Wheeled autonomous following robot and control method thereof

By combining a wheeled autonomous following robot with a depth vision camera, UWB module and multi-line lidar, the problem of poor acquisition of target information and simple path planning in the existing technology is solved, and it can still follow and autonomy after losing the target. The path is planned to improve the passing of the chassis.

CN119065380APending Publication Date: 2024-12-03SHENYANG AEROSPACE UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411392456.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-08
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The existing follower robots mainly rely on vision technology, and the acquisition of target information is poor, and they cannot continue to follow once the camera is lost; the application scenario is simple and the path planning is not possible independently; the chassis passability is poor, making it difficult to meet actual needs.

Method used

Adopting wheeled autonomous follower robots, combining depth vision cameras, UWB modules and multi-line lidar, through improved D* path planning algorithm and SiamRPN vision tracking algorithm, autonomous path planning and obstacle detection are realized to ensure that targets can still be lost. Follow, and the chassis adopts four-wheel differential drive and an independent suspension system for upper and lower double-wrench arms to improve passability.

Benefits of technology

It can still follow independently after losing the target, and independently plan paths and avoid obstacles in complex scenarios, improving the following accuracy and chassis passability, and meeting practical application needs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119065380A_ABST
    Figure CN119065380A_ABST
Patent Text Reader

Abstract

The invention provides a wheeled autonomous following robot and a control method thereof, and relates to the technical field of mobile robots. The chassis system is used for installing a power mechanism and all parts, and normal driving is guaranteed. The control system comprises a control module and an environment sensing module, and an upper computer in the control module receives sensor data of the environment sensing module, calculates the data, makes a corresponding control decision, sends a control instruction to a lower computer and controls the hub motor to complete the corresponding decision. The environment sensing module comprises a depth vision camera, a UWB module and a multi-line laser radar; the depth vision camera is responsible for identifying and tracking a target; the multi-line laser radar is responsible for scanning and extracting obstacle information; and the UWB module is responsible for continuously acquiring target information after the camera loses the following target so as to achieve the purpose of continuously executing the following task.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mobile robots, and in particular to a wheeled autonomous following robot and a control method thereof. Background Art

[0002] In the military field, the quality of equipment will affect the direction of victory in modern wars. With the continuous development of science and technology and the increasing complexity of mission execution, the types of individual equipment are becoming more and more diverse. When performing special operations, combat readiness patrols and other tasks, soldiers often need to carry weapons, supplies, special equipment and other heavy objects with them, resulting in high load-bearing intensity, slow maneuvering speed, and poor sustained combat capability. To solve the above problems, follow-up robots came into being. Follow-up robots are also widely used in the civilian field. They can replace manual labor and perform tasks such as material transportation and emergency rescue. For example, they can follow staff to carry items in factories, farms, supermarkets and other places. They can also carry rescue equipment and follow rescue personnel to carry out emergency rescue activities after natural disasters such as earthquakes, fires, and floods. In summary, the application of follow-up robots is of great significance in reducing the labor intensity of personnel, improving the quality of work, and improving work efficiency.

[0003] Most of the current following robots rely mainly on visual technology, which is not effective in obtaining target information, and once the camera loses the following target, it cannot continue to complete the following task. At the same time, the application scenarios of existing following robots are generally relatively simple, and they cannot autonomously plan paths to find the following target. The chassis has poor passability, which makes it difficult to meet user needs in actual applications.

[0004] Therefore, providing a wheeled robot and a control method thereof that can autonomously plan paths, autonomously avoid obstacles, obtain accurate target information, and autonomously follow the target after losing the target is an urgent problem to be solved by technical personnel in this field, and it is of great significance to the development and promotion of the application field of robots. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a wheeled autonomous following robot and a control method thereof. The present invention solves the problems that the current following robot mainly relies on visual technology, has poor effect in obtaining target information, and cannot continue to complete the following task once the camera loses the following target; the application scenarios are generally relatively simple, and it is impossible to autonomously plan the path to find the following target; the chassis has poor passability, and it is difficult to meet user needs in actual application.

