perception and Decision Triggering. The perception system transmits the processed environmental perception data to the path generation module 101 via the sensor interface module. The scene decision unit in the path generation module 101 makes judgments based on preset decision rules. For example, when the sum of the width of the moving obstacle ahead and the width of the vehicle is greater than the current road width, and there are no other passable paths ahead, it is determined to be a narrow road meeting scenario, an entry sign is set, and the reversing planning mode is triggered.
[0091] Path generation and lane construction. The path generation module 101 generates a passable path based on a path search algorithm (such as the asynchronous OpenSpace scheme) and divides the path into forward and backward segments. Subsequently, the lane construction module 102 constructs virtual lanes based on the passable path, converting the reversing path into lane information recognizable by the forward planner, including the center line, left and right boundary lines, and lane attributes (such as solid line type). During this process, the system constrains the coordinates of the boundary lines to ensure that the virtual lanes do not exceed the physical boundaries.
[0092] Starting point adaptation and planning calculation: The starting point adaptation module 103 monitors the vehicle's gear status. During gear shifting cycles (e.g., from D to R, or from R to D), it performs adaptation calculations on the planned starting point, adjusting its coordinates and heading angle to meet the input requirements of the forward planning scheme. The decision planning module 104, based on the virtual lane and the adapted planned starting point, uses a mature forward planning scheme to perform decision planning, generating a target trajectory containing position, speed, and time information. During this process, the system applies differentiated constraints based on the gear information, such as limiting maximum speed and acceleration during reversing segments.
[0093] Trajectory Verification and Control Execution: Before being sent to the actuators, the generated target trajectory may optionally undergo kinematic feasibility verification to ensure that parameters such as steering angle and steering speed are within the vehicle's physical limits. The verified target trajectory is then transmitted to the vehicle control unit (VCU) or chassis domain controller. The VCU decomposes the trajectory into specific control commands, including steering control commands, throttle control commands, and brake control commands, and sends them to the corresponding actuators via the vehicle's CAN bus, enabling the vehicle to complete the reversing maneuver along the target trajectory.
[0094] When a vehicle encounters an oncoming vehicle on a rural road or narrow alley, and the road width is insufficient for two-way traffic, the system automatically plans a reversing path to a nearby wide area or avoidance bay, and then plans a forward path after the oncoming vehicle has passed.
[0095] When a vehicle encounters obstacles while searching for or leaving a parking space in a parking lot, the system uses rear map information to find a passable area and plans a reversing trajectory to the target location, eliminating the need for the driver to frequently switch gears and manually operate the system.
[0096] When a vehicle accidentally enters a dead end, the system automatically identifies the closed boundary ahead and plans a multi-segment combination trajectory that includes forward, backward, and forward movements, enabling it to turn around on the spot or turn around with the minimum radius.
[0097] If a sensor malfunction, location loss, or planning trajectory verification failure is detected during the planning process, the system will automatically exit the intelligent reversing mode, issue an alarm to prompt the driver to take over the vehicle, and control the vehicle to a smooth stop.
[0098] On the human-machine interface, the system can display the planned virtual lane boundaries and target trajectory in real time, allowing the driver to understand the vehicle's intentions. During reversing, if a sudden obstacle is detected entering the planned path, the system will immediately replan or apply emergency braking, and will display obstacle information through the interface.
[0099] The vehicle provided in this embodiment, by integrating the cruise scenario reversing trajectory planning system, can significantly improve its reversing ability in complex scenarios without altering the original forward planning architecture. It not only achieves automated reversing in parking lots and public roads, but also ensures real-time planning and trajectory smoothness through asynchronous search and starting point adaptation technologies, reducing the driver's workload and improving driving safety and comfort.
[0100] Fourthly, embodiments of this application provide a reversing trajectory planning device for cruise scenarios. The reversing trajectory planning device for cruise scenarios can be a personal computer (PC), a laptop computer, a server, or other devices with data processing capabilities.
[0101] In this embodiment, the reversing trajectory planning device for cruise scenarios may include a processor, a memory, a communication interface, and a communication bus.
[0102] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.
[0103] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting internal components of the reversing trajectory planning device in cruise scenarios, as well as interfaces used for interconnecting the reversing trajectory planning device with other devices (such as other computing devices or user devices). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user devices can be displays, keyboards, etc.
[0104] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.
[0105] The processor can be a general-purpose processor, which can call the cruise scene reversing trajectory planning program stored in the memory and execute the cruise scene reversing trajectory planning method provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the cruise scene reversing trajectory planning program is called can be referred to in the various embodiments of the cruise scene reversing trajectory planning method of this application, and will not be repeated here.
[0106] Fifthly, embodiments of this application also provide a computer-readable storage medium.
