Track inspection robot based on SLAM algorithm

Abstract

The invention discloses a track inspection robot based on an SLAM algorithm. Relates to the technical field of inspection robots, and comprises an inspection vehicle module, an upper computer module and a network communication module, the inspection vehicle module and the upper computer module are in communication connection through the network communication module, and the upper computer module is used for issuing an inspection task, checking the state of the inspection vehicle and displaying an inspection map; and the inspection vehicle module receives the inspection task issued by the upper computer module and plans an inspection path based on an SLAM algorithm. According to the track inspection robot, through the multi-level software system architecture of the central controller and the upper computer, the intelligent level of track inspection is comprehensively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of inspection robots, and more particularly to a track inspection robot based on SLAM algorithm. Background Art

[0002] With the rapid development of railway transportation, traditional track inspection methods face challenges such as low efficiency, high costs, and high labor intensity, especially in inclement weather conditions. Rail maintenance has traditionally relied primarily on manual inspections. Workers, armed with specialized tools, patrol the tracks on foot, often failing to detect and address potential track faults in a timely manner. As track mileage continues to increase, the limitations of this traditional inspection method become increasingly apparent.

[0003] While some track inspection equipment has been developed domestically and internationally, these devices have a low level of automation and are often unable to adapt to diverse track gauges and complex track environments. For example, traditional track inspection vehicles often require frequent stops to set up stations, resulting in limited measurement efficiency and vulnerability to inclement weather. Furthermore, these devices lack obstacle detection and inspection flexibility, making it impossible to adjust inspection strategies in real time to accommodate unexpected situations.

[0004] While advanced inspection equipment from abroad has achieved some technological breakthroughs, such as the introduction of image measurement and intelligent obstacle avoidance, its overall cost is high, and widespread adoption is difficult. Despite years of development, most inspection technologies remain largely unmechanized and intelligent, resulting in low inspection efficiency, high labor intensity, and limited application. Therefore, designing a SLAM-based rail inspection robot to enhance the level of intelligent inspection capabilities is a pressing issue for those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a track inspection robot based on the SLAM algorithm, which comprehensively improves the intelligence level of track inspection through the multi-level software system architecture of the central controller and the host computer.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a track inspection robot based on the SLAM algorithm, including: an inspection vehicle module, a host computer module and a network communication module. The inspection vehicle module and the host computer module are connected for communication through the network communication module. The host computer module is used to issue inspection tasks, check the status of the inspection vehicle and display the inspection map; the inspection vehicle module receives the inspection tasks issued by the host computer module, and plans the optimal inspection path based on the SLAM algorithm.

[0007] Preferably, the inspection vehicle module includes a scanning detection system, a path navigation and control system, and a chassis motion system;

[0008] The scanning detection system is used to obtain the motion path and the surrounding environment information of the motion path, and transmit the motion path and the surrounding environment information of the motion path to the path navigation and control system;

[0009] The path navigation and control system performs path planning and decision-making based on the motion path and the surrounding environment information of the motion path and the inspection tasks issued by the host computer, and generates specific motion instructions;

[0010] The chassis motion system is used to receive and execute motion instructions to drive the inspection vehicle module to move along the planned path.

[0011] Preferably, the path navigation and control system includes a SLAM synchronous positioning and map construction algorithm model module, a path planning module, an operation control model module, a dynamic obstacle avoidance module, a data fusion module, a communication module and an intelligent control module;

[0012] The SLAM synchronous positioning and mapping algorithm model module uses the ORB-SLAM3 algorithm to perform real-time map construction and positioning based on the surrounding environment information to obtain basic information of the motion trajectory;

[0013] The path planning module uses the A* algorithm and the Dijkstra algorithm to perform path planning based on the basic information of the motion trajectory to determine the driving path;

[0014] The operation control model module uses the PID operation control model to control the robot to track the driving path;

[0015] The dynamic obstacle avoidance module detects the position of obstacles in real time through sensors and dynamically adjusts the driving path in combination with path planning;

[0016] The data fusion module fuses the various data from the above modules to improve the accuracy and reliability of sensor detection and optimize the accuracy of positioning and map construction;

[0017] The communication module is used to receive the inspection tasks issued by the host computer module and the instruction information of each module in the path navigation and control system, and transmit them to the intelligent control module;

[0018] The intelligent control module is used to integrate the command information of each module in the path navigation and control system, perform comprehensive analysis and decision-making, and output comprehensive control instructions to the chassis motion system, while feeding back the control status to the host computer module through the communication module.