[0006] The technical solution provided by the present invention is as follows:

[0007] On the one hand, a wheeled autonomous following robot includes a chassis system and a control system;

[0008] The chassis system specifically includes a body module, a power module, a powertrain, a suspension module and a protection module;

[0009] The vehicle body module is composed of four wheels and a vehicle body frame; the wheels use off-road tires with a diameter of 300 mm;

[0010] The power module is composed of a number of lithium batteries connected in series and installed inside the vehicle body frame; the output end of the power module is connected to the control system and the powertrain;

[0011] The powertrain uses four 500W wheel hub motors, the output ends of the four wheel hub motors are connected to the four wheels respectively, and the total power of the wheel hub motors in series is P=2kW; the autonomous following robot adopts four-wheel differential drive;

[0012] The suspension module adopts an upper and lower double wishbone independent suspension system, which is composed of an upper wishbone, a connector, a shock absorber and a lower wishbone; the lower end of the shock absorber is connected to the lower wishbone, the upper end of the shock absorber is connected to the vehicle body frame, the upper wishbone is connected to the lower wishbone through a connector, the center of the connector is connected to the wheel hub, and the wheel is driven by the wheel hub motor;

[0013] The protection module is two detachable anti-collision beams installed in front and behind the vehicle body frame;

[0014] The control system specifically includes a control module and an environment perception module;

[0015] The control module includes a host computer and a slave computer. The host computer adopts a Jetson Xavier NX development board and is installed above the body frame; the slave computer adopts an STM32 single-chip microcomputer and is installed inside the body frame;

[0016] The upper computer and the lower computer use USB to transmit data. The upper computer first receives data from the environment perception module, then calculates the data and makes corresponding control decisions, and then sends control instructions to the lower computer. The lower computer controls the wheel hub motor to complete the corresponding decision-making function.

[0017] The environment perception module is installed above the vehicle body frame, and specifically includes a depth vision camera, a UWB module and a multi-line laser radar; the multi-line laser radar is located behind the depth vision camera, and after the target information and obstacle information are extracted, the environment perception module sends the extracted information data to the host computer; the multi-line laser radar is set on a bracket to form a height difference with the depth vision camera;

[0018] The UWB module consists of two base stations and a tag. The base stations are installed on both sides of the depth vision camera. The tag is carried by the target to be followed. The UWB module uses the received signal angle method to obtain the target position information. The principle is that when the UWB base station receives the signal of the UWB tag, it calculates its angle to locate the UWB tag.

[0019] On the other hand, a control method of a wheeled autonomous following robot is implemented by the aforementioned wheeled autonomous following robot, and specifically comprises the following steps:

[0020] Step S1, obtaining position information of a following target and a wheeled autonomous following robot;

[0021] Step S2, according to the position information of the follow target and the wheeled autonomous follower robot, use the improved D* path planning algorithm to perform global path planning, plan a smooth path that can reach the set following distance near the target, and follow this path to reach the set following distance near the follow target, and the following distance can be set manually;

[0022] The improved D* path planning algorithm is a nine-direction search method. Specifically, the nine-direction search method establishes a coordinate system with the current position of the follower robot as the origin, discards the search direction opposite to the quadrant where the target point is located according to the positional relationship between the target point and the follower robot, and adds two search directions in the quadrant where the target point is located, and adopts a method of fusing B-spline curves for path smoothing;

[0023] Step S3: After reaching the vicinity of the follow target, the target starts to move, and the follower robot uses the improved SiamRPN visual tracking algorithm to track the target and control the wheeled follower robot to follow the target;

[0024] If the target of the visual tracking algorithm is lost during the following process, proceed to S4; if the target is not lost, proceed directly to S6;

[0025] During the following process, a multi-line laser radar is used to detect obstacles. If an obstacle appears, S5 is performed. If no obstacle appears, S6 is performed.

