Autonomous navigation bending operation four-limb robot and operation method thereof

Abstract

The invention provides an autonomous navigation bending operation four-limb robot and an operation method thereof, and the robot comprises a mobile chassis module which comprises a chassis, and a differential driving wheel set and a follow-up wheel set which are arranged below the chassis; a navigation positioning module; a target identification module; the operation execution module comprises a lower hanging type multi-degree-of-freedom structure arranged on the chassis, the lower hanging type multi-degree-of-freedom structure comprises a lifting push rod and a rotating platform arranged on the lower portion of the lifting push rod, and operation parts are symmetrically arranged below the rotating platform; the control module comprises a chassis control unit and an operation execution module control unit, and the chassis control unit receives the path point information issued by the navigation positioning module through a proportional control strategy and issues linear velocity and angular velocity instructions to control the chassis to move; and the operation execution module control unit performs closed-loop control on an operation component to execute operation through a control algorithm. The robot can be used for automatic pushing operation of an operation target, and unmanned, high-precision, safe and efficient systematized operation is achieved.

Description

Technical Field

[0001] This invention belongs to the technical field of robotics, specifically relating to an autonomous navigation, bending-over-operation four-limbed robot and its operation method. Background Technology

[0002] With the rapid development of robotics technology, intelligent equipment designed to meet the diverse operational needs of agriculture and industry is receiving increasing attention. In agricultural settings, tasks such as turning, ridging, covering, and backfilling involve frequent bending-down movements, are labor-intensive and repetitive, and require precise control and proper posture. Traditional manual operations are inefficient and difficult to standardize, especially in complex environments such as large fields, orchards, and greenhouses, where uneven operation and difficulty in monitoring results are common problems.

[0003] In industrial and emergency rescue scenarios, there are also many dangerous operations that require the intervention of robots, such as handling unexploded ordnance, transferring leaked barrels, handling high-temperature or radioactive hazardous materials, and clearing post-disaster debris. These scenarios require robots to have high obstacle-crossing capabilities, multi-degree-of-freedom control, and remote perception and feedback capabilities.

[0004] While some robot systems in the prior art possess basic mobility and manipulation capabilities, they generally suffer from the following problems:

[0005] The platform lacks a generalized structure for multi-scenario operations, has limited functionality, and makes it difficult to flexibly change operation tools according to the task.

[0006] The autonomous navigation and environmental adaptability are weak, especially in areas with complex terrain or severe positioning obstruction, and the operation path planning and execution stability are poor.

[0007] The lack of accurate target identification and closed-loop control mechanism makes it easy for operation actions to deviate or repeat, failing to meet the requirements of precise operation.

[0008] The operating mechanisms mostly use single-degree-of-freedom open-loop control, which cannot adapt to the needs of spatial attitude adjustment and high-precision bulldozing and turning, and the execution results are difficult to be fed back and judged.

[0009] Therefore, there is an urgent need for a general-purpose operating platform that integrates high chassis obstacle crossing capability, autonomous navigation, target recognition, multi-degree-of-freedom operation and closed-loop control, so as to adapt to various application scenarios such as agriculture, mining, industrial safety and emergency rescue, and achieve intelligent autonomous operation with high precision, high reliability and high safety. Summary of the Invention

[0010] To address the aforementioned technical problems in the prior art, this application provides an autonomous navigation quadruped robot capable of bending over and its operating method.

[0011] According to a first aspect of the present invention, an autonomous navigation, bending-over, quadruped robot is provided, comprising:

[0012] A mobile chassis module, including a chassis, and a differential drive wheel set and a follower wheel set disposed under the chassis;

[0013] The navigation and positioning module adopts a navigation technology that integrates wireless positioning with real-time mapping by LiDAR without pre-mapped navigation.

[0014] The target recognition module uses an image recognition model combined with a two-dimensional LiDAR for composite recognition to determine the spatial position and attitude of the target center.

[0015] The operation execution module includes an under-mounted multi-degree-of-freedom structure mounted on the chassis. The under-mounted multi-degree-of-freedom structure includes a lifting push rod and a rotating platform mounted on the lower part of the lifting push rod. The operation components are symmetrically arranged under the rotating platform.

[0016] The control module includes a chassis control unit and a work execution module control unit. The chassis control unit receives path point information from the navigation and positioning module through a proportional control strategy and issues control commands to control the chassis movement. The work execution module control unit controls the displacement of the working components through a closed-loop control algorithm. This application provides an autonomous navigation, bending-over, four-limbed robot system integrating mobility, perception, recognition, and work capabilities, enabling high-precision autonomous operation and reducing human intervention.

[0017] In some specific embodiments, the chassis control unit simplifies the motion model scene to a two-dimensional plane and extracts the robot's position and orientation: (x, y, θ); in response to the chassis control unit receiving the global coordinates (x, y, θ) of the path points from the navigation and positioning module... obj ,y obj ,θ obj When the drive is enabled, the chassis control unit subscribes to position messages in the odometry topic of the navigation and positioning module. The position messages include the robot's spatial coordinates and a quadruple array representing the robot's spatial orientation. The yaw angle is calculated to obtain the robot's own position and attitude (x). robot ,y robot ,θ robot This solution can accurately obtain the robot's current position and orientation, providing precise input for subsequent path tracking and task alignment.

[0018] In some specific embodiments, the chassis control unit receives waypoint information from the navigation and positioning module through a proportional control strategy, specifically including:

[0019] Based on the robot's own position and orientation (x) robot ,y robot ,θ robot) and target path point position and pose (x obj ,y onj ,θ obj The required rotation angle is calculated using the arctangent function and normalized. Then, the angular velocity around the z-axis is calculated using proportional control until the rotation is successful.