[0019] Preferably, the ORB-SLAM3 algorithm performs real-time map construction and positioning based on surrounding environment information, and the formula for its feature extraction and matching method is as follows:

[0020] f(x)=ORB(I t )

[0021] Among them, I t is the input image at the current time t; f(x) is the feature point matrix extracted from the image;

[0022] The pose estimation formula is as follows:

[0023] x t =PnP(f(x t-1 ),f(x t ))

[0024] Among them, x t-1 ,x t are the previous state and current state of the robot respectively;

[0025] The map update formula is as follows:

[0026]

[0027] Among them, M t The map at the current time consists of multiple key frames f k The feature points of ∪ represent the integration of feature points.

[0028] Preferably, the A* algorithm is used to plan the optimal driving path, and the Dijkstra algorithm is used to plan the shortest driving path; wherein the cost function for planning the optimal driving path is as follows:

[0029] f(n)=g(n)+h(n)

[0030] Where f(n) is the total cost from the starting point to the target location through node n, including the path taken and the estimated path; g(n) is the actual cost from the starting point to node n, calculated based on the terrain or known map path; h(n) is the heuristic estimate from node n to the target location;

[0031] The formula for calculating the shortest driving path cost is as follows:

[0032] f′(n)=g′(n);

[0033] Where f′(n) represents the total cost from the starting point to node n; g′(n) represents the actual cumulative cost from the starting point to node n;

[0034] The path planning module generates the optimal path for the inspection vehicle to travel on the track according to the calculation results of the formula.

[0035] Preferably, the PID operation control model calculation formula is as follows:

[0036]

[0037] Where e(t) is the error at the current moment, K p , K i , K d are the proportional coefficient, integral coefficient, and differential coefficient.

[0038] Preferably, the dynamic obstacle avoidance module uses an obstacle avoidance algorithm when an obstacle is detected, and its obstacle avoidance formula is as follows:

[0039] if d<d threshold ,then adjust direction toθ

[0040] Wherein, d represents the distance to the obstacle, θ represents the adjusted direction, and the movement speed and driving direction are adjusted in real time in combination with the detection results of the scanning detection system to achieve obstacle avoidance.

[0041] Preferably, the data fusion module adopts a data fusion algorithm, the formula is as follows:

[0042] X fusion =X predicted +K·(Z-HX predicted )

[0043] Among them, X fusion It combines the estimated and actually observed sensor information to provide real-time correction; K is the Kalman gain, which controls the correction amplitude; Z is the actual measurement data; H is the observation model, which is used to map the prediction to the actual observation and is used to fuse the detection data from the scanning detection system.

[0044] Preferably, the chassis motion system has an adjustable gauge structure, including symmetrically arranged traveling seat units, and a gauge adjustment screw drive module is arranged in the fixed-side traveling seat unit.

[0045] Preferably, the scanning detection system is a three-dimensional laser scanning device, comprising a high-strength housing and an internal integrated equipment cabin; the scanning unit and the control unit are arranged in layers in the internal integrated equipment cabin;

[0046] The scanning unit includes laser radar, industrial camera, and infrared camera;

[0047] The control unit includes a data router and controller and a system power supply;

[0048] Each unit is fixed by a modular bracket. The bracket body adopts a connection column and distributed connection point design and is connected to the inspection vehicle module through a damping hinge.

