Deviation correction and obstacle avoidance control method for large-scale multi-crawler walking device

Abstract

The invention relates to the technical field of unmanned walking devices, in particular to a deviation correction and obstacle avoidance control method for a large-scale multi-crawler walking device, and the method comprises the steps: carrying out the real-time detection of an obstacle and the continuous advancing judgment through an intelligent sensing system based on the three-dimensional laser radar point cloud data of an intelligent holder; when the track offset is larger than a set threshold value, multi-crawler walking device kinematics model analysis and multi-crawler walking device deviation correction judgment are carried out through an automatic deviation correction system, and a rapid adjustment strategy and a precise deviation correction control strategy are formulated; and when an obstacle which cannot pass or affects the passing stability exists in the front, a new advancing path is planned through the automatic obstacle avoidance system. According to the invention, remote obstacle intelligent identification and high-precision positioning and orientation of the multi-crawler walking device can be realized, a safe path for avoiding obstacles is planned by applying an automatic obstacle avoidance algorithm, and the multi-crawler walking device can make an obstacle avoidance decision autonomously, so that safe and stable driving of the multi-crawler walking device is ensured.

Description

Technical Field

[0001] This invention relates to the field of unmanned walking device technology, specifically to a method for correcting deviations and avoiding obstacles in a large multi-track walking device. Background Technology

[0002] Multi-track walking devices, as key components of heavy transport equipment, are highly integrated and complex electromechanical systems. They are characterized by their large size and weight, large turning radius, and independent power and transmission systems. Due to the complex beach environment, current operation of these devices often relies on operator experience. However, drivers often lack comprehensive control over the equipment and the overall environment, leading to significant deviations between the walking trajectory and the desired path, failing to achieve the expected results. For example, manual operation requires drivers to exert considerable effort to dynamically adjust the steering track angle or speed to achieve the predetermined trajectory, which is extremely difficult.

[0003] The sand and soft soil on the beach surface are mobile, causing uncertainty and instability in the terrain. When multi-tracked vehicles travel on this terrain, they are affected by the resistance of uneven road surfaces, which may cause their trajectory to deviate from the predetermined path. Therefore, large multi-tracked vehicles must monitor and adjust their body posture online in real time.

[0004] Meanwhile, when the multi-tracked walking device encounters tall, continuous, or long obstacles ahead, it needs to automatically avoid them. To this end, based on lidar scanning and environmental model reconstruction, a path to avoid obstacles is planned. During the movement, the device continuously monitors surrounding obstacles. Once an obstacle is detected, the multi-tracked walking device applies the developed automatic obstacle avoidance algorithm to adjust its actions in real time, promptly changing its direction of travel to avoid tall or continuous obstacles.

[0005] For ultra-long multi-tracked walking devices (100m in length, 600 tons in weight, 12 independent hydraulic drives) that travel in unknown, complex, and variable shallow water environments, a large tracked walking device correction and obstacle avoidance control system and method needs to be designed for application to various vehicles that need to travel and work on uneven tidal flat terrain. Summary of the Invention

[0006] The purpose of this invention is to provide a method for correcting deviations and avoiding obstacles in a large multi-tracked walking device, in order to solve the technical problem in the prior art where the driver's overall control over the equipment and the environment is incomplete, resulting in a large deviation between the walking trajectory of the multi-tracked walking device and the desired trajectory, making it difficult to achieve the predetermined trajectory. The specific technical solution is as follows:

[0007] This invention provides a method for correcting deviations and avoiding obstacles in a large multi-tracked walking device. The method uses an intelligent sensing system to achieve data acquisition, data preprocessing, data fusion, and real-time output of the fused pose information of the multi-tracked walking device based on the fusion positioning of IMU and Beidou satellite data.

[0008] The intelligent sensing system uses 3D LiDAR point cloud data from an intelligent gimbal to perform real-time obstacle detection and determine whether to continue moving.

[0009] When the trajectory deviation exceeds the set threshold, the automatic correction system performs kinematic model analysis of the multi-track walking device, determines the correction of the multi-track walking device, and formulates rapid adjustment strategies and precise correction control strategies.

[0010] When there are obstacles ahead that are impassable or affect the stability of the journey, the automatic obstacle avoidance system plans a new path.

[0011] A further improvement of the obstacle avoidance and deviation control method for large multi-track walking devices in this invention lies in the use of an adaptive unscented Kalman filter algorithm to fuse IMU accelerometer data and BeiDou satellite positioning data during data fusion to obtain predicted device position and velocity information. The specific steps are as follows:

[0012] State prediction:

[0013] x k+1 =f(x) k ,u k ) 1);

[0014] Where, x k+1 It is the predicted value at time k+1; f is the system's state transition function; x k It is the predicted value at time k; u k It is the system control input at time k;

[0015] Covariance prediction:

[0016]

[0017] Among them, P k+1 A is the covariance matrix at time k+1; k P is the state transition matrix at time k; k Q is the covariance matrix at time k; k It is the process noise covariance matrix at time k; T is the update cycle time of BeiDou satellite data;

[0018] Measurement and prediction:

[0019] z k+1 =h(x k+1 ) 3);

[0020] Among them, z k+1 is the predicted measurement value at time k+1; h is the observation function;

[0021] Calculate the residuals:

[0022] z″ k+1 =z′ k+1 -z k+1 4);

[0023] Among them, z' k+1 It is the measurement value at time k+1; z” k+1 It is the residual of the state estimate at time k+1;

[0024] Calculate the residual covariance:

[0025]

[0026] Among them, S k+1 H is the measurement residual covariance matrix at time k+1; k+1 It is the observation matrix at time k+1; P k+1 R is the covariance matrix at time k+1; k+1 It is the observation noise covariance matrix at time k+1;

[0027] Calculate the Kalman gain:

[0028]

[0029] Among them, K k+1 It is the Kalman gain at time k+1;

[0030] Updated state estimate:

[0031] x k+1 =x k+1 +K k+1 z″ k+1 7);

[0032] Update the covariance matrix:

[0033] P k+1 =P k+1 -K k+1 H k+1 P k+1 8);

[0034] Update process noise covariance matrix:

[0035]

[0036] Among them, Q k+1 It is the process noise covariance matrix at time k+1; Q kIt is the process noise covariance matrix at time k; α k It is the adaptive coefficient at time k, used to adjust the adjustment rate of process noise;

[0037] Update the observation noise covariance matrix:

[0038]

[0039] Among them, R k β is the observation noise covariance matrix at time k; k It is the adaptive coefficient at time k, used to adjust the adjustment rate of the observation noise.