[0026] The improved SiamRPN visual tracking algorithm is based on the traditional SiamRPN visual tracking algorithm and introduces a template matching and updating strategy, specifically, updating the original target template on the premise of obtaining a trustworthy target;

[0027] Step S4: using the UWB module to obtain the distance and angle information of the target, ensuring that the control module obtains the position information of the target in real time, and controls the wheeled autonomous following robot to continue to follow the target;

[0028] Step S5: using a multi-line laser radar to detect obstacles, and after detecting obstacles, using an improved DWA local path planning algorithm to avoid obstacles;

[0029] The obstacle detection is specifically as follows: first, voxel filtering, boundary detection and ground plane segmentation operations are performed on the collected laser three-dimensional point cloud to obtain the region of interest and non-ground point cloud; then, the three-dimensional point cloud in the region of interest is preliminarily grouped using the K-means algorithm according to the density characteristics; the point cloud in each group is clustered in parallel using the parameter-adaptive DBSCAN algorithm; finally, the cluster clusters that meet the set conditions on the boundary of each group are merged to obtain the center position and three-dimensional size information of the obstacle, thereby completing the obstacle detection.

[0030] The improved DWA local path planning algorithm improves the evaluation function of the traditional DWA path planning algorithm and adds an obstacle angle evaluation sub-function dtheta to the evaluation function. The sub-function is the difference between the current robot motion direction and the angle between the current robot and the nearest obstacle. Its mathematical model is as follows:

[0031] dtheta(v,ω)=|arctan[(obs(y)-X(y)) / (obs(x)-X(x))]-θ|

[0032] Among them, dtheta(v, ω) is the obstacle angle evaluation subfunction; v is the linear velocity of the robot in its direction of motion; ω is the angular velocity of the robot; obs(x) and obs(y) are the horizontal and vertical coordinates of the nearest obstacle; X(x) and X(y) are the horizontal and vertical coordinates of the current position of the robot; θ is the angle between the current robot motion direction and the coordinate axis.

[0033] The evaluation function of the improved DWA algorithm is:

[0034] G(v,ω)=σ[α·heading(v,ω)+β·dist(v,ω)+γ·vel(v,ω)+δ·dtheta(v,ω)]

[0035] Among them, dtheta(v, ω) is the above-mentioned obstacle angle evaluation sub-function; heading(v, ω) is the target azimuth evaluation sub-function; dist(v, ω) is the obstacle gap evaluation sub-function; vel(v, ω) is the speed evaluation sub-function; α, β, γ, δ are the weight coefficients of each evaluation function, and σ is a constant.

[0036] Step S6: Continue to follow the target until the target reaches the end point and the task is completed.

[0037] The beneficial effects of adopting the above technical solution are:

[0038] The present invention provides a wheeled autonomous following robot and a control method thereof. Compared with the prior art, the present invention adopts a method of integrating vision and UWB modules to solve the problem that the current following robot mainly relies on vision technology, has poor effect in obtaining target information, and cannot continue to complete the following task once the camera loses the following target. At the same time, the present invention also solves the problems that the current following robot application scenarios are generally relatively simple, and it is impossible to autonomously plan the path to find the following target; the chassis has poor passability and is difficult to meet user needs in actual application.

[0039] The present invention also provides a control method for a wheeled autonomous following robot, which is implemented based on the above-mentioned wheeled autonomous following robot and solves the same technical problem.

[0040] The control method of the present invention has good portability and can meet the requirements of different hardware configurations and different tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of the overall structure of a wheeled autonomous following robot is provided for an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a chassis system of a wheeled autonomous following robot is provided for an embodiment of the present invention;

[0043] Figure 3 Provide a flow chart of a wheeled autonomous following robot performing a following task for an embodiment of the present invention;

[0044] Figure 4 Provides a target relative position and search direction selection diagram for an improved D* path planning algorithm for an embodiment of the present invention;

[0045] Figure 5 Provides a comparison chart of the D* path planning algorithm search strategy before and after improvement for the embodiment of the present invention;

[0046] Figure 6 Provide an improved SiamRPN network structure diagram for an embodiment of the present invention;

[0047] Figure 7 A schematic diagram of a UWB module measuring distance based on a received signal angle method is provided for an embodiment of the present invention;

[0048] Figure 8 A framework diagram of a light radar obstacle detection method is provided for an embodiment of the present invention.

[0049] In the figure, 1-chassis system, 2-control system, 11-body module, 111-off-road tire, 112-body frame, 12-power module, 13-powertrain, 14-suspension module, 15-protection module, 141-upper cross arm, 142-connector, 143-shock absorber, 144-lower cross arm, 21-upper computer, 22-lower computer, 23-depth vision camera, 24-UWB module, 241-base station one, 242-base station two, 243-tag, 25 multi-line laser radar. DETAILED DESCRIPTION

[0050] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0051] On the one hand, a wheeled autonomous following robot, e.g. Figure 1 , Figure 2 As shown, it includes a chassis system 1 and a control system 2;

[0052] The chassis system 1 is as follows Figure 1 As shown, it specifically includes a body module 11, a power module 12, a power assembly 13, a suspension module 14 and a protection module 15;

[0053] The vehicle body module 11 is composed of four wheels and a vehicle body frame; the wheels use off-road tires with a diameter of 300 mm; the off-road tires 111 can effectively improve the climbing and off-road capabilities of the chassis system, ensure that the tires can get good contact with the ground, so that the following robot maintains a good adhesion effect, and then ensure sufficient driving force to achieve the following functions such as forward, backward and steering;

[0054] The power module 12 is composed of a number of 48V lithium batteries connected in series and installed inside the vehicle body frame; the output end of the power module is connected to the control system and the powertrain;

[0055] The power assembly 13 selects four 500W wheel hub motors, the output ends of the four wheel hub motors are connected to the four wheels respectively, and the total power of the wheel hub motors in series is P=2kW; the autonomous following robot adopts a four-wheel differential drive, which can realize the function of turning on the spot;

[0056] The suspension module 14 adopts an upper and lower double wishbone independent suspension system, which is composed of an upper wishbone 141, a connector 142, a shock absorber 143 and a lower wishbone 144; the lower end of the shock absorber 143 is connected to the lower wishbone 144, and the upper end of the shock absorber 143 is connected to the vehicle body frame 112, so that it does not bear the lateral force, has good turning stability, small roll, and a simple structure; the upper wishbone 141 is connected to the lower wishbone 144 through the connector 142, the center of the connector is connected to the wheel hub, and the wheel is driven by the wheel hub motor; the independent suspension enables the chassis of the autonomous following robot to have a good shock absorption effect;

[0057] The protection module 15 is two detachable anti-collision beams installed in front and behind the vehicle body frame, which ensure that the autonomous following robot still has a safety anti-collision function when the system fails, and ensure the integrity of the internal components of the robot.

[0058] The function of the chassis system is to support and install the robot's power mechanism and various component assemblies to form the overall shape of the robot, and to receive driving force to make the robot move and ensure normal driving;

[0059] The control system 2 specifically includes a control module and an environment perception module;

[0060] The control module is a robot main control system designed with ROS as the core communication architecture, combined with remote software to achieve remote control of the autonomous following robot;

[0061] The control module includes a host computer 21 and a slave computer 22. The host computer adopts a Jetson Xavier NX development board and is installed above the body frame; the slave computer adopts an STM32 single-chip microcomputer and is installed inside the body frame; it is responsible for communicating with the host computer, obtaining corresponding decision instructions, converting the instructions into PWM square wave signals and sending them to the motor controller to complete motion control;

[0062] The upper computer 21 and the lower computer 22 use USB to transmit data. The upper computer 21 first receives the data from the environment perception module, then calculates the data and makes corresponding control decisions, and then sends the control instructions to the lower computer 22. The lower computer 22 controls the wheel hub motor to complete the corresponding decision-making function.

[0063] The environment perception module is responsible for extracting target information and obstacle information;

[0064] The environment perception module is installed above the vehicle body frame, and specifically includes a depth vision camera 23, a UWB module 24 and a multi-line laser radar 25; the multi-line laser radar 25 is located behind the depth vision camera 23. After the target information and obstacle information are extracted, the environment perception module sends the extracted information data to the host computer; in order to avoid scanning the front depth vision camera, the multi-line laser radar 25 is set on a bracket to form a height difference with the depth vision camera;