[0049] It can be seen from the above technical solution that, compared with the prior art, the present invention discloses a track inspection robot based on the SLAM algorithm, which has the following beneficial effects:

[0050] 1. Control System Integration: This invention utilizes a multi-layered software system architecture comprised of a central controller and a host computer, comprehensively enhancing the intelligent capabilities of rail inspection. The central controller is responsible for core tasks such as chassis motion control, camera photography, laser ranging, and communications, while the host computer provides a user interface for issuing inspection tasks, viewing vehicle status, and displaying inspection maps. This architecture enables vehicles to efficiently perform inspection tasks and enable real-time interaction with users.

[0051] 2. Robot Operating System Application: Utilizing the ROS framework improves system versatility and reusability while reducing development costs. ROS's open source nature and broad community support provide flexibility and scalability for implementing robotic functions, making it easier to integrate new technologies and features into the system.

[0052] 3. Precision Positioning Module: This module analyzes LiDAR and ultra-wideband (UWB) data and applies filtering algorithms to optimize positioning information, ensuring the accuracy and stability of the robot's positioning. The ROS ACML package further improves the reliability of positioning information, enabling the inspection vehicle to maintain accurate positioning in complex environments.

[0053] 4. Advanced Navigation Module: Autonomous navigation is achieved using SLAM technology, which utilizes cameras and depth sensors for environmental exploration and optical processing. Using neural networks and statistical methods, the robot performs real-time path planning, supporting the coordinated coordination of manual control, local path planning, and global path planning. This navigation module enables the inspection vehicle to flexibly adapt to dynamic environmental changes and achieve efficient route planning.

[0054] 5. Multifunctional Inspection Module: Upon receiving an inspection mission, the inspection vehicle automatically plans its inspection route using a multi-sensor system. During the inspection process, the vehicle can identify horizontal obstacles and high-risk equipment in real time and adjust the camera angle based on the mission requirements to ensure optimal capture. This module can perform complex inspection tasks, such as checking for branches approaching conductors, tilted poles, broken conductors, lightning conductors, and grounding wires, as well as measuring current and voltage at key locations such as stations, signal towers, and contact boxes, thereby improving inspection accuracy and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0056] Figure 1A schematic block diagram of the structure of a track inspection robot based on the SLAM algorithm provided by the present invention;

[0057] Figure 2 This is a structural block diagram of the path navigation and control system provided by the present invention;

[0058] Figure 3 A structural block diagram of the scanning control system provided by the present invention;

[0059] Figure 4 A schematic diagram of the structure of the scanning control system provided by the present invention;

[0060] Figure 5 A structural block diagram of the chassis motion system provided by the present invention;

[0061] Figure 6 A schematic structural diagram of the chassis motion system provided by the present invention;

[0062] Among them, 1 is the sensing device, 101 is the laser radar, 102 is the industrial camera, 103 is the infrared camera, 104 is the ultrasonic sensor, 105 is the inertial measurement unit, 106 is the positioning sensor, 107 is the data router and controller; 2 is the path planning module; 3 is the scanning control system, 301 is the built-in power supply, 302 is the cooling fan, 303 is the back cover, 304 is the start button, 305 is the power socket, and 306 is the connecting column; 4 is the SLAM synchronous positioning and map construction algorithm model module; 5 is the path navigation and control system; 6 is the track Road inspection robot; 7 is the host computer module; 8 is the network communication module; 9 is the inspection vehicle module, 901 is the trapezoidal screw, 902 is the optical axis, 903 is the travel component, 904 is the track wheel motor, 905 is the protective cover, 906 is the base platform, 907 is the tube body clamping seat, 908 is the track wheel, 909 is the lead screw motor, and 910 is the rib; 10 is the intelligent control module, 1001 is the dynamic obstacle avoidance module, and 1002 is the data fusion module; 11 is the system power supply; 12 is the chassis motion system, 1201 is the dual wheel side motor, and 1202 is the adjustable distance chassis. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0064] See also Figure 1As shown, an embodiment of the present invention discloses a track inspection robot based on the SLAM algorithm, including: an inspection vehicle module, a host computer module and a network communication module. The inspection vehicle module and the host computer module are connected for communication via the network communication module. The host computer module is used to issue inspection tasks, check the status of the inspection vehicle and display the inspection map; the inspection vehicle module receives the inspection tasks issued by the host computer module and plans the optimal inspection path based on the SLAM algorithm.