[0040] A further improvement of the obstacle avoidance and deviation control method for large multi-tracked walking devices in this invention lies in that, when outputting the fused pose information of the multi-tracked walking device in real time, the initial pose information of the multi-tracked walking device is first measured, and then the gyroscope data of the IMU is denoised and calibrated to obtain the pose information of the device in real time. The specific calculation formula is as follows:

[0041]

[0042] Where, δ xt1 δ is the rotation angle of the device around the X-axis at the initial position. yt1 δ is the rotation angle of the device around the Y-axis at the initial position. zt1 δ is the rotation angle of the device around the Z-axis at the initial position; xt2 δ is the rotation angle of the device around the X-axis at its current position; yt2 δ is the rotation angle of the device around the Y-axis at its current position; zt2 ω is the rotation angle of the device around the Z-axis at its current position; xt1 ω is the angular velocity of the device about the X-axis as measured by the gyroscope at the initial position. yt1 ω is the angular velocity of the device about the Y-axis as measured by the gyroscope at the initial position. zt1 dt is the angular velocity of the device around the Z-axis as measured by the gyroscope at the initial position; dt is the update time of the gyroscope's measurement data.

[0043] A further improvement of the obstacle avoidance and deviation control method for the large multi-tracked walking device of the present invention lies in the following: when the intelligent sensing system performs real-time obstacle detection and determines whether to continue walking based on the three-dimensional lidar point cloud data of the intelligent gimbal, the point cloud data is first collected and transmitted. Then, an improved ground slope separation algorithm is used to separate the ground point cloud from the non-ground point cloud. Next, an improved Euclidean clustering algorithm is used to perform cluster analysis on the non-ground point cloud obstacles. Finally, the obstacle information ahead is compared with the pre-set obstacle threshold that the multi-tracked walking device can pass based on the real-time detection output, and a decision is made on whether to pass. If yes, the multi-tracked walking device is controlled to pass; otherwise, the automatic obstacle avoidance system is triggered to plan a new path.

[0044] A further improvement of the obstacle avoidance and correction control method for the large multi-track walking device of the present invention lies in that, when performing kinematic model analysis of the multi-track walking device through the automatic correction system, the multi-track walking device is an integrated structure, and the left or right track is set to have the same driving speed, which is equivalently simplified to a two-wheel differential drive motion model. The equivalent simplified forward kinematic model is expressed as follows:

[0045]

[0046] The equivalent simplified inverse kinematics model is expressed as:

[0047]

[0048] d LR =βd wb 14);

[0049] Among them, w c v is the angular velocity of the center of mass. c v is the linear velocity of the center of mass; β is the equivalent coefficient; l The linear velocity of the left drive wheel is derived from the velocity decomposition of the center of mass; v r To determine the linear velocity of the right drive wheel based on the velocity decomposition of the center of mass; d wb d is the wheel spacing; LR This is the equivalent wheel spacing.

[0050] A further improvement of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention is that, during the deviation correction judgment process of the multi-track walking device, when the height of the obstacle passing through the road surface does not exceed the preset ground unevenness threshold, the driving device continues to move and passes through the obstacle without making any deviation correction adjustments during the passage process.

[0051] After the multi-track walking device passes, it makes a correction judgment based on the position and posture data. If the trajectory deviation is greater than the set threshold, the automatic correction system is triggered to make adjustments. The program decides whether to execute the fast adjustment strategy or the precise correction strategy first, depending on the degree of deviation from the predetermined trajectory. If the deviation is greater than the extreme value, the fast adjustment strategy and the precise correction control strategy are combined for automatic correction.

[0052] A further improvement of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention lies in the following when formulating a rapid adjustment strategy:

[0053] The left and right wheels of the multi-track walking device rotate at the same speed but in opposite directions, so that the entire multi-track walking device can deflect in place around the center of gravity.

[0054] Based on the deflected heading angle, the speed of the left and right track drive wheels is adjusted according to predetermined parameters to gradually reduce the lateral error of the multi-track travel device until it is adjusted to within the lateral error threshold of the precise correction control strategy.

[0055] Using the stationary deflection drive method, drive control is applied to the left and right track drive wheels to eliminate heading angle deviation until the heading angle error threshold of the precise correction control strategy is reached.

[0056] A further improvement of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention lies in the following: when formulating a precise deviation correction control strategy:

[0057] The lateral and heading angle errors of the center of gravity are synchronously controlled and adjusted according to predetermined parameters. The lateral and heading angle adjustment amounts are calculated, the inverse kinematics model of the multi-track walking device is called, and the actual speed control values ​​of the left and right track drive wheels are output.

[0058] The position of the multi-tracked walking device is updated by calling a pure tracking algorithm, enabling the multi-tracked walking device to perform precise correction. The implementation method is as follows:

[0059] The point with the smallest angle between the multi-track walking device's direction of travel and the predetermined path is selected as the tracking target point, and the calculation relationship is as follows:

[0060]

[0061] but:

[0062]

[0063] Among them; l d α represents the distance from the current center of mass of the multi-tracked traveling device to the tracking point; α represents the angle between the current heading angle of the multi-tracked traveling device and the tracking point; R represents the radius of travel of the multi-tracked traveling device; e h For the lateral error of the tracking point; v cIndicates the linear velocity of the center of mass of the multi-track walking device; w c Indicates the angular velocity of the center of mass of the multi-track walking device;

[0064] Based on the position of the tracking target point and the position of the center of mass of the multi-track walking device, the desired adjustment angle is calculated. Based on the desired adjustment angle, the desired pose of the multi-track walking device at the next moment is updated as the target for the next wheel speed adjustment.

[0065] A further improvement of the obstacle avoidance and deviation control method for large multi-tracked walking devices of the present invention lies in that, when planning a new travel path through an automatic obstacle avoidance system, based on the current travel route of the multi-tracked walking device and the detection data of the obstacle by the lidar, the left width and right width of the obstacle on the travel route are calculated with the current travel route as the center:

[0066]

[0067] Among them, w 左 w is the width of the obstacle on the left side of the driving path. 右 w is the width of the obstacle on the right side of the driving path. 障碍物 The shortest widths of the obstacle on the left and right sides of the driving path;

[0068]

[0069] Where L is the width of the multi-tracked walking device; safe_w is the lateral safety width between the multi-tracked walking device and the obstacle; distance is the current distance between the multi-tracked walking device and the obstacle; theda is the angle of the multi-tracked walking device's crab-like rotation to avoid the obstacle; H is the crab-like walking distance of the multi-tracked walking device after the crab-like rotation to avoid the obstacle; and W is the lateral obstacle avoidance width of the multi-tracked walking device.

[0070] A further improvement of the obstacle avoidance control method for large multi-track walking device of the present invention is that the walking path of the multi-track walking device is replanned based on the angle of the target's in-situ crab-like rotation, as well as the distance and lateral obstacle avoidance width of the target's obstacle avoidance after rotation.

[0071] The current multi-track walking device aims to achieve the desired degree of rotation in place using a crab-like motion. First, it performs a crab-like rotation in place; then, it moves forward to complete the calculated obstacle avoidance distance; next, it performs another crab-like rotation in place; finally, it moves forward along the newly planned path.

[0072] The application of the technical solution of the present invention has the following beneficial effects:

[0073] This invention presents a method for controlling the deviation and obstacle avoidance of a large multi-tracked walking device. Through the coordinated operation of an intelligent sensing system, an automatic deviation correction system, and an automatic obstacle avoidance system, it solves the technical problem in existing technologies where the driver's comprehensive control over the equipment and the overall environment leads to significant deviations between the multi-tracked walking device's trajectory and the desired trajectory, making it difficult to achieve the predetermined path. The intelligent sensing system of this invention can comprehensively perceive obstacle information in the external environment, as well as the position and attitude information of the multi-tracked walking device, achieving intelligent long-distance obstacle identification and high-precision positioning and orientation of the multi-tracked walking device. The automatic deviation correction strategy can promptly respond to and correct deviations, adjusting the multi-tracked walking device's trajectory or speed in the environment to guide it towards the expected direction, adapting to different working environments and conditions. Based on obstacle data detected by lidar, an automatic obstacle avoidance algorithm is applied to plan safe paths to avoid obstacles, enabling the multi-tracked walking device to autonomously make obstacle avoidance decisions, ensuring its safe and stable operation.

[0074] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0075] The accompanying drawings, which form part of this application, 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 undue limitation of the invention. In the drawings:

[0076] Figure 1 This is a schematic diagram of the sensor arrangement for the obstacle avoidance and deviation correction control method of the large multi-track walking device of the present invention.

[0077] Figure 2 This is a schematic diagram of step a of the ground slope separation algorithm in the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention;

[0078] Figure 3 This is a schematic diagram of step b of the ground slope separation algorithm in the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention.

[0079] Figure 4 This is a simplified equivalent motion model diagram of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention.

[0080] Figure 5 This is a flowchart of steps 1 and 2 of the rapid adjustment strategy of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention.

[0081] Figure 6 This is a flowchart of step 3 of the rapid adjustment strategy of the obstacle avoidance and deviation correction control method for the large multi-track walking device of the present invention.

[0082] Figure 7 This is a flowchart of the precise correction strategy of the large multi-track walking device obstacle avoidance control method of the present invention;

[0083] Figure 8 This is a pure tracking variable flowchart of the obstacle avoidance and deviation correction control method for the large multi-tracked walking device of the present invention;

[0084] Figure 9 This is a flowchart of the obstacle avoidance procedure of the obstacle avoidance control method for the large multi-track walking device of the present invention.

[0085] Figure 10 This is a schematic diagram of the crab-like rotation angle of the obstacle avoidance and deviation control method for the large multi-track walking device of the present invention.

[0086] Among them, 1. Platform; 2. LiDAR; 3. Beidou satellite data receiver; 4. IMU sensor; 5. Vehicle body; 501. Chassis 1; 502. Chassis 2; 503. Chassis 3; 504. Chassis 4; 505. Chassis 5; 506. Chassis 6; 507. Chassis 7; 508. Chassis 8; 509. Chassis 9; 510. Chassis 10; 511. Chassis 11; 512. Chassis 12; 6. Ramp. Detailed Implementation

[0087] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0088] See Figures 1-10 As shown, a method for correcting deviations and avoiding obstacles in a large multi-tracked walking device is proposed. This method uses an intelligent sensing system to achieve data acquisition, data preprocessing, data fusion, and real-time output of the fused pose information of the multi-tracked walking device based on the fusion positioning of IMU and Beidou satellite data.

[0089] The intelligent sensing system uses 3D LiDAR point cloud data from an intelligent gimbal to perform real-time obstacle detection and determine whether to continue moving.

[0090] When the trajectory deviation exceeds the set threshold, the automatic correction system performs kinematic model analysis of the multi-track walking device, determines the correction of the multi-track walking device, and formulates rapid adjustment strategies and precise correction control strategies.

[0091] When there are obstacles ahead that are impassable or affect the stability of the journey, the automatic obstacle avoidance system plans a new path.

[0092] This invention includes an intelligent sensing system operating in a beach environment: a vehicle-mounted fusion "perception-orientation-positioning" system integrating "LiDAR-BeiDou-Inertial Navigation" was designed. This LiDAR can operate in harsh environments with dense fog and dust, with a detection range exceeding 300m. A LiDAR system based on an intelligent gimbal was developed to quickly acquire scanning point cloud data in front of multi-tracked devices, and the geometric dimensions of obstacles are obtained through point cloud filtering algorithms. A combined navigation system integrating BeiDou satellites and inertial navigation (satellite-inertial navigation combined positioning system) was established, and an unscented Kalman filter algorithm was developed to solve the problems of BeiDou satellite signal loss and inertial navigation error accumulation in dynamic environments.

[0093] The Unscented Kalman Filter (UKF) is a nonlinear filtering method that uses the Unscented Transform (UT) to handle the propagation of the mean and covariance of a nonlinear system.

[0094] This invention discloses an automatic correction method for a multi-track walking device: Based on forward and inverse kinematics theory, a kinematic model of the long, heavy-duty multi-track walking device is established. The speeds of the track wheels on both sides of the multi-track walking device are controlled, and the movement direction of the multi-track walking device is adjusted using the differential speed mode of the inner and outer dual wheels to achieve the effect of walking correction. To achieve fast and stable correction walking, a comprehensive correction method of "fast first, then slow" is invented and calculated on an industrial control computer. The calculation results are sent to the controller to control the opening of the proportional valve, thereby changing the track speed.

[0095] The automatic obstacle avoidance algorithm of the multi-track walking device of this invention: When the multi-track walking device encounters a large obstacle ahead, it plans a path in advance to avoid the obstacle. Based on the obstacle geometry model scanned by LiDAR, the developed automatic obstacle avoidance algorithm of the multi-track walking device is applied to plan a path to avoid the obstacle, adjust the direction of travel, and avoid tall or continuous obstacles.

