A local path planning obstacle avoidance method and device based on a drivable area
By acquiring the robot dog's pose and environmental information, and using the drivable area boundary and obstacle parameters to determine the translation direction, the problem of the robot dog quickly avoiding obstacles and returning to the planned path under near-distance obstacles is solved, thus improving the robot dog's obstacle avoidance robustness and motion continuity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUARUAN TECH CO LTD
- Filing Date
- 2026-05-06
- Publication Date
- 2026-07-21
AI Technical Summary
Existing technologies for obstacle avoidance in robot dogs suffer from several drawbacks. In scenarios involving sudden obstacles at close range, emergency obstacle avoidance trajectories are difficult to generate quickly, there is a lack of flexible obstacle avoidance adjustment mechanisms such as lateral movement, and it is difficult to quickly return to the original planned path after the obstacle is cleared. These issues result in insufficient continuity of movement and safety.
By acquiring the pose information, environmental perception information, and drivable area map of the mobile robot, the obstacle avoidance trigger conditions are determined and the robot enters the translation and obstacle avoidance mode. The translation direction is determined by using the boundary of the drivable area and the lateral distribution parameters of the obstacles, and the lateral translation is executed. After the obstacles are cleared, the local path planning is restored.
It enables the robot to quickly form emergency obstacle avoidance actions in the event of sudden obstacles at close range, ensuring the continuity and safety of movement, avoiding stagnation, and quickly returning to the original planned path, thereby improving the robot dog's obstacle avoidance robustness in complex environments.
Smart Images
Figure CN122431387A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of path planning and obstacle avoidance technology, and in particular to a local path planning and obstacle avoidance method and apparatus based on drivable areas. Background Technology
[0002] As the demand for autonomous driving of robot dogs (quadruped robots) in complex environments increases, robot dogs typically need to perform local path planning based on environmental perception and positioning information during driving, and achieve real-time obstacle avoidance on the local planned path, thereby ensuring the safety and continuity of driving.
[0003] In related technologies, existing local path planning and obstacle avoidance schemes for autonomous driving / mobile robots mainly employ the Dynamic Window Method (DWA) and the Artificial Potential Field Method. The DWA method filters feasible speed spaces based on vehicle dynamics constraints, predicts short-term trajectories by sampling speed combinations, and selects the optimal path by combining safety and efficiency indicators. The Artificial Potential Field Method constructs a virtual potential field environment, using the repulsive force of obstacles on the vehicle and the attractive force of the target point as path guidance to drive the vehicle to generate a smooth obstacle avoidance trajectory. In addition, some schemes also employ optimization algorithms to solve for the optimal trajectory that satisfies the constraints by constructing an objective function. However, existing local path planning technologies applied to robot dogs have significant limitations in adaptability: traditional methods such as dynamic window methods and artificial potential field methods cannot quickly generate emergency obstacle avoidance trajectories that are adapted to the robot dog's motion characteristics when encountering sudden obstacles at close range, because trajectory generation relies on preset sampling or virtual potential field guidance, which can easily cause the robot dog to stop. At the same time, these methods lack flexible adjustment mechanisms under sudden close-range obstacles, and cannot avoid obstacles through emergency actions such as lateral movement. They can only stop and wait or forcibly detour, which not only affects the continuity of movement, but may also cause posture imbalance due to being forced to stop. Moreover, it is difficult to quickly return to the original planned path after the obstacle is cleared, and the robustness of adapting to the dynamic obstacle avoidance scenario of robot dogs is insufficient.
[0004] Therefore, in the local path planning and obstacle avoidance of robot dogs, the difficulty in quickly generating emergency obstacle avoidance trajectories in the case of sudden obstacles at close range, the lack of flexible obstacle avoidance adjustment mechanisms such as lateral movement, and the difficulty in quickly returning to the original planned path after the obstacle is cleared have become urgent problems to be solved. Summary of the Invention
[0005] This application provides a local path planning and obstacle avoidance method and apparatus based on drivable areas, aiming to solve the problems in the existing technology of local path planning and obstacle avoidance in robot dogs, such as the difficulty in quickly generating emergency obstacle avoidance trajectories in the case of sudden obstacles at close range, the lack of flexible obstacle avoidance adjustment mechanisms such as lateral movement, and the difficulty in quickly returning to the original planned path after the obstacle is cleared.