[0065] Specifically, the inspection vehicle module includes a scanning detection system, a path navigation and control system, and a chassis motion system;

[0066] The scanning detection system is used to obtain the motion path and the surrounding environment information of the motion path, and transmit the motion path and the surrounding environment information of the motion path to the path navigation and control system;

[0067] The path navigation and control system performs path planning and decision-making based on the motion path and the surrounding environment information of the motion path and the inspection tasks issued by the host computer, and generates specific motion instructions;

[0068] The chassis motion system is used to receive and execute motion instructions to drive the inspection vehicle module to move along the planned path.

[0069] See also Figure 2 As shown, specifically, the path navigation and control system includes a SLAM synchronous positioning and map construction algorithm model module, a path planning module, an operation control model module, a dynamic obstacle avoidance module, a data fusion module, a communication module and an intelligent control module;

[0070] The SLAM synchronous positioning and mapping algorithm model module uses the ORB-SLAM3 algorithm to perform real-time map construction and positioning based on the surrounding environment information to obtain basic information of the motion trajectory;

[0071] The path planning module uses the A* algorithm and the Dijkstra algorithm to perform path planning based on the basic information of the motion trajectory to determine the driving path;

[0072] The operation control model module uses the PID operation control model to control the robot to track the driving path;

[0073] The dynamic obstacle avoidance module detects the position of obstacles in real time through sensors and dynamically adjusts the driving path in combination with path planning;

[0074] The data fusion module fuses the various data from the above modules to improve the accuracy and reliability of sensor detection and optimize the accuracy of positioning and map construction;

[0075] The communication module is used to receive the inspection tasks issued by the host computer module and the instruction information of each module in the path navigation and control system, and transmit them to the intelligent control module;

[0076] The intelligent control module is used to integrate the command information of each module in the path navigation and control system, perform comprehensive analysis and decision-making, and output comprehensive control instructions to the chassis motion system, while feeding back the control status to the host computer module through the communication module.

[0077] In a specific embodiment of the present invention, the host computer module 7 is used in conjunction with the inspection vehicle module 9, the network communication module 8 is used for ROS+SSH communication, and the host computer module 7 is used for the interface display of the SLAM synchronous positioning and map construction algorithm model module 4, issuing inspection tasks, checking the inspection vehicle status and displaying the inspection map; when the inspection vehicle module 9 is running on a normal road section, the path navigation and control system 5 scans and processes the driving path of the inspection vehicle module 9, and plans the driving path and speed; according to the optimal driving path provided by the scanning control system 3, the path navigation and control system 5 controls the inspection vehicle module 9 to track the optimal driving path.

[0078] Further, such as Figure 3-4 As shown, the scanning control system 3 includes a sensing device 1, a built-in power supply 301, a cooling fan 302, a back cover 303, a start button 304, a power socket 305, and a connecting column 306; the sensing device 1 includes: a laser radar 101, an industrial camera 102, an infrared camera 103, an ultrasonic sensor 104, an inertial measurement unit 105, a positioning sensor 106, a data router and a controller 107; a SLAM synchronous positioning and mapping algorithm model module 4 is used to perform real-time mapping and positioning of the above-mentioned environmental information, plan the optimal driving path and the shortest driving path, and perform specific control through the intelligent control module 10;

[0079] Further, such as Figure 5-6As shown, the chassis motion system 12 is used to provide driving support and power transmission for the inspection vehicle module 9; the chassis motion system 12 includes a dual wheel motor 1201, an adjustable distance chassis 1202, a trapezoidal screw 901, an optical axis 902, a travel component 903, a track wheel motor 904, a protective cover 905, a base platform 906, a tube clamping seat 907, a track wheel 908, a screw motor 909 and a rib 910; the dual wheel motor 1201 drives the left and right track wheels 908 respectively, and the track wheel motor 904 and the dual wheel motor 1201 cooperate to provide the robot 6 with forward, braking and steering energy force; the adjustable-distance chassis 1202 drives the trapezoidal screw 901 through the screw motor 909 to adjust the track width to adapt to different track specifications; the optical axis 902 and the travel component 903 provide precise guidance for the chassis; the ribs 910 and the protective cover 905 are installed on the base platform 906 to enhance the structural rigidity and protect the internal components; the tube clamping seat 907 can firmly clamp the guide rail or tube body during operation to improve driving stability; the chassis motion system 12 drives the inspection vehicle 9 to move along the track according to the optimal driving path and the shortest driving path, realizing the map construction and autonomous navigation functions based on the SLAM algorithm.