[0096] Based on the above physical configuration, the automatic obstacle avoidance technology of the present invention enables the multi-track walking device to be more efficient in performing tasks, and can automatically adjust the path without manual intervention, saving time and resources.

[0097] (I) Intelligent Sensing System:

[0098] The intelligent sensing system is equipped with three main types of sensors: LiDAR, an inertial measurement unit (IMU), and BeiDou satellites. Figure 1 As shown, the Beidou satellite data receiver 3 and IMU sensor 4 are installed in the middle of platform 1, and the lidar 2 is installed at the front end of platform 1.

[0099] The specific implementation is as follows:

[0100] ① The following steps are achieved through the fusion positioning of inertial navigation system (IMU) and BeiDou satellite data:

[0101] Step 1: Data collection.

[0102] Raw position and acceleration data are collected from the inertial navigation system (IMU) and BeiDou satellites. The IMU data includes the output of accelerometer and gyroscope data, which is used to measure the linear acceleration and angular velocity of the device. The device's position and time information can be received from the BeiDou satellites.

[0103] Step 2: Sensor data preprocessing.

[0104] Preprocessing of data from the inertial navigation system (IMU) and BeiDou satellites includes noise reduction, calibration, and alignment.

[0105] Step 3: Data fusion.

[0106] An adaptive unscented Kalman filter algorithm is used to fuse IMU accelerometer data and BeiDou satellite positioning data to obtain predicted device position and velocity information. The specific steps are as follows:

[0107] (1) State prediction:

[0108] x k+1 =f(x) k ,u k ) 1);

[0109] Where, x k+1 It is the predicted value at time k+1; f is the system's state transition function; x k It is the predicted value at time k; u k It is the system control input at time k;

[0110] (2) Covariance prediction:

[0111]

[0112] Among them, P k+1 A is the covariance matrix at time k+1; k P is the state transition matrix at time k; k Q is the covariance matrix at time k; k It is the process noise covariance matrix at time k; T is the update cycle time of BeiDou satellite data;

[0113] (3) Measurement and prediction:

[0114] z k+1 =h(xk+1 ) 3);

[0115] Among them, z k+1 is the predicted measurement value at time k+1; h is the observation function;

[0116] (4) Calculate the residuals:

[0117] z″ k+1 =z′ k+1 -z k+1 4);

[0118] Among them, z' k+1 It is the measurement value at time k+1; z” k+1 It is the residual of the state estimate at time k+1;

[0119] (5) Calculate the residual covariance:

[0120]

[0121] Among them, S k+1 H is the measurement residual covariance matrix at time k+1; k+1 It is the observation matrix at time k+1; P k+1 R is the covariance matrix at time k+1; k+1 It is the observation noise covariance matrix at time k+1;

[0122] (6) Calculate the Kalman gain:

[0123]

[0124] Among them, K k+1 It is the Kalman gain at time k+1;

[0125] (7) Update the state estimate:

[0126] x k+1 =x k+1 +K k+1 z″ k+1 7);

[0127] (8) Update the covariance matrix:

[0128] P k+1 =P k+1 -K k+1 H k+1 P k+1 8);

[0129] (9) Update the process noise covariance matrix:

[0130]

[0131] Among them, Qk+1 It is the process noise covariance matrix at time k+1; Q k It is the process noise covariance matrix at time k; α k It is the adaptive coefficient at time k, used to adjust the adjustment rate of process noise;

[0132] (10) Update the observation noise covariance matrix:

[0133]

[0134] Among them, R k β is the observation noise covariance matrix at time k; k It is the adaptive coefficient at time k, used to adjust the adjustment rate of the observation noise.

[0135] Step 4: Real-time output of the fused pose information of the multi-track walking device:

[0136] First, the initial pose information of the multi-track walking device is measured. Then, the gyroscope data of the inertial navigation system (IMU) is denoised and calibrated to obtain the pose information of the device in real time. The specific calculation formula is as follows:

[0137]

[0138] Where, δ xt1 δ is the rotation angle of the device around the X-axis at the initial position. yt1 δ is the rotation angle of the device around the Y-axis at the initial position. zt1 δ is the rotation angle of the device around the Z-axis at the initial position; xt2 δ is the rotation angle of the device around the X-axis at its current position; yt2 δ is the rotation angle of the device around the Y-axis at its current position; zt2 ω is the rotation angle of the device around the Z-axis at its current position; xt1 ω is the angular velocity of the device about the X-axis as measured by the gyroscope at the initial position. yt1 ω is the angular velocity of the device about the Y-axis as measured by the gyroscope at the initial position. zt1 dt is the angular velocity of the device around the Z-axis as measured by the gyroscope at the initial position; dt is the update time of the gyroscope's measurement data.

[0139] ② Real-time obstacle detection and decision-making for continued movement based on 3D LiDAR point cloud data from an intelligent gimbal:

[0140] To address the real-time obstacle detection problem for multi-tracked walking devices navigating in a beach environment, a Euclidean clustering algorithm (a clustering algorithm based on Euclidean distance) is used for real-time obstacle detection. Based on the obstacle information, a decision is made as to whether to continue along the predetermined path. The specific implementation steps are as follows:

[0141] Step 1: First, point cloud data is acquired and transmitted. The LiDAR is mounted on an intelligent gimbal, and the gimbal is controlled by a host computer system to tilt and rotate (rotation angle ±30°) to reduce blind spots in the overhead direction. The LiDAR acquires data on the environment within 1-300m ahead at a frame rate of 20fps. Specifically, the acquired point cloud data can be transmitted to the intelligent sensing system via a high-speed data transmission interface (such as Ethernet).

[0142] Step 2: Use an improved ground slope separation algorithm to separate ground points from non-ground point clouds.

[0143] Step a, as follows Figure 2 As shown:

[0144] a. Project the point cloud onto the XY plane, with the projection range being the horizontal field of view (FOV). H The lidar is divided into S sector regions based on its angular resolution Δα, i.e.

[0145]

[0146] Determine the sector S where the point is located n This requires calculating the ratio of the angle between the line segment formed by the points within the region and the central origin, and the positive direction of the X-axis, to the angular resolution Δα. n The calculation formula is as follows:

[0147]

[0148] Where i is the number of point clouds; y i x is the Y-coordinate value of the point cloud; i The x-coordinate value is the point cloud coordinate.

[0149] Steps b and c, as follows Figure 3 The diagram shown illustrates the improved ground slope separation calculation using auxiliary lines drawn on slope 6.