[0006] Firstly, a local path planning obstacle avoidance method based on a drivable area, the method comprising:
[0007] The pose information, environmental perception information, and drivable area map of the mobile robot are acquired, wherein the drivable area map includes the drivable area boundary.
[0008] During the process of controlling the mobile robot to move along the planned route in local path planning, it is determined whether the obstacle avoidance triggering conditions are met based on the environmental perception information.
[0009] When the obstacle avoidance triggering condition is met, the mobile robot is controlled to enter the translational obstacle avoidance mode;
[0010] In the translational obstacle avoidance mode, the translational space parameters of the mobile robot in the left and right directions are determined based on the boundary of the drivable area, and the lateral distribution parameters of the obstacles relative to the mobile robot are determined based on the environmental perception information.
[0011] The translation direction is determined based on the translational space parameters and the lateral distribution parameters, and the mobile robot is controlled to perform lateral translation along the translation direction.
[0012] During the lateral translation process, the changes in obstacles along the translation path are continuously monitored and the translation direction is updated based on the monitoring results;
[0013] When the obstacle avoidance condition is met, the lateral translation stops and the local path planning is resumed to generate a subsequent driving trajectory. The mobile robot is then controlled to move along the subsequent driving trajectory. The driving direction corresponding to the planned driving route is the tangential direction of the planned driving route determined by the local path planning at the current position of the mobile robot.
[0014] In the above scheme, optionally, the pose information includes at least one of position, linear velocity, angular velocity, heading angle and attitude angle; the environmental perception information includes at least one of the position, size and contour point set of obstacles; and the pose information, the environmental perception information and the drivable area map are synchronized in time.
[0015] Optionally, in the above scheme, the obstacle avoidance triggering conditions include: the obstacle obstructs the planned driving route, and the shortest straight-line distance between the mobile robot and the obstacle is less than a preset safe distance threshold.
[0016] Optionally, in the above scheme, the preset safe distance threshold is associated with the current speed of the mobile robot, and the greater the current speed, the greater the preset safe distance threshold.
[0017] Optionally, in the above scheme, the translatable spatial parameters include a first translatable distance and a second translatable distance, and determining the first translatable distance and the second translatable distance includes:
[0018] Based on the real-time positioning of the mobile robot, the shortest distance from the left outer contour of the mobile robot to the boundary of the left drivable area is calculated as the first translational distance.
[0019] The shortest distance from the right outer contour of the mobile robot to the right drivable area boundary is calculated as the second translational distance.
[0020] Based on the comparison results of the first and second translatable distances with the first passage threshold, the passability in the left and right directions is determined, wherein the first passage threshold is related to the width of the mobile robot.
[0021] Optionally, in the above scheme, the lateral distribution parameters include a first lateral relative distance and a second lateral relative distance, and determining the first lateral relative distance and the second lateral relative distance includes:
[0022] Establish a local coordinate system with the center of mass of the mobile robot as the origin. The longitudinal axis of the local coordinate system is consistent with the driving direction corresponding to the planned driving route, and the lateral axis is perpendicular to the longitudinal axis.
[0023] The left and right endpoints of the obstacle in the horizontal axis direction are determined based on the contour point set of the obstacle;
[0024] The first lateral relative distance is determined based on the coordinate difference between the left endpoint and the centroid in the lateral axis direction, and the second lateral relative distance is determined based on the coordinate difference between the right endpoint and the centroid in the lateral axis direction.
[0025] In the above scheme, optionally, determining the translation direction includes:
[0026] If passage is only possible from the left, determine to shift to the left; if passage is only possible from the right, determine to shift to the right.
[0027] When passage is possible in both left and right directions: when the difference between the first lateral relative distance and the second lateral relative distance is greater than a preset difference threshold, it is determined to move to the side with the smaller lateral relative distance; when the difference between the first lateral relative distance and the second lateral relative distance is not greater than the preset difference threshold, it is determined to move to the side with the larger possible translation distance.
[0028] Optionally, controlling the mobile robot to perform lateral translation along the translation direction includes: determining an initial translation speed, wherein the initial translation speed does not exceed a preset proportion of a preset maximum lateral translation speed.
[0029] Optionally, the above scheme may also include the following during the lateral translation process:
[0030] The attitude angle of the mobile robot is adjusted based on the output of the inertial measurement unit so that the orientation of the mobile robot body and the boundary of the drivable area meet the preset parallelism constraint.