[0080] In one embodiment of the present invention, the scanning control system 3 integrates environmental perception and data preprocessing functions. A laser radar 101 acquires high-precision point clouds through pulse ranging; an industrial camera 102 and an infrared camera 103 collect visible light and infrared images, respectively; an ultrasonic sensor 104 detects obstacles at close range; an inertial measurement unit 105 provides attitude information; a positioning sensor 106 outputs odometry or absolute position; and a data router and controller 107 aggregates and preprocesses the sensor data. The system hardware includes a built-in power supply 301, a cooling fan 302, a rear cover 303, a start button 304, a power socket 305, and connecting posts 306 to ensure power supply, heat dissipation, and securement of the device.

[0081] Furthermore, the laser radar 101 realizes three-dimensional modeling of the target object through pulse laser ranging, providing high-precision spatial data. The laser ranging adopts the following form: Where d is the target distance; c is the speed of light; and t is the laser round-trip time.

[0082] Furthermore, the industrial camera 102 is used to synchronously capture surface texture information, ensuring image quality under varying lighting conditions. Image acquisition is performed using the following format: I = f(x, y), where I represents the captured image and x, y represent pixel coordinates. The combination of the industrial camera 102 and the LiDAR 101 enables high-precision modeling of the track environment. Image quality is ensured under varying lighting conditions by optimizing the resolution and exposure time of the industrial camera 102.

[0083] Furthermore, the infrared camera 103 provides auxiliary imaging function in low light environment, enhancing the perception ability under poor light conditions. Infrared imaging adopts the following forms: Where T is the target temperature; ε is the target surface emissivity; σ is the Stefan-Boltzmann constant; A is the target surface area; and T0 is the background temperature.

[0084] Specifically, the ORB-SLAM3 algorithm performs real-time map construction and positioning based on surrounding environment information. The formula for its feature extraction and matching method is as follows:

[0085] f(x)=ORB(I t )

[0086] Among them, I t is the input image at the current time t; f(x) is the feature point matrix extracted from the image;

[0087] The pose estimation formula is as follows:

[0088] x t =PnP(f(x t-1 ),f(x t ))

[0089] Among them, x t-1 ,x t are the previous state and current state of the robot respectively;

[0090] The map update formula is as follows:

[0091]

[0092] Among them, M t The map at the current time consists of multiple key frames f k The feature points of ∪ represent the integration of feature points.

[0093] In a specific embodiment of the present invention, the path navigation and control system 5 is the core module for realizing autonomous inspection of the robot. The SLAM synchronous positioning and mapping algorithm model module 4 combines the ORB-SLAM3 algorithm for positioning and mapping; ORB-SLAM3 is an advanced visual SLAM algorithm specially designed to handle real-time positioning and mapping problems in complex dynamic environments. The algorithm captures images through a camera and extracts ORB feature points for matching, thereby achieving accurate estimation of the robot's posture and continuous updating of the environment map; the ORB-SLAM3 algorithm includes three main steps: feature extraction, posture estimation, and map construction. First, the image captured by the camera is input into the feature extraction module to extract fast and rotation-invariant ORB feature points. Then, these feature points are used to match between images, and the robot's posture is calculated using the PnP algorithm. Finally, based on the matched feature points, ORB-SLAM3 continuously updates the environment map and generates a three-dimensional map composed of multiple key frames; during the path navigation process, the ORB-SLAM3 algorithm receives the map and posture information output by the SLAM synchronous positioning and mapping algorithm model module 4 through real-time updated map information, calls the A* algorithm and Dijkstra algorithm in the path planning module 2, and optimizes the navigation path of the robot from the starting point to the target position. When an obstacle is detected, the obstacle avoidance algorithm 1001 in the intelligent control module 10 is triggered to dynamically adjust the current path or speed to achieve obstacle avoidance.