[0020] The straight-line distance is calculated using the formula for the distance between two points in a plane, and then the linear velocity is calculated using proportional control until the target point is reached. This two-stage proportional control process based on real-time pose calculation improves the stability and accuracy of path tracking, effectively reducing attitude errors and navigation drift.

[0021] In some specific embodiments, the formula for calculating the required rotation angle using the arctangent function is θ. target =arctan2(y obj -y robot ),(x obj -x robot The normalization process uses the formula Δθ = arctan2(sin(θtarget-θrobot),cos(θtarget-θrobot)). The absolute value of Δθ is compared with the allowable error to determine whether the steering was successful.

[0022] In some specific embodiments, the proportional control formula is u(t)=Kp*e(t), where Kp represents the proportional coefficient, e(t) represents the error between the current target value and the actual value, and u(t) is the output speed.

[0023] In some specific embodiments, the chassis control unit issues linear velocity and angular velocity commands at a predetermined control frequency and sets a speed limit.

[0024] In some specific embodiments, the job execution module control unit uses a microcontroller as its main control core. Using a microcontroller as an embedded main control chip provides a platform with high real-time performance and high reliability, adapting to multi-degree-of-freedom job control requirements.

[0025] In some specific embodiments, the job execution module control unit establishes bidirectional communication with the host computer through the microcontroller's built-in serial port module. The control data frame of the serial port protocol includes a frame header, operation identifier, operation enable symbol, operation target value, check bit, and frame trailer. The structured serial port protocol enables multi-parameter control and real-time feedback of the working mechanism, ensuring efficient and reliable instruction exchange and status synchronization between the host computer and the slave computer.

[0026] In some specific embodiments, the work execution module control unit uses a control algorithm to perform closed-loop control on the lifting push rod, ensuring that its extension and retraction amount accurately tracks the target extension and retraction amount set by the host computer. The calculation formula for the PID control algorithm is as follows: Δu(k) is the control input, e(k) is the current error, and K is the control variable. p K i K d These represent the proportional, integral, and derivative coefficients, respectively, with Δt being the sampling time interval. A PID closed-loop control algorithm is used to precisely adjust the push rod displacement, enhancing the adaptability of the working mechanism to targets at different heights and improving the accuracy and stability of the action execution.

[0027] According to a second aspect of the present invention, a method for operating an autonomously navigated, prone quadruped robot is provided, utilizing the robot as described above, comprising:

[0028] S1: Receive work location and task instructions;

[0029] S2: The navigation and positioning module utilizes a navigation technology that integrates wireless positioning with real-time mapping using LiDAR. Based on the detected surrounding environment information, its own position, and the work site, it performs real-time path planning and publishes the robot's motion path points to the mobile chassis module.

[0030] S3: The mobile chassis module receives path point information and, through the chassis control unit, uses a proportional control strategy to issue control commands to drive the differential drive wheel set and follower wheel set. It controls the robot to move to the work point by first adjusting the direction and then moving in a straight line. The navigation and positioning module continuously corrects and plans the next path point during the movement.

[0031] S4: After the robot arrives at the work site, the target recognition module uses the forward-looking industrial camera to collect images, uses the image recognition model to detect and select the work target in the image, and combines the ground contour and depth data provided by the two-dimensional LiDAR to determine the precise spatial position (X,Y,Z) and attitude of the center of the work target through the coordinate fusion algorithm.

[0032] S5: The operation execution module drives the under-mounted multi-degree-of-freedom structure to perform operations based on the task instructions and the target position and attitude determined by the target recognition module. The operation execution includes driving the lifting push rod to adjust the working height, driving the rotating platform to adjust the working direction, and driving the working components to perform corresponding operation operations.

[0033] The autonomous navigation, prone-operating quadruped robot and its operating method provided by this invention, through the integration of RTK and LiDAR-based navigation without pre-mapping, YOLOv5 and laser point cloud target recognition, proportional control and PID closed-loop control strategies, constructs an intelligent platform integrating perception, planning, decision-making and high-precision operation. The high ground clearance and differential-speed servo chassis design adapt to the needs of navigating complex terrains; the under-mounted three-degree-of-freedom operating mechanism is compact, flexible in operation, and can achieve high-precision operation. The system possesses high environmental adaptability, robust target recognition, and high operational accuracy, enabling stable operation in complex scenarios such as crop turning in agricultural fields, backfilling blast holes in mines, and material cleanup at disaster sites. It significantly improves operational efficiency, reduces labor costs, and ensures operational safety, possessing broad engineering application prospects and promotional value. Attached Figure Description

[0034] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments and, together with the description, serve to explain the principles of the invention. Other embodiments and many anticipated advantages of the embodiments will be readily recognized as they become better understood through reference to the following detailed description. Elements in the drawings are not necessarily to scale. The same reference numerals refer to corresponding similar parts.

[0035] Figure 1 This is a schematic diagram of the structure of an autonomous navigation, bending-over, four-limbed robot according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the mobile chassis module of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0037] Figure 3 This is a schematic diagram of the operation execution module of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention;

[0038] Figure 4 This is a flowchart illustrating the operation of an autonomous navigation, bending-over, four-limbed robot according to an embodiment of the present invention.

[0039] Figure 5 This is a diagram of the pose message structure in the navigation and positioning module of an autonomous navigation, bending-over four-limbed robot according to a specific embodiment of the present invention.

[0040] Figure 6 This is a diagram of the Twist message structure in the navigation and positioning module of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0041] Figure 7This is a flowchart of the chassis control algorithm for an autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention.