[0031] The lateral distance between the mobile robot and the boundary of the drivable area is controlled to be no less than a preset boundary safety distance threshold, wherein the preset boundary safety distance threshold is related to the vehicle width;
[0032] When a new obstacle is detected entering the translation path and the distance between the mobile robot and the new obstacle is less than the translation mode safety threshold, the mobile robot is controlled to decelerate or stop, and the translation space parameters and the lateral distribution parameters are re-determined to update the translation direction.
[0033] Secondly, a local path planning obstacle avoidance device based on a drivable area, the device comprising:
[0034] The acquisition module is used to acquire pose information, environmental perception information, and a map of the drivable area including the boundaries of the drivable area.
[0035] The triggering determination module is used to determine whether the obstacle avoidance triggering conditions are met based on the environmental perception information during the local path planning process, and to trigger the entry into the translational obstacle avoidance mode when the obstacle avoidance triggering conditions are met.
[0036] The parameter calculation module is used to determine the translatable space parameters in the left and right directions based on the boundary of the drivable area in the translation obstacle avoidance mode, and to determine the lateral distribution parameters of the obstacles relative to the mobile robot based on the environmental perception information.
[0037] The decision module is used to determine the translation direction based on the translational spatial parameters and the lateral distribution parameters;
[0038] The execution and monitoring module is used to control the mobile robot to perform lateral translation along the translation direction, and continuously monitor the changes of obstacles on the translation path to update the translation direction;
[0039] The regression module is used to stop lateral translation and resume local path planning to generate subsequent driving trajectories when the obstacle avoidance conditions are met.
[0040] Compared with the prior art, this application has at least the following beneficial effects:
[0041] Based on further analysis and research of existing technical problems, this application recognizes that existing technologies in robot dog local path planning and obstacle avoidance suffer from several issues: difficulty in quickly generating emergency obstacle avoidance trajectories in near-field sudden obstacle scenarios; lack of flexible obstacle avoidance adjustment mechanisms such as lateral movement; and difficulty in quickly returning to the original planned path after obstacle removal. This application addresses these issues by continuously acquiring pose information, environmental perception information, and a map of the drivable area including its boundaries during the local path planning motion of the mobile robot along the planned route. Based on this, the system determines whether obstacle avoidance triggering conditions are met. When the triggering conditions are met, the system no longer relies on the original local planning to continue searching for a feasible trajectory on the planned route affected by obstacles. Instead, it directly switches to a translational obstacle avoidance mode. In this mode, the drivable area boundaries are used to determine the lateral translational space parameters, while simultaneously utilizing the surrounding environment... The environment perception information determines the lateral distribution parameters of obstacles relative to the robot, and determines the translation direction based on two types of parameters, thereby controlling the robot to perform lateral translation along that direction. This process is equivalent to, when an obstacle obstructs the planned driving route and the distance is close, prioritizing the search for a "lateral escape path" within the current drivable area that does not cross the area occupied by the obstacle, enabling the robot to perform an emergency obstacle avoidance action. At the same time, during the lateral translation, the changes in obstacles on the translation path are continuously monitored and the translation direction is updated according to the monitoring results, so that when the environment changes further or new obstacles appear, the robot can still adjust the bypass action in time without stopping. When the obstacle avoidance condition is met, the lateral translation stops and the local path planning is restored to generate the subsequent driving trajectory, so that the robot returns to the local planning control link of the planned driving route after bypassing the obstacle and continues to move forward.
[0042] This solution employs a closed-loop mechanism of "trigger switching, determining the translation direction based on the drivable area and the lateral distribution of obstacles, dynamic updating during the translation process, and restoring the local planning after completion." This mechanism enables the robot to quickly form and execute lateral obstacle avoidance actions even in scenarios where sudden obstacles at close range obstruct the planned route, and traditional local planning struggles to generate feasible obstacle avoidance trajectories in a timely manner, easily leading to forced stops. After avoiding obstacles, the robot automatically returns to the local planned trajectory for tracking. This addresses the problems in existing technologies, such as "difficulty in quickly generating emergency obstacle avoidance trajectories under sudden obstacles at close range, lack of flexible adjustment mechanisms, and difficulty in quickly returning to the original planned path after obstacle avoidance." Attached Figure Description
[0043] Figure 1 A flowchart illustrating a local path planning obstacle avoidance method based on drivable areas provided in one embodiment of this application;
[0044] Figure 2 A flowchart illustrating a local path planning and obstacle avoidance algorithm for a robot dog based on a drivable area, provided in one embodiment of this application;
[0045] Figure 3 A diagram of the drivable area provided for one embodiment of this application. Detailed Implementation
[0046] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0047] In one embodiment, such as Figure 1 As shown, a local path planning obstacle avoidance method based on drivable areas is provided, including the following steps:
[0048] The pose information, environmental perception information, and drivable area map of the mobile robot are acquired, wherein the drivable area map includes the drivable area boundary.