[0150] b. Those located in the same sector S n Points within the region are sorted by distance, and the value of point P within the region is calculated. i The angle between the point P' and the central axis of the sector is rotated and offset to the vertical central plane of the sector (sector S0), forming a new point P'. i .

[0151] Set the distance range as Where D j is the interval distance value, j is the interval number, and the points in the region are divided equally according to the distance interval range.

[0152] c. Since the actual ground is not a perfectly flat surface, a pre-set ground slope threshold ε is used. min And obtain the distance from points within the region to the central origin, and calculate the ground height threshold H. mim .

[0153] H mim The calculation formula is as follows:

[0154]

[0155] If the Z-coordinate value of the point cloud is within the height threshold, it can be determined as a ground point.

[0156] d. Simultaneously, to further accurately distinguish between ground points and non-ground points, a ground height threshold H is used. mim Based on the comparison, a slope threshold β between adjacent intervals is introduced. min Calculate the slope α of points within adjacent intervals. If it is less than the slope threshold β of adjacent intervals... min If so, it can be determined as a ground point. β min The calculation formula is as follows:

[0157]

[0158] In the formula: z i The Z-coordinate value is the point cloud coordinate.

[0159] Step 3: By improving the Euclidean clustering algorithm, cluster analysis is performed on non-ground point cloud obstacles. The KD-Tree Euclidean clustering algorithm is used for nearest neighbor search, combined with an improved Euclidean clustering algorithm, to cluster each non-ground point cloud into a single obstacle point cloud group. Three-dimensional bounding boxes are then labeled and distinguished, and the maximum XYZ axis coordinates of the three-dimensional boundaries are calculated. The obstacle's position, width, height, and distance information are determined, completing real-time obstacle detection.

[0160] Improving the Euclidean clustering algorithm mainly involves optimizing two key parts: the calculation of the adaptive threshold and a fast clustering algorithm based on mean drift.

[0161] Adaptive Threshold Calculation: Traditional Euclidean clustering algorithms typically use a fixed threshold to determine whether two points belong to the same cluster, which can lead to inaccurate clustering results. To address this issue, an improved algorithm proposes an adaptive threshold calculation method. This method first calculates the distances between all points in the point cloud data and arranges these distances in ascending order. Then, based on the distribution of distances, an appropriate percentile is selected as the threshold. In this way, the threshold can be adjusted according to the actual situation, thereby improving the accuracy of clustering.

[0162] A Fast Clustering Algorithm Based on Mean Shift: Traditional Euclidean clustering uses an iterative method for clustering, which is inefficient when processing large-scale point cloud data. To address this issue, an improved algorithm introduces a fast clustering algorithm based on mean shift. This algorithm first selects a cluster center and calculates the local density based on that center. Then, it adjusts the position of the cluster center according to changes in the local density. By iterating this process repeatedly until the cluster centers no longer change significantly, the final clustering result is obtained. Compared to traditional iterative methods, the fast clustering algorithm based on mean shift can improve clustering efficiency while maintaining accuracy.

[0163] Step 4: Based on the real-time detection output of obstacle information ahead (including obstacle length, width, height, and distance data), and the pre-set obstacle threshold (width threshold w) for the multi-track walking device, min Height threshold h min The system compares the data and determines whether to proceed. If yes, it controls the multi-track walking device to pass; otherwise, it triggers the automatic obstacle avoidance system to plan a new path.

[0164] (II) Automatic correction strategy for multi-track walking devices:

[0165] The sand and soft soil on the beach surface are mobile, causing uncertainty and instability in the terrain. When a multi-tracked walking device travels on this terrain, it is affected by the resistance of uneven road surfaces, which may cause the travel trajectory to deviate from the predetermined trajectory. Therefore, it is necessary to formulate an automatic correction strategy for the multi-tracked walking device.

[0166] ① Kinematic model analysis of multi-track walking device:

[0167] The multi-track walking device is a one-piece structure (with components connected by long pin shafts). The left (or right) track is set to have the same drive speed, which can be simplified to a two-wheel differential drive motion model, such as... Figure 4As shown in the diagram, COM is the center of mass, M1, M2, M3, and M4 represent the four-wheel drive power sources, X and Y represent the positive directions of the coordinate system with the center of mass COM as the origin, R and L represent the equivalent differential wheels on the left and right sides, ICR represents the equivalent rotation center, and r c This represents the equivalent radius of rotation. The equivalent simplified forward kinematics model is expressed as:

[0168]

[0169] The equivalent simplified inverse kinematics model is expressed as:

[0170]

[0171] d LR =βd wb 14);

[0172] Where: w c v is the angular velocity of the center of mass (COM). c v is the linear velocity of the center of mass; β is the equivalent coefficient; l To decompose the velocity based on the center of mass, the linear velocity of the left drive wheel is obtained; v r To determine the linear velocity of the right drive wheel based on the velocity decomposition of the center of mass; d wb d is the wheel spacing; LR This is the equivalent wheel spacing.

[0173] ② Multi-track walking device correction judgment:

[0174] During the movement of the multi-track walking device, its position and posture data need to be monitored in real time. When the height of the road surface unevenness obstacle (slope obstacle / pothole obstacle) does not exceed the preset ground unevenness threshold, in order to ensure the overall stability of the multi-track walking device, the drive device continues to move and pass through the obstacle without making any correction adjustments during the process;

[0175] After the multi-track walking device passes, it makes a correction judgment based on the position and posture data. If the trajectory deviation is greater than the set threshold, the automatic correction system is triggered to make adjustments. The program decides whether to execute a fast adjustment strategy or a precise correction strategy based on the degree of deviation from the predetermined trajectory (the degree is described as heading angle deviation and lateral deviation). The fast adjustment strategy is for heading angle ≥ 2° or lateral error ≥ 2m, and the precise correction strategy is for heading angle < 2° and lateral error < 2m. If the deviation is greater than the extreme value (i.e. the heading angle or lateral error is greater than the set threshold), the fast adjustment strategy and the precise correction control strategy are combined for automatic correction.

[0176] ③ Quickly adjust the strategy, which consists of three steps:

[0177] Step 1: According to the kinematic model calculation results of the multi-track walking device in ①, in order to quickly adjust the travel heading to the predetermined trajectory direction, it is necessary to set the left and right walking wheels of the multi-track walking device to have the same rotation speed and opposite direction, so as to realize that the multi-track walking device as a whole deflects around the center of mass in place.