[0094] Specifically, the A* algorithm is used to plan the optimal driving path, and the Dijkstra algorithm is used to plan the shortest driving path. The cost function for planning the optimal driving path is as follows:

[0095] f(n)=g(n)+h(n)

[0096] Where f(n) is the total cost from the starting point to the target location through node n, including the path taken and the estimated path; g(n) is the actual cost from the starting point to node n, calculated based on the terrain or known map path; h(n) is the heuristic estimate from node n to the target location;

[0097] The formula for calculating the shortest driving path cost is as follows:

[0098] f′(n)=g′(n);

[0099] Where f′(n) represents the total cost from the starting point to node n; g′(n) represents the actual cumulative cost from the starting point to node n;

[0100] The path planning module generates the optimal path for the inspection vehicle to travel on the track according to the calculation results of the formula.

[0101] In one embodiment of the present invention, the path navigation module 2 includes the A* algorithm and the Dijkstra algorithm. The A* algorithm is suitable for optimal path search in dynamic environments, while the Dijkstra algorithm is suitable for shortest path planning in static environments. The obstacle observation and prediction module uses multiple input sensing devices 1 to detect external environmental information in real time and feeds obstacle location and status information back to the path planning module to ensure the safety and efficiency of path planning.

[0102] Specifically, the calculation formula of the PID operation control model is as follows:

[0103]

[0104] Where e(t) is the error at the current moment, K p , K i , K d are the proportional coefficient, integral coefficient, and differential coefficient.

[0105] Furthermore, u(t) in the PID control model is the controller output, or the controlled variable. In a real-world control system, it is the signal that acts on the controlled object (such as a motor or valve). It is the overall control command calculated by the PID controller based on the current error e(t) and its integral and differential terms. This output signal is sent to the actuator to ensure that the system's actual output is as close to the desired value as possible.

[0106] Specifically, when the dynamic obstacle avoidance module detects an obstacle, it uses an obstacle avoidance algorithm, and its obstacle avoidance formula is as follows:

[0107] if d<d threshold ,then adjust direction toθ

[0108] Wherein, d represents the distance to the obstacle, θ represents the adjusted direction, and the movement speed and driving direction are adjusted in real time in combination with the detection results of the scanning detection system to achieve obstacle avoidance.

[0109] In a specific embodiment of the present invention, the intelligent control system 10 includes an obstacle avoidance algorithm 1001, which identifies the location and boundaries of obstacles by real-time analysis of sensor data such as the laser radar 101 and the ultrasonic sensor 104.

[0110] Furthermore, the intelligent control module 10 also includes a data fusion module 1002: based on Kalman filtering or extended Kalman filtering, the inertial measurement unit 105, the positioning sensor 106 and the odometer data (when using a camera (such as ORB-SLAM), the motion trajectory is estimated by matching the feature points of continuous frame images to obtain visual odometer data) to improve the accuracy of posture and speed estimation; during driving, the dual wheel motor 1201 and the track wheel motor 904 are closed-loop adjusted through the PID control model to ensure that the track inspection robot 6 strictly tracks the planned trajectory and maintains stable operation. The path navigation and control system 5 realizes motion control and navigation functions through real-time data fusion. ORB-SLAM combines sensors such as lidar and performs data fusion through the Kalman filter to improve the accuracy of sensor information.

[0111] Specifically, the data fusion module adopts a data fusion algorithm, and the formula is as follows:

[0112] X fusion =X predicted +K·(Z-HX predicted )

[0113] Among them, X fusion It combines the estimated and actually observed sensor information to provide real-time correction; K is the Kalman gain, which controls the correction amplitude; Z is the actual measurement data; H is the observation model, which is used to map the prediction to the actual observation and is used to fuse the detection data from the scanning detection system.

[0114] