Autonomous following method and system of leg-foot robot

Abstract

The invention belongs to the technical field of robots. According to the autonomous following method and system of the leg-foot robot, the leg-foot robot provided with a laser radar communicates with a to-be-followed mobile terminal, the leg-foot robot and the mobile terminal carried by patrol personnel are each provided with a UWB module, a target point of the mobile terminal is converted into a robot coordinate system, and then the robot coordinate system is started; the method comprises the following steps: carrying out downsampling on laser radar three-dimensional point cloud voxels, screening near-field points, filtering heights, projecting into a two-dimensional obstacle boundary point set, and expanding obstacle points; and taking a minimum value by combining the lateral safety threshold value and the same-direction safety threshold value to obtain an effective safety distance, judging the reachability of a target point, and selecting an optimal middle point from an obstacle gap to generate a track when the target point is safe. According to the method, the autonomous following stability and safety of the leg-foot robot in a complex environment are improved, the calculation load is reduced, and the path planning rationality is enhanced.

Description

Technical Field

[0001] This invention relates to the field of robotics, specifically to a method and system for autonomous following of a legged robot. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In patrol scenarios, legged robots can remotely collect image or video information about equipment and people moving within the patrol area. However, the current operation mode, which relies on manual remote control, requires a patrol person, exacerbating the already strained patrol resources. In scenarios with a large number of people moving around, autonomous following can be used to reduce the burden on patrol resources.

[0004] However, existing autonomous following methods (such as visual image-based tracking) have obvious limitations. When pedestrians frequently pass between patrol personnel and legged robots, they can easily cause temporary obstruction of visual information, leading to the loss of the tracking target and affecting the legged robot's continuous and stable following. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for autonomous following of legged robots. It designs a multi-stage local key point path planning method guided by a two-dimensional obstacle boundary point set, which does not require a global map, has strong real-time performance, achieves multi-level obstacle avoidance, and improves the accuracy and safety of legged robot following.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for autonomous following of a legged robot.

[0007] A method for autonomous following by a legged robot, wherein the legged robot is equipped with a lidar sensor, and the control terminal of the legged robot is communicatively connected to the mobile terminal to be followed, includes the following process: Acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; Acquire and preprocess the 3D point cloud data from the lidar sensor; The preprocessed 3D point cloud data is represented as a 2D obstacle boundary point set. Each obstacle point in the 2D obstacle boundary point set is dilated to obtain dilated obstacle points. Determine the effective safe distance of the inflated obstacle points in the target direction. If the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, the obstacle corresponding to that inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable. If all inflated obstacle points are greater than or equal to the effective safe distance, the target point is considered to be safely reachable. When the target point is considered to be safely reachable, a gap with a distance between adjacent points greater than a set threshold is found in the set of two-dimensional obstacle boundary points. The optimal midpoint is obtained in the gap, and the following motion trajectory of the legged robot is generated based on the robot's starting point, midpoint, and target point.

[0008] In one implementation of the first aspect of the present invention, the effective safe distance of the obstacle point in the target direction is: ,in, This represents the safe distance threshold that ensures the robot maintains a safe lateral distance. This represents the safe distance threshold that ensures the obstacle point is located behind the target point. Representing the There are several obstacle points.

[0009] In one implementation of the first aspect of the present invention, acquiring and preprocessing the three-dimensional point cloud data from a lidar sensor includes: The point cloud space composed of 3D point cloud data is divided into multiple voxels. Each voxel retains only one point to achieve downsampling, and only points within a set range of the center of the legged robot are retained.

[0010] In one implementation of the first aspect of the present invention, the preprocessed three-dimensional point cloud data is represented as a two-dimensional obstacle boundary point set, including: Points with heights greater than the first height threshold and those with heights less than the second height threshold are filtered out. The filtered point cloud is then projected onto a two-dimensional plane. Only the point closest to the center of the legged robot is retained in each angular direction, resulting in an obstacle outline centered on the robot. The points on the obstacle outline constitute a two-dimensional obstacle boundary point set.

[0011] In one implementation of the first aspect of the present invention, each obstacle point in the two-dimensional obstacle boundary point set is expanded to obtain expanded obstacle points, including: For each obstacle point, expand it in multiple directions to generate a batch of offset points as expanded obstacle points.

[0012] In one implementation of the first aspect of the present invention, obtaining the optimal intermediate point in the gap includes: obtaining multiple candidate points in the gap, and selecting the candidate point with the lowest cost as the optimal intermediate point according to the cost function. The cost function is: ; in, For the sake of value, The target distance weights are indicated by the superscript. Represents distance, subscript Represents weight, Let be the Euclidean distance from the candidate point to the target point. It is a candidate point. It is the target point. The weight is the distance from the origin. Let be the distance from the candidate point to the current position of the legged robot. The target angle deviation weight is indicated by the superscript. Representative perspective, The angle difference between the candidate point and the target point. The origin angle weight, The angle from the candidate point to the current position of the legged robot.

[0013] Secondly, the present invention provides an autonomous following system for legged robots.

[0014] An autonomous following system for a legged robot, wherein the legged robot is equipped with a lidar sensor, and the control terminal of the legged robot is communicatively connected to the mobile terminal to be followed, comprising: The target acquisition unit is configured to: acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; The point cloud preprocessing unit is configured to acquire and preprocess the 3D point cloud data from the lidar sensor. The obstacle point inflation unit is configured to: represent the preprocessed 3D point cloud data as a 2D obstacle boundary point set, and inflate each obstacle point in the 2D obstacle boundary point set to obtain an inflated obstacle point; The safety judgment unit is configured to: determine the effective safe distance of the inflated obstacle points in the target direction; if the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, then the obstacle corresponding to the inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable; if all inflated obstacle points are greater than or equal to the effective safe distance, then the target point is considered to be safely reachable. The trajectory generation unit is configured to: when the target point is considered to be safely reachable, find gaps in the set of two-dimensional obstacle boundary points where the distance between adjacent points is greater than a set threshold, obtain the optimal intermediate point in the gap, and generate the following motion trajectory of the legged robot based on the robot's starting point, intermediate point and target point.

[0015] Thirdly, the present invention provides an autonomous following system for legged robots.

[0016] A legged robot autonomous following system includes a legged robot and a wristband for workers to wear. Both the wristband and the legged robot are equipped with UWB modules. The UWB module of the legged robot can receive the target point sent by the UWB module of the wristband. The control terminal of the legged robot is communicatively connected to the UWB module of the legged robot, and the control terminal is configured to execute the autonomous following method of the legged robot according to the first aspect of the present invention.

[0017] Fourthly, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the autonomous following method for a legged robot according to the first aspect of the present invention.

[0018] Fifthly, the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and to execute the legged robot autonomous following method of the first aspect of the present invention.

[0019] Compared with the prior art, the beneficial effects of the present invention are: This invention ensures the accuracy of target positioning by converting the target point of the mobile terminal to the robot coordinate system, providing clear directional guidance for following. After preprocessing the 3D point cloud data of the LiDAR, it is transformed into a 2D obstacle boundary point set based on polar coordinates, simplifying the processing complexity of environmental data. At the same time, obstacle point inflation is equivalent to building a safe buffer zone for obstacles, effectively avoiding collision risks. By judging the effective safe distance, it can accurately identify whether the target point is reachable, preventing the robot from entering dangerous areas. When the target point is safely reachable, the optimal intermediate point is selected from the gaps between obstacles, and the trajectory is generated by combining the starting point, intermediate point, and target point, which ensures that the path can bypass obstacles and ensures the rationality of the trajectory. The whole process makes full use of the environmental perception capability of LiDAR. Through a series of data processing and path planning steps, it overcomes the problems of susceptibility to occlusion and weak obstacle avoidance capability in traditional following methods, significantly improving the stability and safety of autonomous following of legged robots, allowing the robot to continuously and stably follow the target in complex environments.

[0020] This invention sets two safety distance thresholds for different positional relationships between obstacles and target points, and combines them to form an effective safety distance. When the angle between the obstacle and the target direction is large, the focus is on ensuring lateral safety distance; when the obstacle and the target point are close to the same direction, the emphasis is on ensuring that the obstacle is behind the target point. This differentiated safety distance setting makes safety judgments more aligned with actual scenario needs, avoiding safety loopholes that may occur with a single safety threshold under different obstacle layouts. By accurately defining the effective safety distance, the robot can more accurately determine whether the target point is in a safe and reachable state, effectively avoiding collision risks under different positional relationships, making safety protection during the following process more comprehensive and reliable, further improving the practicality and safety of the autonomous following method, and ensuring the robot moves stably in complex obstacle distribution environments.

[0021] This invention utilizes voxel downsampling to divide the point cloud space into multiple voxels, retaining only one point per voxel. This significantly reduces redundant information in the 3D point cloud data, lowers the computational burden on subsequent data processing and path planning, and improves the overall efficiency of the method. Simultaneously, by retaining only points within a predetermined range around the robot's center and eliminating irrelevant data far from the robot, it avoids interference from irrelevant information in environmental judgment, allowing the robot to focus on key environmental areas around itself. This targeted preprocessing method achieves efficient data simplification while ensuring accurate environmental perception. It provides high-quality, timely data support for subsequent steps such as constructing 2D obstacle boundary point sets and determining safe distances, ensuring the entire following method can quickly respond to environmental changes and improve real-time following capabilities.

[0022] This invention filters out points exceeding a set height range, eliminating obstacle information irrelevant to robot movement and focusing on key obstacles affecting robot movement. This simplifies the scope of data processing. Projecting the filtered point cloud onto a two-dimensional plane further reduces data processing complexity and facilitates subsequent obstacle contour analysis. Only the point closest to the robot's center is retained in each angle direction, accurately outlining the core contours of obstacles. This allows the robot to clearly understand the actual distribution of surrounding obstacles. This processing method ensures the accuracy of obstacle information while simplifying the data processing flow, providing reliable obstacle foundation data for subsequent steps such as obstacle expansion and gap search. It ensures the robot can accurately identify passable areas, improves the scientific nature of obstacle avoidance decisions, and enhances its adaptability to complex obstacles during following.

[0023] This invention expands each obstacle point in multiple directions to generate offset points, creating a comprehensive safety buffer zone for the obstacle. This multi-directional expansion method fully considers the robot's size and movement requirements, avoiding collision risks caused by irregular obstacle edges or deviations in the robot's movement trajectory. The expanded obstacle points better meet actual safety protection needs, allowing the robot to maintain sufficient safety clearance from obstacles during path planning. Even in environments with dense obstacle distribution, the expanded obstacle points can accurately determine passable space. This approach effectively improves the comprehensiveness and safety of obstacle judgment, ensuring that the robot can flexibly avoid various obstacles during following, enhancing the reliability and safety of autonomous following.

[0024] This invention uses a cost function to select candidate points with the lowest cost from gaps as intermediate points. The cost function comprehensively considers factors such as target distance, self-position, and angle deviation, ensuring that the selected intermediate points are both close to the target direction and maintain the robot's motion rationality. The intermediate points are extrapolated outward along the boundary normal vector or away from obstacles to a safe radius, further ensuring that the intermediate points are in unobstructed free space. This avoids safety hazards caused by intermediate points being too close to obstacles. This method of obtaining intermediate points results in a smooth and continuous following trajectory that effectively avoids obstacles, balancing following efficiency and safety. Through scientific candidate point selection and intermediate point optimization, the robot's path is more in line with motion laws, avoiding sudden path changes or dangerous sections, significantly improving the stability and reliability of autonomous following motion, and ensuring that the robot can follow the target efficiently and safely.

[0025] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0026] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0027] Figure 1 A schematic diagram of the principle of an autonomous following system for a legged robot provided as an exemplary embodiment of the present invention; Figure 2 A flowchart illustrating an exemplary embodiment of the present invention provides a method for autonomous following of a legged robot. Figure 3 A flowchart illustrating a local path planning method based on obstacle boundary search, provided as an exemplary embodiment of the present invention; Figure 4A schematic diagram of the principle of an autonomous following system for a legged robot provided as another exemplary embodiment of the present invention; Figure 5 A schematic diagram of a computer device provided for another exemplary embodiment of the present invention; Among them, 1. Legged robots; 2. Pedestrians; 3. Patrol personnel. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0030] To address the issue of insufficient stability in human-following operations for legged robots during patrol scenarios, this invention proposes an autonomous following system for legged robots. By fusing multi-source information from UWB and LiDAR, it possesses strong anti-interference capabilities in dynamic environments, maintaining stable and continuous tracking even under conditions of frequent pedestrian traffic. This invention integrates the core control logic into a UWB wristband, allowing patrol personnel to switch the legged robot's operating mode (such as standby, follow, or mode switching) with a single button on the wristband. This eliminates the need for carrying a remote control and complex button operations, enabling mode switching to be completed with one hand, shortening response time for unexpected tasks, and thus improving patrol efficiency.

[0031] Specifically, such as Figure 1 As shown, two UWB modules are installed on the back of the legged robot 1, one in front and one behind. At the same time, the patrol personnel 3 wears a UWB signal transmitting wristband. During the autonomous following process, the legged robot 1 perceives and acquires the surrounding environment in real time through LiDAR. The position of the UWB signal emitted by the wristband worn by the patrol personnel 3 is used as the target point and input into a multi-stage local key point path planning and obstacle avoidance algorithm guided by a polar coordinate obstacle point set. The system collects LiDAR point cloud data in real time, perceives dynamic obstacles in the surrounding environment (such as pedestrians 2 walking through), and updates the path dynamically accordingly to ensure that the legged robot can safely and flexibly avoid obstacles and move forward.

[0032] Specific follow control logic, such as Figure 2 As shown, the process includes the following: S201: Patrol personnel wear wristbands and activate them to control the legged robot to switch modes. Simultaneously, the wristbands serve as a target signal for the legged robot to follow. Specific mode switching may include: Standby mode: The robot is stationary, the follow algorithm is turned off, and only the sensors are kept running at low power (the LiDAR sampling frequency is reduced to 5Hz). Follow mode: Activates the autonomous following algorithm, follows the target according to the planned trajectory, with a maximum speed of 0.8 m / s; Manual fine-tuning mode: Temporarily pauses automatic following, and controls the robot to move forward / backward / turn via the wristband buttons (step length 0.1m, turning angle 5° / time). Complex terrain gait mode: For terrains such as stairs and steep slopes, switch to an adaptive gait (shorter stride and slower speed).

[0033] One-handed switching can be achieved in the following way: The wristband's hardware design features three physical buttons arranged vertically on the side (8mm in diameter, 15mm apart) for single-finger operation. The button logic is as follows: Short press button 1: Standby mode → Follow mode (switch response time ≤ 0.1s); Press and hold button 2 (≥3 seconds): Follow mode → Manual fine-tuning mode (it will automatically return to follow mode after releasing); Double-click key 3: Trigger complex terrain gait mode (lasts 10 seconds, automatically exits if no operation is performed); Feedback mechanism: After a successful switch, the wristband vibrates once (intensity 0.5g), and the corresponding LED light illuminates (standby - red, follow - green, manual - yellow, complex terrain - blue).

[0034] S202: Following the algorithm, the LiDAR acquires surrounding perception information in real time, the wristband emits a target signal, and after the UWB module receives the signal, it calculates the position of the wristband relative to the UWB module. The position is expressed in the local coordinate system of the legged robot. The target signal is input into the local path planning method based on obstacle boundary search. Based on the surrounding perception information, a passable path is planned in real time, and the legged robot moves along the current path. S203: Activate the lidar inertial odometry to provide the legged robot's real-time position in the world coordinate system. Provide the world coordinate system position to the monitoring backend for monitoring the legged robot's real-time position. S204: The legged robot will move along the planned path, with its maximum speed limited to within 0.8m / s. At the same time, in order to avoid collisions with the target personnel, the system sets a minimum following distance of 0.5m to ensure that the legged robot always maintains a safe distance from the personnel.

[0035] This invention integrates multiple sensors, such as LiDAR and UWB, through this modular integrated design. By using a UWB wristband worn on the wrist of patrol personnel, it provides a stable target following signal for the legged robot, overcoming the problem of visual information loss caused by factors such as pedestrian occlusion in pure vision tracking methods. This significantly improves the stability and robustness of target following. The UWB wristband continuously provides the location of the target personnel, effectively preventing the tracking target from being lost.

[0036] To achieve better following for legged robots, this invention also proposes an autonomous following method for legged robots, namely a local path planning method based on obstacle boundary search, such as... Figure 3 As shown, the process includes the following: S301: Receives the target signal emitted by the patrol personnel's wristband and converts the target point represented by the target signal into the coordinate system of the legged robot.

[0037] S302: Subscribe to the point cloud information of the LiDAR and preprocess the point cloud (remove invalid points and filter).

[0038] Noise and out-of-range points are removed from the original point cloud; the point cloud space is divided into small grids (voxels), and only one point is retained in each grid to achieve downsampling; only points in a certain area centered on the legged robot are retained to reduce irrelevant data; the LiDAR operates at a frequency of 10Hz, and the connection detection and path replanning process is also periodically executed at a frequency of 10Hz to ensure that the local motion planning is updated in real time and to guarantee the real-time performance of the planning.

[0039] S303: Represent a 3D point cloud as a set of 2D obstacle boundary points.

[0040] Filter out points that are too high or too low, project the point cloud onto a 2D plane, and retain only the nearest point in each angle direction to obtain an obstacle outline centered on the robot: (1) (2); In the formula, ( , ) represents the point cloud coordinates of the obstacle. , ) is the first The polar coordinates of the obstacle points.

[0041] S304: Dilate the obstacle point cloud.

[0042] Legged robots have a certain size. If the path is planned solely based on the original obstacle boundaries, it might be too close to the target, increasing the risk of collision. Expanding the obstacles essentially adds a "safety buffer zone," ensuring the robot always has sufficient clearance. The expansion effect is achieved by generating a batch of offset points (up, down, left, right, four diagonals, eight directions) for each obstacle point, as detailed below: (3); In the formula, This is the point cloud offset. This is the set of offset vectors used when expanding obstacle points. Specifically, δ∈[0.2m, 0.8m], with a default value of 0.4m. The robot's own size (maximum circumscribed circle radius R) + safety redundancy (ΔR) means that δ≥R+ΔR. For example, if the robot's body width is 0.6m (R=0.3m) and the safety redundancy ΔR=0.1m, then the default value is δ=0.4m.

[0043] S305: Safety inspection.

[0044] The system determines whether the target point is obstructed by an obstacle. If the path is not direct, it enters detour mode. By calculating the orientation angle of the target point in the robot coordinate system, the angle difference between the target point and the obstacle point is obtained. Combined with the robot's safe radius and the distance to the target point, the safe boundary of the obstacle at different angles is derived.

[0045] Specifically, when the angle between the obstacle and the target direction is large, the robot needs to maintain a safe lateral distance, at which point the safety threshold is required. for: (4); in, This represents the difference in direction angle between the obstacle point and the target point. This represents the safe radius of the legged robot itself. Specifically, the safe radius of the legged robot can be obtained in the following way: (5); in, This represents the maximum circumscribed circle radius of the robot's body (determined by mechanical design parameters, such as 0.3m). This represents the motion error compensation amount (considering robot gait error and trajectory tracking error, with a value of 0.1m). Represents a fixed safety redundancy; r∈[0.3m, 0.8m], dynamically adjusted according to robot load and speed (r increases by 0.1m when speed ≥ 0.6m / s; r increases by 0.05m when load ≥ 5kg); when the obstacle point and the target point are close to the same direction, it is necessary to ensure that the obstacle is behind the target point, at which point the safety threshold is... for: (6) in, The Euclidean distance represents the distance from the target point to the center point of the legged robot.

[0046] Combining the two constraints, we can obtain the first... Effective safe distance of each obstacle point in the target direction : (7); If any obstacle point is smaller than the effective safe distance, the obstacle is considered to be located in the danger zone between the robot and the target point, and the target point is determined to be unreachable directly; otherwise, if all obstacle points do not meet this condition, the target point is considered to be safely reachable.

[0047] S306: Boundary gap search and candidate point generation.

[0048] Find the "gap" with a large distance between adjacent points in the set of obstacle boundary points. If the distance between two points is greater than a certain set threshold, it means that there is a gap between the two obstacles. The two sides of the gap are the source of the candidate point positions.

[0049] S307: The midpoint is calculated.

[0050] The intermediate point is the optimal point selected from the candidate points. Based on the cost function, the point with the minimum cost is chosen as the intermediate point. Then, the robot's safe radius is extrapolated along the boundary normal vector or from the direction away from the obstacle to ensure the intermediate point lies within free space, guaranteeing trajectory safety. The cost function is as follows: (8); In the formula, For the sake of value, The target distance weights are indicated by the superscript. Represents distance, subscript Represents weight, Let be the Euclidean distance from the candidate point to the target point. It is a candidate point. It is the target point. The weight is the distance from the origin. Let be the distance from the candidate point to the current position of the legged robot. The target angle deviation weight is indicated by the superscript. Representative perspective, Candidate points With the target point angular difference, The origin angle weight, Candidate points The angle to the current position of the legged robot.

[0051] The methods for solving the above cost function include: (1) Candidate point generation: uniform sampling in the gaps between obstacles (sampling interval 0.1m) to generate an initial candidate point set; (2) Gradient descent method is used to minimize the cost function, with an iteration step size of 0.05m. The iteration termination condition is "value difference between the two generations ≤ 0.01" or the number of iterations ≤ 10 (to ensure real-time performance); the candidate points obtained by solving are extrapolated along the normal vector of the obstacle boundary r (safe radius) to ensure that the intermediate point is in the unobstructed free space.

[0052] Based on the Analytic Hierarchy Process (AHP), the weights of each element are set by prioritizing "accuracy of following direction" (higher weights for target distance and angle deviation) while also considering "motion efficiency" (lower weights for position and angle). Specifically, the target distance weight... The weight is: 0.4-0.6 (core weight, prioritizing proximity to the target); target angle deviation weight. The value is: 0.3-0.4 (secondary core, ensuring consistent direction); origin distance weight. The value is 0.1-0.2 (auxiliary, to shorten the motion distance); origin angle weight. The value is 0.05-0.1 (assistance, reducing steering angle).

[0053] S308: Trajectory Generation and Optimization.

[0054] After selecting the intermediate point, the robot needs to reach the intermediate point or final target point from the current position along a safe, smooth, and executable path. By connecting the starting point, intermediate point, and target point in sequence according to the above key point sequence, a preliminary trajectory is generated. Cubic spline interpolation is used to smooth and generate the trajectory.

[0055] Experiments have verified that, in a complex scenario with an obstacle density of 5 obstacles / m² and a pedestrian crossing frequency of 10 times / minute, after continuous following for 1 hour, the number of target loss instances is ≤3, the relock time for each instance is ≤0.3s, the overall system delay (from target signal change to trajectory update) is ≤0.1s, meeting the real-time threshold (≤0.2s) for autonomous following of industrial robots, the average following distance error is ≤0.2m, the maximum deviation is ≤0.5m, and no collision accidents occur.

[0056] The method proposed in this invention does not require a global map, has strong real-time performance, can achieve multi-level obstacle avoidance, and generates smooth trajectories. By integrating the core control logic into the UWB wristband, the working mode of the legged robot can be switched with one click through the wristband, eliminating the need to carry a remote control and perform complex key operations. Mode switching can be completed with one hand, shortening the response time for sudden tasks. Combined with the set trajectory generation strategy, patrol efficiency is improved.

[0057] Figure 4 An autonomous following system for a legged robot is shown, comprising: The target acquisition unit 401 is configured to: acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; The point cloud preprocessing unit 402 is configured to acquire and preprocess the three-dimensional point cloud data from the lidar sensor. The obstacle point expansion unit 403 is configured to: represent the preprocessed three-dimensional point cloud data as a two-dimensional obstacle boundary point set, and expand each obstacle point in the two-dimensional obstacle boundary point set to obtain an expanded obstacle point; The safety judgment unit 404 is configured to: determine the effective safe distance of the inflated obstacle point in the target direction; if the distance between the inflated obstacle point and the center of the legged robot is less than the effective safe distance, then the obstacle corresponding to the inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable; if all inflated obstacle points are greater than or equal to the effective safe distance, then the target point is considered to be safely reachable. The trajectory generation unit 405 is configured to: when the target point is considered to be safely reachable, find gaps in the set of two-dimensional obstacle boundary points where the distance between adjacent points is greater than a set threshold, obtain the optimal intermediate point in the gap, and generate the following motion trajectory of the legged robot based on the robot's starting point, intermediate point and target point.

[0058] It is understood that the aforementioned units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of the present invention. The aforementioned units are based on logical functional division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of the present invention, the system may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0059] According to another embodiment of the present invention, the system of this embodiment can be constructed by running a computer program (including program code) capable of performing the steps involved in the corresponding method of the present invention on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the aforementioned computing device through the computer-readable recording medium, and run therein.

[0060] Figure 5 A computer device is shown, which includes a processor 501, a communication interface 502, and a computer-readable storage medium 503. The processor 501, communication interface 502, and computer-readable storage medium 503 can be connected via a bus or other means.

[0061] The communication interface 502 is used to receive and send data. The computer-readable storage medium 503 can be stored in the memory of the electronic device. The computer-readable storage medium 503 is used to store computer programs, which include program instructions. The processor 501 is used to execute the program instructions stored in the computer-readable storage medium 503.

[0062] The processor 501 is the computing and control core of the electronic device. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to achieve the corresponding method flow or corresponding function.

[0063] Processor 501 is configured to perform the following procedure: Acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; Acquire and preprocess the 3D point cloud data from the lidar sensor; The preprocessed 3D point cloud data is represented as a 2D obstacle boundary point set. Each obstacle point in the 2D obstacle boundary point set is dilated to obtain dilated obstacle points. Determine the effective safe distance of the inflated obstacle points in the target direction. If the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, the obstacle corresponding to that inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable. If all inflated obstacle points are greater than or equal to the effective safe distance, the target point is considered to be safely reachable. When the target point is considered to be safely reachable, a gap with a distance between adjacent points greater than a set threshold is found in the set of two-dimensional obstacle boundary points. The optimal midpoint is obtained in the gap, and the following motion trajectory of the legged robot is generated based on the robot's starting point, midpoint, and target point.

[0064] This invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the electronic device and extended storage media supported by the electronic device. The computer-readable storage medium provides storage space for storing the processing system of the electronic device.

[0065] Furthermore, this storage space also contains one or more instructions suitable for loading and execution by the processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory; alternatively, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0066] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to perform the following process: Acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; Acquire and preprocess the 3D point cloud data from the lidar sensor; The preprocessed 3D point cloud data is represented as a 2D obstacle boundary point set. Each obstacle point in the 2D obstacle boundary point set is dilated to obtain dilated obstacle points. Determine the effective safe distance of the inflated obstacle points in the target direction. If the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, the obstacle corresponding to that inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable. If all inflated obstacle points are greater than or equal to the effective safe distance, the target point is considered to be safely reachable. When the target point is considered to be safely reachable, a gap with a distance between adjacent points greater than a set threshold is found in the set of two-dimensional obstacle boundary points. The optimal midpoint is obtained in the gap, and the following motion trajectory of the legged robot is generated based on the robot's starting point, midpoint, and target point.

[0067] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can implement the described functions using different methods for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0068] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic cable, digital cable) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0069] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for autonomous following of a legged robot, characterized in that, The legged robot is equipped with a lidar sensor. The control terminal of the legged robot communicates with the mobile terminal to be followed, including the following process: Acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; Acquire and preprocess the 3D point cloud data from the lidar sensor; The preprocessed 3D point cloud data is represented as a 2D obstacle boundary point set. Each obstacle point in the 2D obstacle boundary point set is dilated to obtain dilated obstacle points. Determine the effective safe distance of the inflated obstacle point in the target direction. If the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, then the obstacle corresponding to the inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable. If all inflated obstacle points are greater than or equal to the effective safe distance, the target point is considered to be safely reachable; When the target point is considered to be safely reachable, a gap with a distance between adjacent points greater than a set threshold is found in the set of two-dimensional obstacle boundary points. The optimal midpoint is obtained in the gap, and the following motion trajectory of the legged robot is generated based on the robot's starting point, midpoint, and target point.

2. The autonomous following method for a legged robot as described in claim 1, characterized in that, The effective safe distance of the obstacle point in the target direction is: ,in, This represents the safe distance threshold that ensures the robot maintains a safe lateral distance. This represents the safe distance threshold that ensures the obstacle point is located behind the target point. Representing the There are several obstacle points.

3. The autonomous following method for a legged robot as described in claim 1, characterized in that, Acquire and preprocess 3D point cloud data from the lidar sensor, including: The point cloud space composed of 3D point cloud data is divided into multiple voxels. Each voxel retains only one point to achieve downsampling, and only points within a set range of the center of the legged robot are retained.

4. The autonomous following method for a legged robot as described in claim 1, characterized in that, The preprocessed 3D point cloud data is represented as a set of 2D obstacle boundary points, including: Points with heights greater than the first height threshold and those with heights less than the second height threshold are filtered out. The filtered point cloud is then projected onto a two-dimensional plane. Only the point closest to the center of the legged robot is retained in each angular direction, resulting in an obstacle outline centered on the robot. The points on the obstacle outline constitute a two-dimensional obstacle boundary point set.

5. The autonomous following method for a legged robot as described in claim 1, characterized in that, Dilatation is performed on each obstacle point in the set of two-dimensional obstacle boundary points to obtain dilatated obstacle points, including: For each obstacle point, expand it in multiple directions to generate a batch of offset points as expanded obstacle points.

6. The autonomous following method for a legged robot as described in claim 1, characterized in that, Finding the optimal intermediate point in the gap includes: obtaining multiple candidate points in the gap, and selecting the candidate point with the lowest cost as the optimal intermediate point according to the cost function; The cost function is: ; in, For the sake of value, The target distance weights are indicated by the superscript. Represents distance, subscript Represents weight, Let be the Euclidean distance from the candidate point to the target point. It is a candidate point. It is the target point. The weight is the distance from the origin. Let be the distance from the candidate point to the current position of the legged robot. The target angle deviation weight is indicated by the superscript. Representative perspective, The angle difference between the candidate point and the target point. The origin angle weight, The angle from the candidate point to the current position of the legged robot.

7. A legged robot autonomous following system, characterized in that, The legged robot is equipped with a lidar sensor, and its control terminal is communicatively connected to the mobile terminal to be followed, including: The target acquisition unit is configured to: acquire the target signal from the mobile terminal and convert the target point represented by the target signal into the coordinate system of the legged robot; The point cloud preprocessing unit is configured to acquire and preprocess the 3D point cloud data from the lidar sensor. The obstacle point inflation unit is configured to: represent the preprocessed 3D point cloud data as a 2D obstacle boundary point set, and inflate each obstacle point in the 2D obstacle boundary point set to obtain an inflated obstacle point; The safety judgment unit is configured to: determine the effective safe distance of the inflated obstacle points in the target direction; if the distance between an inflated obstacle point and the center of the legged robot is less than the effective safe distance, then the obstacle corresponding to the inflated obstacle point is considered to be located in the danger zone between the legged robot and the target point, and thus the target point is determined to be unreachable; if all inflated obstacle points are greater than or equal to the effective safe distance, then the target point is considered to be safely reachable. The trajectory generation unit is configured to: when the target point is considered to be safely reachable, find gaps in the set of two-dimensional obstacle boundary points where the distance between adjacent points is greater than a set threshold, obtain the optimal intermediate point in the gap, and generate the following motion trajectory of the legged robot based on the robot's starting point, intermediate point and target point.

8. A legged robot autonomous following system, characterized in that, The device includes a legged robot and a wristband for workers to wear. Both the wristband and the legged robot are equipped with UWB modules. The UWB module of the legged robot can receive the target point sent by the UWB module of the wristband. The control terminal of the legged robot is communicatively connected to the UWB module of the legged robot, and the control terminal is configured to execute the autonomous following method of the legged robot according to any one of claims 1-6.

9. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program that, when executed by the processor, implements the autonomous following method for a legged robot as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 6 for the autonomous following method of a legged robot.

Images