[0042] Figure 8 This is a structural block diagram of a specific embodiment of the Diff Drive Controller of the present invention;

[0043] Figure 9 This is a schematic diagram of a two-wheel differential motion model according to a specific embodiment of the present invention;

[0044] Figure 10 This is a 12-byte data packet in a fixed format for serial communication control commands of the chassis control unit of an autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention.

[0045] Figure 11 This is a 12-byte data packet in a fixed format for a serial communication query command of the chassis control unit of an autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention.

[0046] Figure 12 This is a flowchart illustrating the typical command structure and transmission process of the control unit of the autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention.

[0047] Figure 13 This is a delay distribution diagram of a test system for an autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention.

[0048] Figure 14 This is a line graph showing the success rate and average delay of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0049] Figure 15 This is a diagram illustrating the embedded PID code implementation of an autonomous navigation, bending-over, four-limbed robot, according to a specific embodiment of the present invention. Detailed Implementation

[0050] In the following detailed description, reference is made to the accompanying drawings, which form part of the detailed description and are illustrated by specific illustrative embodiments in which the invention may be practiced. In this regard, directional terms such as “top,” “bottom,” “left,” “right,” “up,” “down,” etc., are used with reference to the orientation of the described figures. Because components of the embodiments can be positioned in several different orientations, directional terms are used for illustrative purposes and are by no means limiting. It should be understood that other embodiments may be utilized or logical changes may be made without departing from the scope of the invention. Therefore, the following detailed description should not be taken in a limiting sense, and the scope of the invention is defined by the appended claims.

[0051] Figure 1 A schematic diagram of the structure of an autonomous navigation, prone-operating quadruped robot according to an embodiment of the present invention is shown, as follows: Figure 1 As shown, the main mechanical structure of the robot includes a mobile chassis module 1 and a work execution module 2. The mobile chassis module 1 is used for the robot to move autonomously in the working environment, and the work execution module 2 is located under the chassis and is suitable for bending over to complete tasks such as scraping, pushing and backfilling.

[0052] Figure 2 A schematic diagram of the mobile chassis module of an autonomous navigation, prone-operating quadruped robot according to a specific embodiment of the present invention is shown, as follows: Figure 2 As shown, a pair of differential drive wheels 11 are installed at the front of the chassis, each driven by a dual-drive servo motor; a pair of follower wheels 12 are installed at the rear, which can automatically adjust the deflection angle according to the steering direction of the front wheels to improve overall handling stability and obstacle crossing ability. Figure 3 This diagram illustrates the structure of the work execution module of an autonomous navigation, prone-operating quadruped robot according to a specific embodiment of the present invention. Figure 3 As shown, the operation execution module 2 includes a vertical lifting push rod 21, a rotating platform 22, and symmetrically arranged operation components below the rotating platform. In this embodiment, the operation components are electric bulldozer blades 23 and 24. The push rod 21 can adjust the height of the operation components, and the rotating platform 22 can achieve ±180° horizontal rotation adjustment, allowing the bulldozing mechanism to adjust its working orientation while the robot is stationary. The two electric bulldozer blades 23 and 24 are symmetrically installed below the rotating platform for specific scraping and backfilling operations. In some other embodiments, the operation components can also be replaced with other mechanisms according to different application scenarios, expanding the application scenarios through a detachable structure.

[0053] In a specific embodiment, the mobile chassis module 1 is based on a customized high ground clearance chassis design, with a ground clearance of 1.15m, a wheelbase of 1.2m, a vehicle length of 1.5m, and a total weight of approximately 240KG. It can effectively traverse irregular soil piles and gravel obstacles. The two front wheels are differential drive wheels 11, using a dual-drive servo motor driver (model DSDAS48150-2E) in conjunction with two servo motors to achieve a differential drive structure. Steering and agile maneuverability are achieved by adjusting the speed difference between the two servo motors. The rear wheels are follow-up wheels 12, which automatically adjust the deflection angle according to the direction of the front wheels during movement, improving handling stability and terrain adaptability.

[0054] In addition to the aforementioned mechanical structure, the robot is equipped with a navigation and positioning module, a target recognition module, and a control module. The navigation and positioning module employs a pre-mapped navigation technology that combines RTK high-precision differential positioning with real-time mapping using LiDAR, eliminating the need for pre-built environmental maps. Upon startup, the system can identify the surrounding environment contours in real time and construct a local cost map. Combined with the absolute coordinates provided by RTK, it achieves stable autonomous path planning and obstacle avoidance. This approach adapts to dynamic changes in mining terrain and is suitable for agricultural and industrial operations in open environments. The target recognition module uses a YOLOv5 image recognition model combined with 2D LiDAR for composite recognition to determine the spatial position and orientation of the target center. The robot's forward-facing industrial camera acquires images in real time as it approaches the target area. The YOLOv5 model detects and selects boreholes in the images. Combined with ground contour and depth data provided by LiDAR, a coordinate fusion algorithm determines the precise spatial position (X, Y, Z) and orientation of the target center. This solution is adaptable to strong light, dust, or occlusion conditions, significantly improving the robustness of target recognition. The control module includes a chassis control unit and a work execution module control unit. The chassis control unit receives path point information from the navigation and positioning module through a proportional control strategy and issues linear velocity and angular velocity commands to control the chassis movement. The work execution module control unit uses a PID control algorithm to control the displacement of the electric bulldozer blades in a closed loop.

[0055] The following is combined with Figure 4 The diagram illustrates the workflow of an autonomous navigation, prone-operating quadruped robot according to an embodiment of the present invention. The working principles and processes of each module of the robot are explained in detail, such as... Figure 4 As shown, the specific workflow includes:

[0056] S1: Receive work location and task instructions;

[0057] S2: The navigation and positioning module utilizes RTK high-precision differential positioning and LiDAR real-time mapping fusion to perform real-time path planning based on the detected surrounding environment information, its own position and work point, and publishes the robot's motion path points to the mobile chassis module.

[0058] S3: The mobile chassis module receives path point information and uses a proportional control strategy through the chassis control unit to issue linear velocity and angular velocity commands to drive the differential drive wheel set and follower wheel set. The robot is controlled to move to the work point by first adjusting the direction and then moving in a straight line. The navigation and positioning module continuously corrects and plans the next path point during the movement.

[0059] S4: After the robot arrives at the work site, the target recognition module uses the forward-looking industrial camera to collect images, uses the YOLOv5 image recognition model to detect and select the work target in the image, and combines the ground contour and depth data provided by the two-dimensional LiDAR to determine the precise spatial position (X,Y,Z) and attitude of the center of the work target through the coordinate fusion algorithm.

[0060] S5: The operation execution module drives the under-mounted multi-degree-of-freedom structure to perform operations based on the task instructions and the target position and attitude determined by the target recognition module through the operation execution module control unit; the operation execution includes driving the lifting push rod to adjust the working height, driving the rotating platform to adjust the working direction, and driving the electric bulldozer blade to perform scraping and backfilling operations.

[0061] In a specific embodiment, the control of the entire chassis is based on the ROS2 Diff Drive Controller. Considering that the robot chassis runs on the ground, the motion model scene is simplified to a two-dimensional plane, extracting only (x, y, θ) to represent its position and orientation (head direction). When the mobile chassis module receives the global coordinates (x, y, θ) of the path point from the navigation and positioning module... obj ,y obj ,θ obj When the driver is enabled, it first subscribes to the location messages published by the topic / odom (odometer) in the navigation and positioning module. The structure of this message is as follows: Figure 5 The pose message structure diagram in the navigation and positioning module shows that the Point position part represents the robot's spatial coordinates, and the Quaternion orientation part, a quaternion array, represents the robot's spatial orientation. As needed, the position coordinates position.x and position.y are read to determine the planar position (x, y). Then, the quaternion array quaternion = (orientation.x, orientation.y, orientation.z, orientation.w) is read, and the quaternion is converted to Euler angles using the formula θ = arctan2((2(w·z+x·y), 1-2(y)). 2 +z 2 )) , calculate the yaw angle, and obtain the robot's own position and attitude (x robot ,y robot ,θ robot ).

[0062] Figure 6 This diagram illustrates the Twist message structure in the navigation and positioning module of an autonomous navigation, prone-operating quadruped robot according to a specific embodiment of the present invention. Figure 6As shown, the message includes both linear velocity and angular velocity, represented by Vector3 for the x, y, and z vector directions. Considering the robot chassis can only steer or move linearly along the x-axis, only linear.x (forward or backward velocity in m / s) and angular.z (angular velocity around the z-axis in rad / s) are published to the topic / cmd_vel.

[0063] In a specific embodiment, the mobile chassis employs a proportional control strategy, generating control signals by adjusting the error proportionally, as shown in the formula u(t) = Kp * e(t), where Kp represents the proportional coefficient, e(t) represents the error between the current target value and the actual value, and u(t) is the output speed. This formula is a function with time as the variable. During each topic posting cycle, the linear or angular velocity that the robot chassis should output for the next movement is calculated, and the robot moves accordingly, enabling it to smoothly move to the target point. To prevent the output speed from exceeding the safety threshold due to excessive proportional coefficients or errors, the mobile module sets a speed upper limit during movement. When the output u(t) exceeds this upper limit, the robot still moves according to the speed upper limit.

[0064] The specific control is carried out in two steps: first, adjusting the direction, and then moving towards the target point. First, based on its own position and attitude (x... robot ,y robot ,θ robot ) and target path point position and pose (x obj ,y obj ,θ obj Using the arctangent function, the angle between the line connecting two points in the plane and the x-axis is obtained. Subtracting the current yaw angle gives the required rotation angle θ. target , such as formula θ target =arctan2(y obj -y robot ),(x obj -x robotThe angle is normalized using the arctangent function to obtain the final parameter Δθ = arctan2(sin(θtarget-θrobot),cos(θtarget-θrobot)), making its range [-π,π]. The absolute value of Δθ is compared with the allowable error (e.g., 0-5°) to determine if the turn was successful. If unsuccessful, angular.z is calculated and published to the topic / cmd_vel. This process is repeated at a predetermined frequency (50Hz in this embodiment), continuously publishing new angular.z until the mobile chassis is deemed to have turned successfully and can proceed in a straight line towards the target point. The straight-line distance d between the mobile chassis coordinates and the target path point is calculated using the two-point distance formula. d is compared with the error. If the target point has not been reached, linear.x is calculated and published to the topic / cmd_vel. This process is repeated at a frequency of 50Hz, continuously publishing new linear.x. During this period, the yaw angle is still used as a premise, and adjustments are made continuously until the target path point is reached. The complete control algorithm logic is as follows: Figure 7 The diagram shows a flowchart of the chassis control algorithm for an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0065] The theoretical output value obtained through the above steps will be converted into actual motor speed output using the ROS2 official differential drive controller, the Diff Drive Controller. The Diff Drive Controller, part of the ROS2_Controllers package, is a highly integrated modular controller designed specifically for two-wheeled differential drive robots, providing a complete functional chain from high-level speed commands to low-level motor control. The specific framework is as follows... Figure 8 The diagram illustrates a specific embodiment of the Diff Drive Controller of the present invention. This module primarily uses its subscription ` / cmd_vel` to convert the two-wheel differential speed model into the speeds of each drive wheel. The two-wheel differential speed model is as follows: Figure 9 The diagram illustrates a two-wheel differential motion model according to a specific embodiment of the present invention. Where w is the wheelbase (width); u left The linear velocity of the left wheel; u right ω is the linear velocity of the right wheel. b,z v is the angular velocity about the z-axis of the vehicle body coordinate system; b,x The velocity along the x-axis of the vehicle body coordinate system; y b Let x be the y-axis of the vehicle body coordinate system; w y w Let x and y be the world coordinate system axes; θ be the turning angle. When v l =v r At that time, the robot moves along a straight line, vb,x =v l =v r When v l ≠v r At that time, the forward kinematics formula is: To find the inverse kinematics, the formula is v. left =v b,x -ω b,z w / 2, v right =v b,x +ω b,z w / 2. Using the forward kinematics formula, the velocity and rotational angular velocity can be calculated from the linear velocities of the left and right wheels. Through cumulative calculations, the continuous position and angle changes of the robot, as well as its position and orientation at any given time point (x...), can be obtained. robot ,y robot ,θ robot ). Publish it to the topic / Odom to assist RTK in determining the robot's position after movement. By solving the inverse kinematics, the linear.x and angular.z files published in the / Cmd_vel topic, i.e., v, can be determined. b,x ω b,z The speed is converted into the linear velocity of the two wheels. Finally, through linear velocity-angular velocity conversion, it is converted into the rotational speed of the two wheels. According to the driver's technical documentation, the host computer can communicate with the driver via RS232. This module selects a serial port rate of 115200bps, no parity check, and Hex transmission and reception. Data transmission and reception are performed in data frames, each consisting of 12 8-bit bytes. The data format is divided into control and query types.

[0066] The control format, mainly used to enable the dual-channel driver and set the target speed, is shown in Table 1. In the data frame, the first bit E0 is the control identifier, and the second bit is the enable signal: 00 for dual-channel disabled, 01 for channel A enabled, 02 for channel B enabled, and 03 for dual-channel enabled. Bits 3 and 4 are not enabled. Bits 5 to 8 are for channel A speed input, and bits 9 to 12 are for channel B speed input. The value FF FF D8 F0—00 002710 (hexadecimal) corresponds to the motor's negative rated speed—positive rated speed (percentage). The positive speed portion is directly converted from 1000 (decimal), while the negative speed portion uses a padded form: 1 00 0000 00-00 00 27 10 (hexadecimal).

[0067] Table 1

[0068]

[0069] The query format is mainly used to query the current operating status of the motor, as shown in Table 2. The first bit, ED, in the data frame is the query identifier, and the second bit is the query type. The specific numerical meanings are as follows: 0x00: System control status; 0x01: Motor rotor position electrical angle (0-1000°); 0x02: Motor speed (with sign, rpm); 0x03: Motor winding current; 0x04: Motor rotor position mechanical angle (0-1000°); 0x05: Controller power supply voltage; 0x06: Controller temperature; 0x07: Motor fault status; 0x08: Motor rotor position mechanical position (position control mode); 0x10: System version number. Bits 3 to 12 are distinguished according to the query type.

[0070] Table 2

[0071]

[0072] This driver can use the Hall effect signal inside the motor as rotor position feedback, and in conjunction with an external incremental encoder (1000-2500 lines) or various rotor position sensor signals to control motor movement, it can achieve internal speed closed-loop and position closed-loop control. Considering that it is used to drive a differential chassis and the servo motor uses magnetic encoder feedback, the speed closed-loop control mode can better support the Twist control flow in ROS2, and the control is continuous, responsive, and easier to implement path planning and obstacle avoidance. The position mode is suitable for positioning in high-precision machining scenarios. Therefore, speed closed-loop control is adopted as the control mode in this application. The control implementation is developed based on the Python language, reading the drive wheel speed output from the Diff Drive Controller, and sending commands through serial communication to control the driver. This includes functions such as motor speed control, status query, and stop control.

[0073] The serial communication section uses the Python pyserial package, and the control commands are in a fixed format of 12-byte data packets, such as... Figure 10 The illustration shows a fixed-format 12-byte data packet for serial communication control commands of the chassis control unit of an autonomous navigation, prone-operating, four-limbed robot according to a specific embodiment of the present invention. The fixed-format 12-byte data packet for query commands is shown below. Figure 11 The illustration shows a fixed-format 12-byte data packet for the serial communication query command of the chassis control unit of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention. By combining the instruction `val = int(rpm / self.max_rpm*10000)` to convert the rotational speed to a percentage of the rated speed, and then to a hexadecimal number, the drive wheel speed read from the Diff DriveController can be input into the driver.

[0074] In a specific embodiment, all push rods in the work execution module adopt a DC motor and reducer structure, using an H-bridge circuit constructed with BTS7960 chips, supporting bidirectional drive and a maximum current of 43A. Two digital signals are output to IN1 and IN2 of the BTS7960 to control the push rod's movement direction, and a PWM signal is output to the enable pin to control the advance speed. Each push rod integrates a potentiometer, which outputs an analog voltage synchronously with the push rod's movement. The drive section of the rotating platform uses a combination of a stepper motor, a planetary reducer, and a stepper motor driver. Pulse signals are output to control the stepping rotation, and direction signals set the rotation direction. By controlling the number and frequency of pulses, precise rotation within a range of ±180° can be achieved.