[0049] During the process of controlling the mobile robot to move along the planned route in local path planning, it is determined whether the obstacle avoidance triggering conditions are met based on the environmental perception information.
[0050] When the obstacle avoidance triggering condition is met, the mobile robot is controlled to enter the translational obstacle avoidance mode;
[0051] In the translational obstacle avoidance mode, the translational space parameters of the mobile robot in the left and right directions are determined based on the boundary of the drivable area, and the lateral distribution parameters of the obstacles relative to the mobile robot are determined based on the environmental perception information.
[0052] The translation direction is determined based on the translational space parameters and the lateral distribution parameters, and the mobile robot is controlled to perform lateral translation along the translation direction.
[0053] During the lateral translation process, the changes in obstacles along the translation path are continuously monitored and the translation direction is updated based on the monitoring results;
[0054] When the obstacle avoidance condition is met, the lateral translation stops and the local path planning is resumed to generate a subsequent driving trajectory. The mobile robot is then controlled to move along the subsequent driving trajectory. The driving direction corresponding to the planned driving route is the tangential direction of the planned driving route determined by the local path planning at the current position of the mobile robot.
[0055] In one embodiment, the mobile robot may be a quadruped robot, a wheeled robot, or a mobile platform with lateral translation / transverse control capabilities. Pose information may be output by a fusion localization module, and environmental perception information may be obtained by lidar, depth camera, millimeter-wave radar, or multi-sensor fusion; the drivable area map may be output by a SLAM module, and may be represented as the polygonal boundary of the drivable area, the outer contour of the drivable area in a grid map, or a closed polyline composed of boundary point sets.
[0056] When the robot moves along the planned route under local path planning control, it periodically executes obstacle avoidance trigger judgments: the planned route can be discretized into a sequence of trajectory points or a set of trajectory line segments; the obstacle contour can be represented as a point cloud / polygon, and optional safety expansion (such as expanding the radius according to the robot's outer shape) can be performed for conservative judgment. If the judgment meets the trigger condition, the robot switches from "local planning and tracking state" to "translation obstacle avoidance mode". In translation obstacle avoidance mode, the means of translation space in the left and right directions are first determined according to the boundary of the drivable area: for example, the left and right edge points of the robot's outer contour (or circumscribed rectangle) are projected onto the polyline of the drivable area boundary, and the shortest distance to the left and right boundaries is calculated to form "translation space parameters". Then, the distribution parameters of the obstacle relative to the robot in the lateral direction (such as the left / right lateral relative distance) are calculated according to the obstacle contour, and the translation direction is selected according to the preset direction decision logic to generate lateral translation control commands (such as speed command vy, acceleration limit, duration or target lateral displacement, etc.).
[0057] During lateral translation, the robot continuously updates its perception information and monitors the "translation path occupancy": the predicted occupied strip area in the translation direction can be used as the translation path. If a new obstacle is detected entering this strip area, a direction update process is triggered (e.g., after stopping or slowing down, the translatable space parameters and lateral distribution parameters are recalculated, and the translation direction is redefined). When the obstacle avoidance condition is met, the robot stops lateral translation, reactivates the local path planning module, generates the subsequent driving trajectory starting from the current position and using the global path / original local target as constraints, and performs tracking.
[0058] This embodiment introduces a closed-loop mechanism of "triggering, translational obstacle avoidance, monitoring and updating, and restoring local planning after obstacle avoidance" during the local planning process. This enables the robot to obtain executable emergency obstacle avoidance actions even when the planned route is obscured by obstacles at close range, and automatically return to the local planning trajectory tracking process after obstacle avoidance. This improves the continuity of movement and the availability of obstacle avoidance in sudden obstacle scenarios.
[0059] In this embodiment, the pose information includes at least one of position, linear velocity, angular velocity, heading angle, and attitude angle; the environmental perception information includes at least one of the position, size, and contour point set of obstacles; and the pose information, the environmental perception information, and the drivable area map are synchronized in time.
[0060] In this embodiment, the obstacle avoidance triggering conditions include: an obstacle obstructing the planned driving route, and the shortest straight-line distance between the mobile robot and the obstacle being less than a preset safe distance threshold.
[0061] In this embodiment, the preset safe distance threshold is associated with the current speed of the mobile robot, and the higher the current speed, the higher the preset safe distance threshold.
[0062] In this embodiment, the translatable spatial parameters include a first translatable distance and a second translatable distance, and determining the first translatable distance and the second translatable distance includes:
[0063] Based on the real-time positioning of the mobile robot, the shortest distance from the left outer contour of the mobile robot to the boundary of the left drivable area is calculated as the first translational distance.
[0064] The shortest distance from the right outer contour of the mobile robot to the right drivable area boundary is calculated as the second translational distance.
[0065] Based on the comparison results of the first and second translatable distances with the first passage threshold, the passability in the left and right directions is determined, wherein the first passage threshold is related to the width of the mobile robot.
[0066] In this embodiment, the lateral distribution parameters include a first lateral relative distance and a second lateral relative distance, and determining the first lateral relative distance and the second lateral relative distance includes:
[0067] Establish a local coordinate system with the center of mass of the mobile robot as the origin. The longitudinal axis of the local coordinate system is consistent with the driving direction corresponding to the planned driving route, and the lateral axis is perpendicular to the longitudinal axis.
[0068] The left and right endpoints of the obstacle in the horizontal axis direction are determined based on the contour point set of the obstacle;
[0069] The first lateral relative distance is determined based on the coordinate difference between the left endpoint and the centroid in the lateral axis direction, and the second lateral relative distance is determined based on the coordinate difference between the right endpoint and the centroid in the lateral axis direction.
[0070] In this embodiment, determining the translation direction includes:
[0071] If passage is only possible from the left, determine to shift to the left; if passage is only possible from the right, determine to shift to the right.
[0072] When passage is possible in both left and right directions: when the difference between the first lateral relative distance and the second lateral relative distance is greater than a preset difference threshold, it is determined to move to the side with the smaller lateral relative distance; when the difference between the first lateral relative distance and the second lateral relative distance is not greater than the preset difference threshold, it is determined to move to the side with the larger possible translation distance.
[0073] In this embodiment, controlling the mobile robot to perform lateral translation along the translation direction includes: determining an initial translation speed, wherein the initial translation speed does not exceed a preset proportion of a preset maximum lateral translation speed.
[0074] In this embodiment, the process of performing lateral translation also includes:
[0075] The attitude angle of the mobile robot is adjusted based on the output of the inertial measurement unit so that the orientation of the mobile robot body and the boundary of the drivable area meet the preset parallelism constraint.
[0076] The lateral distance between the mobile robot and the boundary of the drivable area is controlled to be no less than a preset boundary safety distance threshold, wherein the preset boundary safety distance threshold is related to the vehicle width;
[0077] When a new obstacle is detected entering the translation path and the distance between the mobile robot and the new obstacle is less than the translation mode safety threshold, the mobile robot is controlled to decelerate or stop, and the translation space parameters and the lateral distribution parameters are re-determined to update the translation direction.
[0078] This embodiment proposes a method to navigate around a robotic dog when an obstacle suddenly appears in front of it, obstructing the planned route and causing the distance between the robot and the obstacle to fall short of the pre-planned distance. The solution involves controlling the robot to move around the obstacle by translation. After the robot is forced to stop, this method calculates the translation direction based on factors such as the distance between the robot and the drivable area boundary provided during SLAM mapping, and the lateral relative distance between the robot and the obstacle. This eliminates the problem of existing local path planning methods failing to successfully navigate obstacles due to limitations.
[0079] In one embodiment, a local path planning obstacle avoidance algorithm based on drivable areas is provided, such as... Figure 2As shown, this method aims to address the problem of traditional local path planning causing the robot dog to stop when a sudden obstacle obstructs its planned route and the safety distance is insufficient. This solution, after the robot dog is forced to stop, comprehensively determines the direction of translational obstacle avoidance by combining key factors such as the drivable area boundary distance generated by SLAM mapping and the lateral relative distance between the robot dog and the obstacle. It then controls the robot dog to complete the translational obstacle avoidance and return to the original locally planned path after completely passing the obstacle, ensuring the continuity and safety of the robot dog's movement. To achieve the above objectives, the implementation method includes the following steps:
[0080] Step S100: Real-time acquisition of robot dog's motion status and environmental information: The robot dog collects its own motion status data and surrounding environment data in real time through its onboard LiDAR, inertial measurement unit (IMU), and vision sensors. The motion status data includes the robot dog's current position, x / y linear velocity, angular velocity, heading angle, and attitude angle; the environmental data includes the position, size, and outline of obstacles in front. Simultaneously, it acquires a global drivable area map output by the SLAM mapping module (the map includes drivable area boundary coordinates and non-drivable area markers), and ensures spatiotemporal consistency between sensor data and map data through a time synchronization mechanism (such as the PTP protocol).