[0178] Step 2: Based on the deflected heading angle, adjust the speed of the left and right track drive wheels according to the predetermined PID (Process Identifier) ​​parameters. The predetermined speed adjustment range of the moving chassis wheel speed is [-10, 10] rad / s. Perform PID speed adjustment to gradually reduce the lateral error of the multi-track walking device until it is adjusted to within the lateral error threshold of the precise correction control strategy.

[0179] Step 3: Using the stationary yaw drive method, drive the left and right track drive wheels to quickly eliminate the heading angle deviation caused by Step 1 until it reaches the heading angle error threshold of the precise correction control strategy.

[0180] like Figure 5 As shown in the figure, the flowchart shows the process of quickly adjusting the strategy. The specific operation of the flowchart is as follows: first, initialization is performed, the multi-track walking device moves, and data is acquired during the movement (including measurement data from each sensor).

[0181] Detect whether the current multi-track walking device has reached the target position. If yes, end the walking; otherwise, return to the multi-track walking device.

[0182] Simultaneously, perform step one to check whether the eccentricity error meets the specified minimum error. If yes, proceed to step three (line B in the figure). If no, use PID to output the eccentricity adjustment amount based on the eccentricity error. Use the eccentricity adjustment amount and inverse kinematics to adjust the walking speed of the left and right wheels. Check whether the deflection angle error meets the specified value range. If yes, proceed to step 2. If no, return to using PID to output the eccentricity adjustment amount based on the eccentricity error.

[0183] Proceed to step two, reduce the walking speed, set the speed of the left and right wheels of the multi-track walking device to be equal, and check whether the eccentricity error meets the specified minimum error. If yes, proceed to step three (line C in the figure). If no, continue walking and return to the point of reducing the walking speed and setting the speed of the left and right wheels of the multi-track walking device to be equal.

[0184] Proceed to step three, such as Figure 6As shown, the system checks whether the deflection angle error meets the specified minimum error. If yes, it returns to the step before step one after data acquisition (line A in the figure). If no, it uses PID to output output based on the deflection angle error and eccentricity error. The output and inverse kinematics are used to adjust the walking speed of the left and right wheels, and then it returns to the step of checking whether the deflection angle error meets the specified minimum error.

[0185] ④ Precise correction and control strategy, the specific process is as follows: Figure 7 As shown:

[0186] Because there is a certain overshoot during rapid adjustment, considering the travel stability of the multi-track walking device, a precise correction strategy is implemented for adjustment within a small trajectory error limit. The precise correction strategy consists of two steps:

[0187] Step 1: Adjust the lateral and heading angle errors of the center of gravity synchronously using PID control according to preset parameters. The predetermined speed adjustment range of the moving chassis wheel speed is [-10, 10] rad / s. Calculate the lateral and heading angle adjustment amounts, call the inverse kinematics model of the multi-track walking device, and output the actual left and right track drive wheel speed control values ​​(forming a small speed difference between the two sides).

[0188] Step 2: Call the pure tracking algorithm (e.g.) Figure 8 As shown), update the position of the multi-track walking device so that it can perform precise correction in a "cut-in" manner, such as... Figure 7 As shown. The implementation method is as follows:

[0189] The point with the smallest angle between the multi-track walking device's direction of travel and the predetermined path is selected as the tracking target point, as shown in the figure (g). x ,g y () is the tracking point on the next predetermined trajectory, and the center of gravity of the multi-track walking device needs to be controlled to pass through this point. For example... Figure 6 As shown, the calculation relationship is as follows:

[0190]

[0191] but:

[0192]

[0193] Among them, l b The distance from the current center of mass of the multi-tracked traveling device to the tracking point is represented by α; α represents the angle between the current heading angle of the multi-tracked traveling device and the tracking point; R represents the radius of travel of the multi-tracked traveling device, e h For the lateral error of the tracking point; v c The linear velocity of the center of mass of the multi-tracked walking device; w c This represents the angular velocity of the center of mass of the multi-track walking device.

[0194] like Figure 8 As shown, the desired adjustment angle is defined as the heading angle that should be adjusted at the next moment. The calculation method is as follows: at the current moment, take the center point of the multi-track walking device as the starting point and the position of the tracking target point (on the predetermined path) as the ending point to draw a tangent line. The angle formed (tangent angle) is the heading angle that should be adjusted at the next moment, which is the desired angle. Figure 8 In this context, `circulararc` represents the rotation angle of the wheels on both sides of the multi-track walking device, allowing the device to travel along an arc passing through the target waypoint, whose coordinates are (g...). x ,g y ), path represents the predetermined travel path of the multi-track walking device, specifically the radius of the arc that the multi-track walking device follows at a given turning angle.

[0195] Based on the position of the tracking target point and the position of the center of mass of the multi-track walking device, the desired adjustment angle is calculated. Based on the desired adjustment angle, the desired pose of the multi-track walking device at the next moment is updated as the target for the next wheel speed adjustment.

[0196] The desired adjustment angle is defined as the heading angle that should be adjusted at the next moment. The calculation method is as follows: at the current moment, take the center of mass of the multi-track walking device as the starting point and the position of the tracking target point (on the predetermined path) as the ending point to draw a tangent line. The angle formed (tangent angle) is the heading angle that should be adjusted at the next moment, which is the desired angle.

[0197] Figure 7 The flowchart for the precise deviation correction control strategy is as follows: First, initial values ​​are set. Then, the lateral deviation and heading angle deviation of the center of gravity are calculated. It is then determined whether the lateral deviation / heading angle deviation exceeds a deviation threshold. If not, the vehicle continues forward until it reaches a designated position. If so, lateral and heading PID synchronous control of the center of gravity is implemented. The PID adjusts the lateral and heading errors, the inverse kinematics model is called to calculate the track wheel speed, the Pure_Pursuit algorithm (pure tracking algorithm) is called to update the vehicle's pose, and it is again determined whether it has reached the designated position. If yes, movement stops; otherwise, it returns to the step of determining whether the lateral deviation / heading angle deviation exceeds the deviation threshold.