[0075] In a specific embodiment, the control system of the working mechanism uses an STM32F103C8T6 microcontroller as the main control core. It reads and parses data frames sent from the host computer to obtain the target output value, and samples the analog voltage output by the potentiometer through an ADC, mapping it to the actual length. An internal PID control algorithm is used to control the push rod to move to the target value in a closed loop. The stepper motor control accuracy depends on its own step angle; only abnormalities such as slippage need to be addressed, so open-loop control is adopted. Therefore, this application independently designed a communication protocol, control strategy, and calibration verification method to ensure the reliability and stability of the system performance, constructing an integrated, high-response, multi-degree-of-freedom coordinated control embedded system platform.

[0076] In a specific embodiment, the main control unit establishes bidirectional communication with the host computer through the built-in USART serial port module of the microcontroller, port numbers PA9 (RXD) and PA10 (TXD). The serial port adopts standard TTL level, baud rate of 115200bps, no parity bit, and Hex for transmission and reception. Since the entire main control involves four independent control signals and three feedback signals, in order to avoid controlling components that may interfere with each other at the same time and to improve communication efficiency, seven 8-bit bytes are used as a control data frame, including frame header byte[0], operation identifier byte[1], operation enable byte[2], operation target value (elongation / contraction, rotation, PID parameter) byte[3:2] (high byte first, low byte last), parity bit byte[5], and frame tail byte[6], as shown in Table 3 below.

[0077] Table 3

[0078]

[0079] The specific data meanings are as follows: (1) Frame header: AA (fixed value). (2) Operation identifier: 00: horizontal push rod 1, 01: horizontal push rod 2, 02: vertical push rod, 03: rotating platform, 04: horizontal push rod 1 and 2. Operation enable symbol: 00: reverse, 01: forward, 02: Kp assignment (proportional parameter), 03: Ki assignment (integral parameter), 04: Kd assignment (differential parameter), 05: Kp reading, 06: Ki reading, 07: Kd reading, 08: displacement reading. Operation target quantity: angle: 00 00-4650 (0-180.00 degrees * 100.0f), length: 00 00-13 88 (0-500.0mm * 10.0f), (3) parameter assignment (* 1000.0f). (4) Check bit: XOR check. (5) Frame tail: 55 (fixed value).

[0080] The component within parentheses in the target quantity of the operation, such as *10.f, indicates that the transmitted data is 10 times the original data. When the microcontroller processes the data, it needs to be divided by this value to restore the original data.

[0081] In order to monitor the system status, this application also designed a communication protocol for the data feedback part. This part of the data is different depending on the application scenario. However, it consists of 6 8-bit bytes as a data frame, including frame header byte[0], return data type byte[1], data value (elongation / contraction, rotation, PID parameter) byte[3:2] (high-order byte first, low-order byte last), check bit byte[4], and frame tail byte[5], as shown in Table 4 below.

[0082] Table 4

[0083]

[0084] The first type is used to respond to control commands transmitted from the host computer. Each response lasts for 1 second with a 0.1 second interval. The corresponding format is as follows: (1) Frame header: AA (fixed value). (2) Return data type: 00: ACK (response), 01: horizontal 1 push rod displacement, 02: horizontal 2 push rod displacement, 03: vertical 3 push rod displacement, 04, 05, 06: horizontal 1 push rod Kp, Ki, Kd, ​​07, 08, 09: horizontal 2 push rod Kp, Ki, Kd, ​​0A, 0B, 0C: vertical push rod Kp, Ki, Kd. (3) Query data or control command confirmation: length: 0000-1388 (0-500.0mm*10.0f), parameter reading value (*1000.0f). (4) Check bit: XOR check. (5) Frame tail: 55 (fixed value).

[0085] The second method is to periodically upload the system status (busy / idle / error) in byte[1] at a frequency of 50 Hz. In the busy state, only query commands are supported, in the idle state, all commands are supported, and in the error state, a continuous alarm is triggered and the microcontroller interrupts all current operations. The specific data definition for each byte is as follows: (1) Frame header: AA (fixed value). (2) System status: IDLE: F0, BUSY: F1, ERROR: FF. (3) Check bit: XOR check. (4) Frame tail: 55 (fixed value).

[0086] Based on the communication protocol design, the host computer part is written in Python and uses the pyserial library to communicate with the STM32. Its main functions include constructing and sending command frames (with automatic verification), parsing the response frames and feedback values ​​returned by the slave computer, starting an independent thread to receive feedback and monitor the status in real time, and providing several function interfaces to control the actuator, rotate the platform, and set or read PID parameters. A typical command construction and sending process is as follows: Figure 12 The diagram illustrates a typical command structure and transmission flowchart for the work execution module control unit of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0087] The lower-level machine uses an STM32 microcontroller (STM32F103) to implement serial port reception and data processing. Serial port reception employs an interrupt callback mechanism. Upon receiving a complete frame (7 bytes), the RxFlag flag is set, and the SerParser() function is called in the main loop to parse the command. Command processing uses a switch-case structure. The target device is selected based on the command ID, and the corresponding action is executed according to the opcode CMD, including setting or reading PID parameters (Kp, Ki, Kd); lever movement control (forward / backward); reading the current physical length (or position); and sending an acknowledgment frame (ACK).

[0088] To evaluate the reliability and real-time performance of the designed communication protocol in actual operation, this application conducted systematic testing and analysis on key indicators such as response success rate, average latency, and latency fluctuation range during serial communication. The test platform was Windows 11, Python 3.9, and the pyserial library; the serial port settings were: baud rate 115200, port COM3, and timeout 0.1 seconds. 1000 commands (e.g., GET_KP) were sent consecutively, and the test was repeated 10 times, with a 1-second interval between each round. The success rate of each response and the round-trip time (in milliseconds) were recorded.

[0089] Test system latency distribution as follows Figure 13The latency distribution diagram of the test system for an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention is shown. The mean (5.47ms) and median (5.48ms) are almost identical, indicating that the communication quality of the system has excellent stability and consistency. Most request latencies in the diagram are concentrated between 5.0-6.0ms, and the distribution of the main latency range is relatively even. Under interference-free testing conditions, communication failures are almost nonexistent. It is worth noting that a small number of request latencies exceed 6ms. These outliers may originate from network jitter or instantaneous system load. However, statistics show a standard deviation of 0.3ms, indicating that the overall latency fluctuation is extremely small, and these outliers have a negligible impact on the entire communication system. The success rate versus average latency line graph is shown below. Figure 14 The diagram shows a line graph illustrating the success rate versus average latency of an autonomously navigated, bend-over quadruped robot according to a specific embodiment of the present invention. The system maintained excellent stability during testing, with a 100% communication success rate across all rounds and minimal fluctuations in average latency per round.

[0090] PID control is a combination of proportional, integral, and derivative control components, widely used in closed-loop feedback control systems. Its main objective is to enable the controlled variable (actuator extension / retraction) to reach the setpoint quickly and stably, while minimizing overshoot, steady-state error, and system oscillation. In discrete systems, the PID control algorithm is expressed as the formula... Simplify it into incremental form Where Δu(k) is the control variable, e(k) is the current error, and K p K i K d These are the proportional, integral, and derivative coefficients, respectively, with Δt representing the sampling time interval. By testing and adjusting the coefficients, the proportional term can achieve a fast response and amplify the error. However, when the error approaches zero, the output also approaches zero, which is insufficient to overcome system disturbances or nonlinear characteristics, resulting in steady-state error. The integral term can accumulate the steady-state error, continuously increasing and increasing the control output until the deviation disappears. This "memory" mechanism allows integral control to compensate for steady-state error, even if proportional control alone is insufficient to eliminate the error. However, this also means that even when the error begins to decrease, the integral term may still maintain a large value. When the actual system output approaches the setpoint, an excessively large integral term can lead to an excessively large control output, causing oscillations. The derivative term predicts the future behavior of the error by observing its rate of change. When the error changes rapidly, the derivative term generates a large signal to counteract this rapid change, thereby suppressing further error growth, inhibiting system abrupt changes and oscillations, and improving system stability.

[0091] In this application, the actual extension / retraction of the actuator (potentiometer feedback) is precisely tracked by the target extension / retraction amount set by the host computer. First, the target extension / retraction amount sent by the host computer is received. This is converted to the current position of the actuator by reading the analog signal from the potentiometer, and the error is calculated. PID control is executed to address proportional error; integral cumulative deviation is used to eliminate long-term errors; and derivative predictive trend is used to suppress oscillations. A PWM output is provided, setting the direction (IN1 / IN2) and speed (PWM duty cycle). A timer drives the PID control output (50Hz) every 20ms. The code implementation is as follows: Figure 15 The diagram shows the embedded PID code implementation of an autonomous navigation, bending-over, four-limbed robot according to a specific embodiment of the present invention.

[0092] This invention structures the input-output relationships of each functional module of the robot system, forming a clear module collaboration system: the navigation and positioning module takes RTK coordinates and laser point clouds as input and outputs path point coordinates and attitude information; the mobile chassis module receives path point information and outputs drive wheel speed and mileage feedback; the target recognition module fuses images and laser point clouds and outputs the spatial coordinates of the target center; the operation execution module outputs potentiometer feedback signals according to the control commands of the push rod and the rotating platform to realize operation status monitoring.

[0093] The autonomous navigation, bending-over four-limbed robot and its operation method described in this application have the following technical advantages: **Multi-scene autonomous high-precision operation:** By integrating RTK differential positioning and LiDAR SLAM navigation technologies, it achieves autonomous path planning and environmental mapping without the need for pre-mapping, adapting to dynamic operation requirements in complex environments such as mines, farmland, greenhouses, and disaster areas. **Strong target recognition and positioning capabilities:** Utilizing the YOLOv5 image recognition algorithm combined with two-dimensional LiDAR point cloud information, it achieves highly robust recognition of the spatial position and attitude of the target, providing support for precise positioning and alignment. **Full closed-loop path tracking and operation control:** Employing a proportional control strategy and ROS2 control framework, it achieves a real-time closed loop of motion planning, error calculation, attitude adjustment, and speed control; the operating mechanism uses PID control to ensure precise execution of lifting and rotational movements. **Modular and expandable operation framework:** The robot adopts a differential chassis combined with an under-mounted three-degree-of-freedom operating structure, which is compact and flexible in operation. The bulldozer blade can be replaced with tools such as a shovel, explosion-proof fork, and rake teeth, supporting various applications in agriculture, industry, and disaster prevention. Improved operational efficiency and enhanced operational safety: The system operates stably with low communication latency and timely response, significantly reducing manual intervention and effectively mitigating personnel exposure risks in high-risk scenarios, thereby improving both operational efficiency and system safety. Flexible system deployment and engineering application potential: Clear and unambiguous information input and output for each module, along with a decoupled hardware and software design for easy maintenance, enable rapid deployment, large-scale replication, and engineering promotion in open environments.

[0094] It is evident that those skilled in the art can make various modifications and alterations to the embodiments of the present invention without departing from the spirit and scope of the invention. In this way, the invention is also intended to cover such modifications and alterations if they fall within the scope of the claims and their equivalents. The word "comprising" does not exclude the presence of other elements or steps not listed in the claims. The simple fact that certain measures are described in mutually different dependent claims does not indicate that a combination of these measures cannot be used for profit. Any reference numerals in the claims should not be considered as limiting the scope.