[0081] Step S200, sudden obstacle trigger condition judgment: When the robot dog is driving normally under the drive of traditional local path planning (such as dynamic window method, artificial potential field method), it calculates the shortest straight distance between itself and the obstacle in front in real time, and compares it with the preset safe distance threshold (determined according to the robot dog's movement speed and braking performance, the higher the speed, the larger the threshold); at the same time, it judges whether the obstacle completely blocks the current planned driving route. When the conditions of "obstacle blocking the planned route" and "actual shortest distance < preset safe distance threshold" are met, the traditional local path planning system cannot generate an effective obstacle avoidance trajectory, the robot dog triggers emergency braking and stops driving, and enters the translation obstacle avoidance mode trigger process.
[0082] Step S300, Step 1: Calculation of the drivable area boundary distance: Based on the drivable area map generated by SLAM mapping, combined with the robot dog's real-time positioning coordinates, the shortest distance from both sides of the robot dog's body (left edge and right edge) to the corresponding side boundary of the drivable area is calculated using the Euclidean distance calculation formula, denoted as the left drivable translation space distance L1 and the right drivable translation space distance L2. Simultaneously, it is determined whether the drivable spaces on both sides meet the robot dog's translation passage conditions: that is, L1 must be greater than or equal to 1.2 times the robot dog's body width (for attitude adjustment redundancy), and L2 must be greater than or equal to 1.2 times the robot dog's body width. If the distance on one side does not meet the requirements, that side is directly excluded as a translation direction. The drivable area is as follows: Figure 3 As shown.
[0083] Step 2: Calculation of the lateral relative distance between the robot dog and the obstacle: Using the current driving direction of the robot dog as the longitudinal reference, establish a local coordinate system (the robot dog's center of mass is the origin, the longitudinal direction is the X-axis, and the lateral direction is the Y-axis). Using the obstacle contour coordinates collected by the sensors, calculate the difference between the coordinates of the left and right endpoints of the obstacle in the Y-axis direction and the coordinates of the robot dog's center of mass in the Y-axis direction. This will give us the lateral relative distance D1 between the robot dog and the left side of the obstacle and the lateral relative distance D2 between the robot dog and the right side of the obstacle, thus clarifying the lateral position distribution of the obstacle relative to the robot dog.
[0084] Step 3: Comprehensive Decision-Making for Translation and Obstacle Avoidance: Combining the distances of the two sides that can be translated in Step 1 with the relative lateral distances obtained in Step 2, a translation direction decision model is constructed. The decision rules are as follows: 1) If only one side of the translation space meets the passage conditions, then that side is directly determined as the translation direction; 2) If both sides of the translation space meet the passage conditions, then the side with the smaller relative lateral distance is preferentially selected as the translation direction (shortening the translation distance and improving obstacle avoidance efficiency); if the relative lateral distances of the two sides are close (the difference is less than a preset threshold, such as 5cm), then the side with the larger translation space distance is selected to ensure safety redundancy during the translation process. After the decision is completed, the translation direction command (left translation / right translation) and the initial translation speed are output (the initial speed does not exceed 50% of the robot dog's maximum lateral translation speed to avoid attitude imbalance).
[0085] Step S400, Translation and Obstacle Avoidance Process Control and Dynamic Obstacle Monitoring: Based on the determined translation direction and initial speed, the robot dog initiates lateral translation movement. During the translation, the attitude angle is adjusted in real time to ensure that the vehicle body is parallel to the boundary of the drivable area and maintains a constant safe distance (not less than 0.2 times the width of the vehicle body) from both sides. At the same time, the surrounding environment is continuously monitored through sensors. If a new obstacle is detected entering the translation path range (the distance between the new obstacle and the robot dog is less than the safety threshold in the translation mode), the robot dog immediately decelerates or stops the translation operation and returns to step S300 to reassess the drivable space and obstacle avoidance direction; if no new obstacle is detected, the translation speed is maintained until the obstacle avoidance completion condition is met.