[0198] (III) Path planning for automatic obstacle avoidance of multi-track walking devices:

[0199] As described above, the multi-track walking device moves along a predetermined straight trajectory on the beach terrain. Based on the information of obstacles (height and width) detected by lidar, if there are obstacles ahead that prevent passage or affect the stability of passage (i.e., the obstacle size exceeds the width threshold w), it will detect them. min Or height threshold h minThis triggers an obstacle avoidance strategy and plans a new path. Therefore, an automatic obstacle avoidance path planning algorithm for a multi-tracked walking device is implemented. The implementation steps are as follows:

[0200] Step 1: Based on the current travel route of the multi-track walking device and the obstacle detection data from the lidar, calculate the left width and right width of the obstacle on the travel route, centered on the current travel route:

[0201]

[0202] Among them, w 左 w is the width of the obstacle on the left side of the driving path. 右 w is the right-hand width of the obstacle on the driving path. 障碍物 The shortest widths of the obstacle on the left and right sides of the driving path;

[0203]

[0204] Where L is the width of the multi-tracked walking device; safe_w is the lateral safety width between the multi-tracked walking device and the obstacle; distance is the current distance between the multi-tracked walking device and the obstacle; theda is the angle of the multi-tracked walking device's crab-like rotation to avoid the obstacle; H is the crab-like walking distance of the multi-tracked walking device after the crab-like rotation; and W is the lateral obstacle avoidance width of the multi-tracked walking device.

[0205] Step 2: Determine the angle theda of the multi-track walking device target's in-situ crab-like rotation according to formulas 17) and 18), as well as the distance H and lateral obstacle avoidance width W of the multi-track walking device target after the rotation. Based on the above data, replan the walking path of the multi-track walking device.

[0206] Step 3: First, calculate the in-situ crab-like rotation angle (theda) of the current multi-track walking device. Then, move forward again to complete the calculated obstacle avoidance distance (theda); next, perform the in-situ crab-like rotation angle; finally, move forward along the newly planned path. See the diagram for the crab-like rotation angle of the multi-track walking device. Figure 10 As shown in the figure, there are twelve chassis under the vehicle body 5, namely chassis 1 501, chassis 2 502, chassis 3 503, chassis 4 504, chassis 5 505, chassis 6 506, chassis 7 507, chassis 8 508, chassis 9 509, chassis 10 510, chassis 11 511, and chassis 12 512.

[0207] See the detailed flowchart of the walking obstacle avoidance procedure. Figure 9As shown in the flowchart, the specific operation process is as follows: First, measure the left and right width ratio of the obstacle on the current driving route, calculate the rotation angle, obstacle avoidance driving distance, and lateral obstacle avoidance width, rotate in place by the angle theda, move forward to complete the calculated obstacle avoidance driving distance, rotate in place by the angle -theda, plan a new driving route, and finally end the walking obstacle avoidance program.

[0208] This invention employs a comprehensive correction method of "fast first, slow later". The first step uses a fast adjustment strategy to make a rough adjustment to the multi-track walking device, and the second step uses a precise correction strategy to make a precise adjustment to the multi-track walking device, so that the walking device can maintain the center of gravity position and the overall deflection angle within the allowable error range during the walking process.

[0209] This invention employs a path planning algorithm for automatic obstacle avoidance of a multi-track walking device. Based on obstacle information and the mechanical characteristics of the multi-track walking device itself, and within a safe obstacle avoidance distance, it calculates the minimum rotation angle and the shortest obstacle avoidance travel distance, ensuring that the multi-track walking device can avoid obstacles and avoid collisions during movement.

[0210] This invention presents a method for controlling the deviation and obstacle avoidance of a large multi-tracked walking device. Through the coordinated operation of an intelligent sensing system, an automatic deviation correction system, and an automatic obstacle avoidance system, it solves the technical problem in existing technologies where the driver's comprehensive control over the equipment and the overall environment leads to significant deviations between the multi-tracked walking device's trajectory and the desired trajectory, making it difficult to achieve the predetermined path. The intelligent sensing system of this invention can comprehensively perceive obstacle information in the external environment, as well as the position and attitude information of the multi-tracked walking device (12 independent tracks), achieving intelligent long-distance obstacle identification and high-precision positioning and orientation of the multi-tracked walking device. The automatic deviation correction strategy can promptly respond to and correct deviations, adjusting the trajectory or speed of the multi-tracked walking device in the environment to guide it towards the expected direction, adapting to different working environments and conditions. Based on obstacle data detected by lidar, an automatic obstacle avoidance algorithm is applied to plan safe paths to avoid obstacles, enabling the multi-tracked walking device to autonomously make obstacle avoidance decisions and ensure its safe and stable operation.

[0211] 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 error correction and obstacle avoidance control of a large multi-track walking device, characterized in that, Includes the following steps: The intelligent sensing system, based on the fusion positioning of IMU and Beidou satellite data, realizes data acquisition, data preprocessing, data fusion, and real-time output of the fused multi-track walking device's posture information. The intelligent sensing system uses 3D LiDAR point cloud data from an intelligent gimbal to perform real-time obstacle detection and determine whether to continue moving. When the trajectory deviation exceeds the set threshold, the automatic correction system performs kinematic model analysis of the multi-track walking device, determines the correction of the multi-track walking device, and implements rapid adjustment and precise correction control strategies. When there are obstacles ahead that are impassable or affect the stability of the journey, the automatic obstacle avoidance system plans a new path.

2. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, During data fusion, an adaptive unscented Kalman filter algorithm is used to fuse the accelerometer data from the IMU and the positioning data from the BeiDou satellite to obtain the predicted device location and velocity information. The specific steps are as follows: State prediction: x k+1 =f(x k ,u k ) 1); Where, x k+1 It is the predicted value at time k+1; f is the system's state transition function; x k It is the predicted value at time k; u k It is the system control input at time k; Covariance prediction: Among them, P k+1 A is the covariance matrix at time k+1; k P is the state transition matrix at time k; k Q is the covariance matrix at time k; k It is the process noise covariance matrix at time k; T is the update cycle time of BeiDou satellite data; Measurement and prediction: z k+1 =h(x k+1 ) 3); Among them, z k+1 is the predicted measurement value at time k+1; h is the observation function; Calculate the residuals: With" k+1 =z k+1 -With k+1 4); Among them, z' k+1 It is the measurement value at time k+1; z” k+1 It is the residual of the state estimate at time k+1; Calculate the residual covariance: Among them; S k+1 H is the measurement residual covariance matrix at time k+1; k+1 It is the observation matrix at time k+1; P k+1 R is the covariance matrix at time k+1; k+1 It is the observation noise covariance matrix at time k+1; Calculate the Kalman gain: Among them, K k+1 It is the Kalman gain at time k+1; Updated state estimate: x k+1 =x k+1 +K k+1 z″ k+1 7); Update the covariance matrix: P k+1 =P k+1 -K k+1 H k+1 P k+1 8); Update process noise covariance matrix: Among them, Q k+1 It is the process noise covariance matrix at time k+1; Q k It is the process noise covariance matrix at time k; α k It is the adaptive coefficient at time k, used to adjust the adjustment rate of process noise; Update the observation noise covariance matrix: Among them, R k β is the observation noise covariance matrix at time k; k It is the adaptive coefficient at time k, used to adjust the adjustment rate of the observation noise.

3. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, When outputting the fused pose information of the multi-track walking device in real time, the initial pose information of the multi-track walking device is first measured, and then the gyroscope data of the IMU is denoised and calibrated to obtain the pose information of the device in real time. The specific calculation formula is as follows: Where, δ xt1 δ is the rotation angle of the device around the X-axis at the initial position. yt1 δ is the rotation angle of the device around the Y-axis at the initial position. zt1 δ is the rotation angle of the device around the Z-axis at the initial position; xt2 δ is the rotation angle of the device around the X-axis at its current position; yt2 δ is the rotation angle of the device around the Y-axis at its current position; zt2 ω is the rotation angle of the device around the Z-axis at its current position; xt1 ω is the angular velocity of the device about the X-axis as measured by the gyroscope at the initial position. yt1 ω is the angular velocity of the device about the Y-axis as measured by the gyroscope at the initial position. zt1 dt is the angular velocity of the device around the Z-axis as measured by the gyroscope at the initial position; dt is the update time of the gyroscope's measurement data.

4. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, When using the intelligent sensing system to perform real-time obstacle detection and decision-making based on the 3D LiDAR point cloud data of the intelligent gimbal, the system first collects and transmits point cloud data. Then, an improved ground slope separation algorithm is used to separate the ground point cloud from the non-ground point cloud. Next, an improved Euclidean clustering algorithm is used to cluster and analyze the non-ground point cloud obstacles. Finally, the system compares the real-time obstacle information with the pre-set obstacle clearance threshold of the multi-track walking device and determines whether to make a passage decision. If yes, the multi-track walking device is controlled to pass; otherwise, the automatic obstacle avoidance system is triggered to plan a new path.

5. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, When performing kinematic model analysis of a multi-track walking device using an automatic correction system, the multi-track walking device is a one-piece structure. With the left or right track set to the same driving speed, it is simplified to a two-wheel differential drive kinematic model. The simplified forward kinematic model is expressed as follows: The equivalent simplified inverse kinematics model is expressed as: d LR =βd wb 14); in; w c v is the angular velocity of the center of mass. c v is the linear velocity of the center of mass. l To decompose the velocity based on the center of mass, the linear velocity of the left drive wheel is obtained; v r To determine the linear velocity of the right drive wheel based on the velocity decomposition of the center of mass; d wb d is the wheel spacing; LR β is the equivalent wheel spacing; β is the equivalent coefficient.

6. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, During the correction judgment process of the multi-track walking device, if the height of the obstacle on the road surface does not exceed the preset ground unevenness threshold, the drive device continues to move and passes through the obstacle without making correction adjustments during the process. After the multi-track walking device passes, it makes a correction judgment based on the position and posture data. If the trajectory deviation is greater than the set threshold, the automatic correction system is triggered to make adjustments. The program decides whether to execute the fast adjustment strategy or the precise correction strategy first, depending on the degree of deviation from the predetermined trajectory. If the deviation is greater than the extreme value, the fast adjustment strategy and the precise correction control strategy are combined for automatic correction.

7. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 6, characterized in that, When rapidly adjusting strategies: The left and right wheels of the multi-track walking device rotate at the same speed but in opposite directions, so that the entire multi-track walking device can deflect in place around the center of gravity. Based on the deflected heading angle, the speed of the left and right track drive wheels is adjusted according to predetermined parameters to gradually reduce the lateral error of the multi-track travel device until it is adjusted to within the lateral error threshold of the precise correction control strategy. Using the stationary deflection drive method, drive control is applied to the left and right track drive wheels to eliminate heading angle deviation until the heading angle error threshold of the precise correction control strategy is reached.

8. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 6, characterized in that, When implementing precise correction and control strategies: The lateral and heading angle errors of the center of gravity are synchronously controlled and adjusted according to predetermined parameters. The lateral and heading angle adjustment amounts are calculated, the inverse kinematics model of the multi-track walking device is called, and the actual speed control values ​​of the left and right track drive wheels are output. The position of the multi-tracked walking device is updated by calling a pure tracking algorithm, enabling the multi-tracked walking device to perform precise correction. The implementation method is as follows: The point with the smallest angle between the multi-track walking device's direction of travel and the predetermined path is selected as the tracking target point, and the calculation relationship is as follows: but: Among them, l d α represents the distance from the current center of mass of the multi-tracked traveling device to the tracking point; α represents the angle between the current heading angle of the multi-tracked traveling device and the tracking point; R represents the radius of travel of the multi-tracked traveling device; e h Indicates the lateral error of the tracking point; v c The linear velocity of the center of mass of the multi-tracked walking device; w c This indicates the angular velocity of the center of mass of the multi-tracked walking device; Based on the position of the tracking target point and the position of the center of gravity of the multi-track walking device, the desired adjustment angle is calculated. Based on the desired adjustment angle, the desired pose of the multi-track walking device at the next moment is updated as the target for the next wheel speed adjustment.

9. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 1, characterized in that, When planning a new travel path using the automatic obstacle avoidance system, based on the current travel route of the multi-track walking device and the obstacle detection data from the lidar, the left width and right width of the obstacle on the travel route are calculated, with the current travel route as the center: in; w 左 w is the width of the obstacle on the left side of the driving path. 右 The right-hand width of the obstacle on the driving path; w 障碍物 The shortest widths of the obstacle on the left and right sides of the driving path; Where L is the width of the multi-tracked walking device; safe_w is the lateral safety width between the multi-tracked walking device and the obstacle. distance is the current distance between the multi-tracked walking device and the obstacle; theda is the angle of the multi-tracked walking device's crab-like rotation to avoid the obstacle; H is the crab-like walking distance of the multi-tracked walking device after the crab-like rotation to avoid the obstacle; W is the lateral obstacle avoidance width of the multi-tracked walking device.

10. The obstacle avoidance and deviation correction control method for a large multi-track walking device according to claim 9, characterized in that, Based on the angle of the multi-track walking device target's in-situ crab-like rotation, as well as the distance and lateral obstacle avoidance width of the multi-track walking device target after rotation, the walking path of the multi-track walking device is replanned. The current multi-track walking device aims to achieve a crab-like rotation degree in place, and first performs a crab-like rotation angle in place. Then move forward to complete the calculated obstacle avoidance distance; then, rotate in place in a crab-like motion; finally, move forward along the newly planned path.

Images