Claims

1. An autonomous navigation, prone-operating quadruped robot, characterized in that, include: A mobile chassis module includes a chassis, and a differential drive wheel set and a follower wheel set disposed under the chassis; The navigation and positioning module adopts a navigation technology without pre-mapped navigation that combines wireless positioning with real-time mapping by LiDAR. The target recognition module uses an image recognition model combined with a two-dimensional LiDAR for composite recognition to determine the spatial position and attitude of the target center. The operation execution module includes an under-mounted multi-degree-of-freedom structure mounted on the chassis. The under-mounted multi-degree-of-freedom structure includes a lifting push rod and a rotating platform mounted below the lifting push rod. The rotating platform is symmetrically equipped with operation components. The control module includes a chassis control unit and a work execution module control unit. The chassis control unit receives path point information issued by the navigation and positioning module through a proportional control strategy and issues control commands to control the movement of the chassis. The work execution module control unit controls the displacement of the work component through a closed-loop control algorithm.

2. The autonomous navigation, prone-operating quadruped robot according to claim 1, characterized in that, The chassis control unit simplifies the motion model scene to a two-dimensional plane and extracts the robot's position and orientation: (x, y, θ); in response to the chassis control unit receiving the global coordinates (x, y, θ) of the path point from the navigation and positioning module... obj ,y obj ,θ obj When the drive is enabled, the chassis control unit subscribes to the position messages in the odometer topic of the navigation and positioning module. The position messages include the robot's spatial coordinates and a four-element array representing the robot's spatial orientation. The yaw angle is calculated to obtain the robot's own position and attitude (x). robot ,y robot ,θ robot ).

3. The autonomous navigation, prone-operating quadruped robot according to claim 2, characterized in that, The chassis control unit receives the waypoint information published by the navigation and positioning module through a proportional control strategy, specifically including: Based on the robot's own position and orientation (x) robot ,y robot ,θ robot ) and target path point position and pose (x obj ,y obj ,θ obj The required rotation angle is calculated using the arctangent function and normalized. Then, the angular velocity around the z-axis is calculated using proportional control until the rotation is successful. The straight-line distance is calculated using the formula for the distance between two points in a plane, and then the linear velocity is calculated by controlling the ratio until the target point is reached.

4. The autonomous navigation, prone-operating quadruped robot according to claim 1, characterized in that, The formula for calculating the required rotation angle using the arctangent function is θ. target =arctan2(y obj -y robot ),(x obj -x robot The normalization process uses the formula Δθ = arctan2(sin(θtarget-θrobot),cos(θtarget-θrobot)). The absolute value of Δθ is compared with the allowable error to determine whether the steering was successful.

5. The autonomous navigation, prone-operating quadruped robot according to claim 3, characterized in that, The proportional control formula is u(t)=Kp*e(t), where Kp represents the proportional coefficient, e(t) represents the error between the current target value and the actual value, and u(t) is the output speed.

6. The autonomous navigation, prone-operating quadruped robot according to claim 3, characterized in that, The chassis control unit issues linear velocity and angular velocity commands at a predetermined control frequency and sets a speed limit.

7. The autonomous navigation, prone-operating quadruped robot according to claim 1, characterized in that, The control unit of the job execution module uses a microcontroller as its main control core.

8. The autonomous navigation, prone-operating quadruped robot according to claim 7, characterized in that, The job execution module control unit establishes bidirectional communication with the host computer through the serial port module built into the microcontroller. The control data frame of the serial port protocol includes a frame header, operation identifier, operation enable symbol, operation target value, check bit, and frame tail.

9. The autonomous navigation, prone-operating quadruped robot according to claim 8, characterized in that, The control unit of the operation execution module performs closed-loop control of the lifting push rod through a control algorithm, ensuring that its extension and retraction amount accurately tracks the target extension and retraction amount set by the host computer. The calculation formula of the control algorithm is as follows: Δu(k) is the control input, e(k) is the current error, and K is the control variable. p K i K d These are the proportional, integral, and derivative coefficients, respectively, and Δt is the sampling time interval.

10. A method for operating an autonomous navigation, bending-over, four-limbed robot, utilizing the robot as described in any one of claims 1-9, characterized in that... include: S1: Receive work location and task instructions; S2: The navigation and positioning module utilizes a navigation technology that integrates wireless positioning and real-time mapping with LiDAR without pre-mapping. Based on the detected surrounding environment information, its own position, and the work point, it performs real-time path planning and publishes the robot's motion path points to the mobile chassis module. S3: The mobile chassis module receives the path point information and issues control commands through the chassis control unit using a proportional control strategy to drive the differential drive wheel set and the follower wheel set. The robot is controlled to move to the work point by first adjusting the direction and then moving in a straight line. The navigation and positioning module continuously corrects and plans the next path point during the movement. S4: After the robot arrives at the work site, the target recognition module uses a forward-looking industrial camera to collect images, uses an image recognition model to detect and select the work target in the image, and combines the ground contour and depth data provided by the two-dimensional lidar to determine the precise spatial position (X,Y,Z) and attitude of the center of the work target through a coordinate fusion algorithm. S5: The operation execution module drives the under-hanging multi-degree-of-freedom structure to perform operations according to the task instructions and the target position and attitude determined by the target recognition module through the operation execution module control unit; the operation execution includes driving the lifting push rod to adjust the working height, driving the rotating platform to adjust the working direction, and driving the working components to perform the corresponding operation.

Images