[0086] Step S500, Obstacle Avoidance Completion Judgment and Path Regression: Real-time judgment is made as to whether the robot dog has completely passed the obstacle. The judgment conditions are: the robot dog's center of mass coordinates on the original travel direction (X-axis) exceeds the X-axis coordinates of the obstacle's endpoint furthest from the robot dog, and the robot dog's body is completely out of the obstacle's obstruction range. When the obstacle avoidance completion conditions are met, the robot dog stops lateral translation and adjusts its heading angle to return to the original planned path's heading direction. Subsequently, the traditional local path planning module is reactivated, using the robot dog's current position as the path planning starting point, connecting it to a local segment of the original global path, generating the subsequent travel trajectory, and the robot dog resumes normal straight-line travel.
[0087] Step S600, Posture and speed calibration after path regression: After the robot dog reverts to the original local planned path, it monitors the vehicle posture in real time through the IMU sensor. If there is a posture deviation (posture angle deviation is greater than a preset threshold, such as 3°), posture calibration is performed by adjusting the leg joint movement. At the same time, according to the speed curve of the original planned path, the driving speed is gradually increased to the target speed to ensure a smooth transition from translation state to normal driving state and avoid body shaking caused by sudden speed changes.
[0088] This embodiment effectively solves the problem of being forced to stop in scenarios with sudden close-range obstacles in traditional local path planning. It achieves efficient obstacle avoidance by intelligently deciding the lateral movement direction, ensuring the continuity of the robot dog's movement and avoiding risks such as posture imbalance caused by stagnation. Based on the accurate decision of the relative distance between the drivable area and obstacles using SLAM, it improves the robustness of obstacle avoidance in complex environments. After avoiding obstacles, it can quickly and smoothly return to the original planned path, taking into account both driving safety and path tracking accuracy.
[0089] In one embodiment, a local path planning obstacle avoidance device based on a drivable area is provided, comprising:
[0090] The acquisition module is used to acquire pose information, environmental perception information, and a map of the drivable area including the boundaries of the drivable area.
[0091] The triggering determination module is used to determine whether the obstacle avoidance triggering conditions are met based on the environmental perception information during the local path planning process, and to trigger the entry into the translational obstacle avoidance mode when the obstacle avoidance triggering conditions are met.
[0092] The parameter calculation module is used to determine the translatable space parameters in the left and right directions based on the boundary of the drivable area in the translation obstacle avoidance mode, and to determine the lateral distribution parameters of the obstacles relative to the mobile robot based on the environmental perception information.
[0093] The decision module is used to determine the translation direction based on the translational spatial parameters and the lateral distribution parameters;
[0094] The execution and monitoring module is used to control the mobile robot to perform lateral translation along the translation direction, and continuously monitor the changes of obstacles on the translation path to update the translation direction;
[0095] The regression module is used to stop lateral translation and resume local path planning to generate subsequent driving trajectories when the obstacle avoidance conditions are met. The specific implementation details of each module can be found in the above description of the limitations of the obstacle avoidance method based on local path planning within drivable areas, and will not be repeated here.
[0096] 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.
Claims
1. A local path planning obstacle avoidance method based on drivable areas, characterized in that, The method includes: The pose information, environmental perception information, and drivable area map of the mobile robot are acquired, wherein the drivable area map includes the drivable area boundary. During the process of controlling the mobile robot to move along the planned route in local path planning, it is determined whether the obstacle avoidance triggering conditions are met based on the environmental perception information. When the obstacle avoidance triggering condition is met, the mobile robot is controlled to enter the translational obstacle avoidance mode; In the translational obstacle avoidance mode, the translational space parameters of the mobile robot in the left and right directions are determined based on the boundary of the drivable area, and the lateral distribution parameters of the obstacles relative to the mobile robot are determined based on the environmental perception information. The translation direction is determined based on the translational space parameters and the lateral distribution parameters, and the mobile robot is controlled to perform lateral translation along the translation direction. During the lateral translation process, the changes in obstacles along the translation path are continuously monitored and the translation direction is updated based on the monitoring results; When the obstacle avoidance condition is met, the lateral translation stops and the local path planning is resumed to generate a subsequent driving trajectory. The mobile robot is then controlled to move along the subsequent driving trajectory. The driving direction corresponding to the planned driving route is the tangential direction of the planned driving route determined by the local path planning at the current position of the mobile robot.