[0065] The UWB module 24 is composed of a base station 1 241, a base station 242 and a tag 243. The two base stations are installed on both sides of the depth vision camera; the tag 243 is carried by the follower; the UWB module 24 uses a method based on the angle of the received signal to obtain the target position information. The principle is that when the UWB base station receives the signal of the UWB tag 243, it calculates its angle to locate the UWB tag 243; Figure 7 As shown. Assume that the coordinates of base station 1 241 are (x1, y1); the coordinates of base station 242 are (x2, y2); θ1, θ2 are the angles at which the signal sent by UWB tag 243 reaches UWB base station 1 241 and base station 2 242. Extend the angle of incidence, and the focus of the two signal extension lines is the coordinate T (x, y) of the location of UWB tag 243. Solving the position of UWB tag 243, we can get:

[0066]

[0067] The control system is responsible for the path planning and tracking control functions of the wheeled autonomous following robot.

[0068] On the other hand, a control method of a wheeled autonomous following robot is implemented by the aforementioned wheeled autonomous following robot, such as Figure 3 As shown, the specific steps include:

[0069] Step S1, obtaining position information of a following target and a wheeled autonomous following robot;

[0070] Step S2, according to the position information of the follow target and the wheeled autonomous follower robot, use the improved D* path planning algorithm to perform global path planning, plan a smooth path that can reach the set following distance near the target, and follow this path to reach the set following distance near the follow target, and the following distance can be set manually;

[0071] The improved D* path planning algorithm aims to solve the problems of too many turning points, a large number of search nodes, and low algorithm operation efficiency in the path obtained by the traditional D* algorithm using the 8-neighborhood search. The 8-neighborhood search method of the traditional D* algorithm is changed to a 9-direction search method. Specifically, the 9-direction search method establishes a coordinate system with the current position of the follower robot as the origin, discards the search direction opposite to the quadrant where the target point is located according to the positional relationship between the target point and the follower robot, and adds two search directions in the quadrant where the target point is located to improve the path quality and algorithm operation efficiency. In order to solve the problem of sharp turning points in the planned path, the method of integrating B-spline curves is adopted to increase the smoothness of the path. The search direction selection is as follows: Figure 4 , Figure 5 shown.

[0072] Step S3: After reaching the vicinity of the follow target, the target starts to move, and the follower robot uses the improved SiamRPN visual tracking algorithm to track the target and control the wheeled follower robot to follow the target;

[0073] If the target of the visual tracking algorithm is lost during the following process, proceed to S4; if the target is not lost, proceed directly to S6;

[0074] During the following process, the multi-line laser radar 25 is used to detect obstacles. If an obstacle appears, S5 is performed; if no obstacle appears, S6 is performed;

[0075] The improved SiamRPN visual tracking algorithm is based on the traditional SiamRPN visual tracking algorithm, and proposes an improved SiamRPN algorithm based on template matching and updating strategy; Figure 6 As shown in the figure, a template matching and updating strategy is introduced. Specifically, the original target template is updated on the premise of obtaining a trustworthy target. This can effectively cope with the continuous tracking task when the target posture changes. It also has good positioning accuracy and tracking success rate for targets in low resolution and high motion blur states, and its robustness is greatly improved compared to the original algorithm.

[0076] Step S4, using the UWB module 24 to obtain the distance and angle information of the target, ensuring that the control module obtains the position information of the target in real time, ensuring that the visual camera still maintains accurate positioning and tracking after losing the target, and controlling the wheeled autonomous following robot to continue to follow the target;

[0077] Step S5, using the multi-line laser radar 25 to detect obstacles, and after detecting obstacles, using the improved DWA local path planning algorithm to avoid obstacles;

[0078] The obstacle detection is specifically as follows: Figure 8As shown in the figure, the collected laser 3D point cloud is first subjected to voxel filtering, boundary detection and ground plane segmentation operations to obtain the region of interest and non-ground point cloud; then, the 3D point cloud in the region of interest is preliminarily grouped using the K-means algorithm according to its density characteristics; the point cloud in each group is clustered in parallel using the parameter-adaptive DBSCAN algorithm; finally, the clusters that meet the set conditions on the boundary of each group are merged to obtain the center position and 3D size information of the obstacle, thereby completing obstacle detection.

[0079] The improved DWA local path planning algorithm aims at the problem that the traditional DWA path planning algorithm sometimes falls into the local optimal solution, resulting in the inability to find the optimal path or even the robot being unable to move forward. The evaluation function of the traditional DWA path planning algorithm is improved, and an obstacle angle evaluation sub-function dtheta is added to the evaluation function. The sub-function is the difference between the current robot movement direction and the angle between the current robot and the nearest obstacle, which is used to prevent the robot from falling into the local optimal problem and improve the path quality. Its mathematical model is as follows:

[0080] dtheta(v,ω)=|arctan[(obs(y)-X(y)) / (obs(x)-X(x))]-θ|

[0081] Among them, dtheta(v, ω) is the obstacle angle evaluation subfunction; v is the linear velocity of the robot in its direction of motion; ω is the angular velocity of the robot; obs(x) and obs(y) are the horizontal and vertical coordinates of the nearest obstacle; X(x) and X(y) are the horizontal and vertical coordinates of the current position of the robot; θ is the angle between the current robot motion direction and the coordinate axis.

[0082] The evaluation function of the improved DWA algorithm is:

[0083] G(v,ω)=σ[α·heading(v,ω)+β·dist(v,ω)+γ·vel(v,ω)+δ·dtheta(v,ω)]

[0084] Among them, dtheta(v, ω) is the above-mentioned obstacle angle evaluation sub-function; heading(v, ω) is the target azimuth evaluation sub-function; dist(v, ω) is the obstacle gap evaluation sub-function; vel(v, ω) is the speed evaluation sub-function; α, β, γ, δ are the weight coefficients of each evaluation function, and σ is a constant.

[0085] The improved DWA path planning algorithm has good results in dynamic environments.

[0086] Step S6: Continue to follow the target until the target reaches the end point and the task is completed.

[0087] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution.

Claims

1. A wheeled autonomous following robot, characterized in that: Including chassis system and control system; The chassis system specifically includes a body module, a power module, a powertrain, a suspension module and a protection module; The vehicle body module is composed of four wheels and a vehicle body frame; the power module is composed of a number of lithium batteries connected in series and installed inside the vehicle body frame; the output end of the power module is connected to the control system and the powertrain; the powertrain selects four 500W wheel hub motors, the output ends of the four wheel hub motors are connected to four wheels respectively, and the total power of the wheel hub motors in series is P=2kW; the autonomous following robot adopts a four-wheel differential drive; the suspension module adopts an upper and lower double wishbone independent suspension system; the protection module is two detachable anti-collision beams installed in front and behind the vehicle body frame; The control system specifically includes a control module and an environment perception module; the control module includes a host computer and a slave computer, the host computer adopts a Jetson Xavier NX development board and is installed above the body frame; the slave computer adopts an STM32 single-chip microcomputer and is installed inside the body frame; the environment perception module is installed above the body frame.

2. A wheeled autonomous following robot according to claim 1, characterized in that: The wheels use off-road tires with a diameter of 300 mm.

3. A wheeled autonomous following robot according to claim 1, characterized in that: The suspension module consists of an upper cross arm, a connector, a shock absorber and a lower cross arm; the lower end of the shock absorber is connected to the lower cross arm, the upper end of the shock absorber is connected to the body frame, the upper cross arm is connected to the lower cross arm through a connector, the center of the connector is connected to the wheel hub, and the wheel is driven by the hub motor.

4. The wheeled autonomous following robot according to claim 1, characterized in that: The upper computer and the lower computer use USB to transmit data. The upper computer first receives data from the environment perception module, then calculates the data and makes corresponding control decisions, and then sends control instructions to the lower computer. The lower computer controls the wheel hub motor to complete the corresponding decision-making function.

5. The wheeled autonomous following robot according to claim 1, characterized in that: The environment perception module specifically includes a depth vision camera, a UWB module and a multi-line laser radar; the multi-line laser radar is located behind the depth vision camera, and after the target information and obstacle information are extracted, the environment perception module sends the extracted information data to the host computer; the multi-line laser radar is arranged on a bracket to form a height difference with the depth vision camera; The UWB module consists of two base stations and a tag. The base stations are installed on both sides of the depth vision camera; the tag is carried by the tracking target; The UWB module distance measurement adopts the method based on the received signal angle to obtain the target position information. The principle is that when the UWB base station receives the signal of the UWB tag, it calculates its angle to locate the UWB tag.

6. A control method for a wheeled autonomous following robot, implemented based on the wheeled autonomous following robot according to claim 1, characterized in that: The following steps are involved: Step S1, obtaining position information of a following target and a wheeled autonomous following robot; Step S2, according to the position information of the follow target and the wheeled autonomous follower robot, use the improved D* path planning algorithm to perform global path planning, plan a smooth path that can reach the set following distance near the target, and follow this path to reach the set following distance near the follow target, and the following distance can be set manually; Step S3: After reaching the vicinity of the follow target, the target starts to move, and the follower robot uses the improved SiamRPN visual tracking algorithm to track the target and control the wheeled follower robot to follow the target; If the target of the visual tracking algorithm is lost during the following process, proceed to S4; if the target is not lost, proceed directly to S6; During the following process, a multi-line laser radar is used to detect obstacles. If an obstacle appears, S5 is performed; if no obstacle appears, S6 is performed; The improved SiamRPN visual tracking algorithm is based on the traditional SiamRPN visual tracking algorithm and introduces a template matching and updating strategy, specifically, updating the original target template on the premise of obtaining a trustworthy target; Step S4: using the UWB module to obtain the distance and angle information of the target, ensuring that the control module obtains the position information of the target in real time, and controls the wheeled autonomous following robot to continue to follow the target; Step S5: using a multi-line laser radar to detect obstacles, and after detecting obstacles, using an improved DWA local path planning algorithm to avoid obstacles; The obstacle detection is specifically as follows: first, voxel filtering, boundary detection and ground plane segmentation operations are performed on the collected laser 3D point cloud to obtain the region of interest and non-ground point cloud; then, the 3D point cloud in the region of interest is preliminarily grouped using the K-means algorithm according to the density characteristics; the point cloud in each group is clustered in parallel using the parameter adaptive DBSCAN algorithm; finally, the cluster clusters that meet the set conditions on the boundary of each group are merged to obtain the center position and 3D size information of the obstacle, thereby completing the obstacle detection; Step S6: Continue to follow the target until the target reaches the end point and the task is completed.

7. The control method of a wheeled autonomous following robot according to claim 6, characterized in that: The improved D* path planning algorithm described in step S2 is a nine-way search method. The nine-way search method specifically establishes a coordinate system with the current position of the follower robot as the origin, discards the search direction opposite to the quadrant where the target point is located according to the positional relationship between the target point and the follower robot, adds two search directions in the quadrant where the target point is located, and adopts the method of fusing B-spline curves to smooth the path.

8. The control method of a wheeled autonomous following robot according to claim 6, characterized in that: The improved DWA local path planning algorithm described in step S5 improves the evaluation function of the traditional DWA path planning algorithm, and adds an obstacle angle evaluation sub-function dtheta to the evaluation function. The sub-function is the difference between the current robot motion direction and the angle between the current robot and the nearest obstacle. Its mathematical model is as follows: dtheta(v,ω)=|arctan[(obs(y)-X(y)) / (obs(x)-X(x))]-θ| Among them, dtheta(v,ω) is the obstacle angle evaluation subfunction; v is the linear velocity of the robot in its direction of motion; ω is the angular velocity of the robot; obs(x) and obs(y) are the horizontal and vertical coordinates of the nearest obstacle; X(x) and X(y) are the horizontal and vertical coordinates of the current position of the robot; θ is the angle between the current robot motion direction and the coordinate axis; The evaluation function of the improved DWA algorithm is: G(v,ω)=σ[a·heading(v,ω)+β·dist(v,ω)+γ·vel(v,ω)+δ·dtheta(v,ω)] Among them, dtheta(v, ω) is the above-mentioned obstacle angle evaluation sub-function; heading(v, ω) is the target azimuth evaluation sub-function; dist(v, ω) is the obstacle gap evaluation sub-function; vel(v, ω) is the speed evaluation sub-function; α, β, γ, δ are the weight coefficients of each evaluation function, and σ is a constant.