[0107] The present application stores a cruise scene reversing trajectory planning program on a computer-readable storage medium, wherein when the cruise scene reversing trajectory planning program is executed by a processor, it implements the steps of the cruise scene reversing trajectory planning method described above.
[0108] The method implemented when the reversing trajectory planning program for the cruise scenario is executed can be referred to in the various embodiments of the cruise scenario reversing trajectory planning method of this application, and will not be repeated here.
[0109] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0110] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.
[0111] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.
[0112] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.
[0113] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.
[0114] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.
[0115] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A method for planning a reversing trajectory in a cruising scenario, characterized in that, The method includes: Generate a passable path based on a path search algorithm; Construct a virtual lane based on the passable path; The starting point of the plan is adapted during the gear shift cycle; Based on the virtual lane and the adapted planning starting point, a forward planning scheme is used for decision-making and planning, and the reversing planning is processed into forward planning.
2. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The generation of a passable path based on the path search algorithm includes the following scenario decision-making steps: The scenario requiring reversing is determined based on a preset rule scheme; When the sum of the width of the moving obstacle ahead and the width of the vehicle is greater than the width of the road, set an entry sign and begin path search and planning; When the vehicle reaches the vicinity of the destination, set an exit flag to exit the route search and planning.
3. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The generation of a passable path based on the path search algorithm includes a boundary decision step: Use the end point of the previous cycle trajectory as the starting point for planning; Find a location on the map behind the vehicle where the lane width is greater than the sum of the vehicle width and the width of the moving obstacle, and set the planned endpoint as the target location near the search boundary within that area; Determine the search boundary, which includes solid lines, physical boundaries, and static obstacles.
4. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The generation of a passable path based on the path search algorithm includes the following path search steps: Search for feasible paths based on an asynchronous path search scheme; Calculate the path from the current node to the destination using a curve; Based on the cost function, the node with the lowest cost is selected from the candidate nodes for expansion.
5. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The generation of a passable path based on the path search algorithm includes a path segmentation step: The search path is segmented according to the direction of movement; Based on whether the vehicle's current position has reached the end of the current segment path, determine whether to set the current segment path as the next segment path.
6. The method for planning reversing trajectory in a cruising scenario as described in claim 5, characterized in that, The path segmentation step also includes a boundary generation step: The current segment path is offset to the left and right by a preset distance to obtain the left and right boundary lines.
7. The method for planning reversing trajectory in a cruising scenario as described in claim 6, characterized in that, The boundary generation step also includes boundary constraints: Constrain the Frenet coordinates of the left and right boundary lines so that the Frenet coordinates of the left and right boundary lines do not exceed the Frenet coordinates of the physical boundary at the corresponding positions.
8. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, Constructing a virtual lane based on the passable path includes: Set the centerline according to the current segment path; Set the boundary lines according to the left and right boundary lines obtained by offset; Set the boundary line type to solid line.
9. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The adaptation of the planning starting point during the gear shift cycle includes: When switching from R to D, the planning starting point coordinates are set to the vehicle's rear axle center positioning coordinates.
10. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The adaptation of the planning starting point during the gear shift cycle includes: When switching from D mode to R mode, an adaptation calculation is performed on the planning starting point; The adaptation calculation includes: The planned starting point heading angle is obtained by rotating the rear axle center heading angle by a preset angle offset. Based on the coordinates of the rear axle center, the direction vector corresponding to the rear axle center heading angle, and the virtual distance, the coordinates of the planning starting point are calculated. The virtual distance is calculated based on the vehicle length and the distance from the rear bumper to the center of the rear axle.
11. The method for planning reversing trajectory in a cruising scenario as described in claim 1, characterized in that, The decision-making and planning using forward programming includes: Set a stop line constraint at the end of the virtual lane; Limit the speed at the end of the planned trajectory to zero.
12. A reversing trajectory planning system for cruising scenarios, characterized in that, The system includes: The path generation module is used to generate passable paths based on path search algorithms. A lane construction module is used to construct virtual lanes based on the passable path; The starting point adaptation module is used to adapt the planned starting point during the gear shifting cycle. The decision planning module is used to make decisions and plans based on the virtual lane and the adapted planning starting point, using a forward planning scheme.
13. The reversing trajectory planning system for cruise scenarios as described in claim 12, characterized in that, The path generation module includes a scene decision unit, a boundary decision unit, a path search unit, and a path segmentation unit; The scenario decision unit is used to determine the scenario that requires reversing based on the rule scheme, and to enter the path search and planning when the triggering conditions are met; The boundary decision unit is used to determine the planning start point, end point, and search boundary; The path search unit is used to search for a passable path based on an asynchronous path search scheme; The path segmentation unit is used to segment the search path according to the direction of movement.
14. A vehicle, characterized in that, Including the reversing trajectory planning system for cruise scenarios as described in claim 12 or 13.
15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the reversing trajectory planning method for the cruise scenario as described in any one of claims 1 to 11.