2. The method according to claim 1, characterized in that, The pose information includes at least one of position, linear velocity, angular velocity, heading angle, and attitude angle; the environmental perception information includes at least one of the position, size, and contour point set of obstacles; and the pose information, the environmental perception information, and the drivable area map are synchronized in time.
3. The method according to claim 1, characterized in that, The obstacle avoidance triggering conditions include: an obstacle obstructing the planned driving route, and the shortest straight-line distance between the mobile robot and the obstacle being less than a preset safe distance threshold.
4. The method according to claim 3, characterized in that, The preset safe distance threshold is related to the current speed of the mobile robot, and the higher the current speed, the higher the preset safe distance threshold.
5. The method according to claim 1, characterized in that, The translatable spatial parameters include a first translatable distance and a second translatable distance, and determining the first translatable distance and the second translatable distance includes: Based on the real-time positioning of the mobile robot, the shortest distance from the left outer contour of the mobile robot to the boundary of the left drivable area is calculated as the first translational distance. The shortest distance from the right outer contour of the mobile robot to the right drivable area boundary is calculated as the second translational distance. Based on the comparison results of the first and second translatable distances with the first passage threshold, the passability in the left and right directions is determined, wherein the first passage threshold is related to the width of the mobile robot.
6. The method according to claim 1, characterized in that, The lateral distribution parameters include a first lateral relative distance and a second lateral relative distance, and determining the first lateral relative distance and the second lateral relative distance includes: Establish a local coordinate system with the center of mass of the mobile robot as the origin. The longitudinal axis of the local coordinate system is consistent with the driving direction corresponding to the planned driving route, and the lateral axis is perpendicular to the longitudinal axis. The left and right endpoints of the obstacle in the horizontal axis direction are determined based on the contour point set of the obstacle; The first lateral relative distance is determined based on the coordinate difference between the left endpoint and the centroid in the lateral axis direction, and the second lateral relative distance is determined based on the coordinate difference between the right endpoint and the centroid in the lateral axis direction.
7. The method according to claim 6, characterized in that, Determining the direction of translation includes: If passage is only possible from the left, determine to shift to the left; if passage is only possible from the right, determine to shift to the right. When passage is possible in both left and right directions: when the difference between the first lateral relative distance and the second lateral relative distance is greater than a preset difference threshold, it is determined to move to the side with the smaller lateral relative distance; when the difference between the first lateral relative distance and the second lateral relative distance is not greater than the preset difference threshold, it is determined to move to the side with the larger possible translation distance.
8. The method according to claim 1, characterized in that, Controlling the mobile robot to perform lateral translation along the translation direction includes: determining an initial translation speed, wherein the initial translation speed does not exceed a preset proportion of a preset maximum lateral translation speed.
9. The method according to claim 1, characterized in that, The process of performing a lateral translation also includes: The attitude angle of the mobile robot is adjusted based on the output of the inertial measurement unit so that the orientation of the mobile robot body and the boundary of the drivable area meet the preset parallelism constraint. The lateral distance between the mobile robot and the boundary of the drivable area is controlled to be no less than a preset boundary safety distance threshold, wherein the preset boundary safety distance threshold is related to the vehicle width; When a new obstacle is detected entering the translation path and the distance between the mobile robot and the new obstacle is less than the translation mode safety threshold, the mobile robot is controlled to decelerate or stop, and the translation space parameters and the lateral distribution parameters are re-determined to update the translation direction.
10. A local path planning obstacle avoidance device based on a drivable area, characterized in that, include: The acquisition module is used to acquire pose information, environmental perception information, and a map of the drivable area including the boundaries of the drivable area. The triggering determination module is used to determine whether the obstacle avoidance triggering conditions are met based on the environmental perception information during the local path planning process, and to trigger the entry into the translational obstacle avoidance mode when the obstacle avoidance triggering conditions are met. The parameter calculation module is used to determine the translatable space parameters in the left and right directions based on the boundary of the drivable area in the translation obstacle avoidance mode, and to determine the lateral distribution parameters of the obstacles relative to the mobile robot based on the environmental perception information. The decision module is used to determine the translation direction based on the translational spatial parameters and the lateral distribution parameters; The execution and monitoring module is used to control the mobile robot to perform lateral translation along the translation direction, and continuously monitor the changes of obstacles on the translation path to update the translation direction; The regression module is used to stop lateral translation and resume local path planning to generate subsequent driving trajectories when the obstacle avoidance conditions are met.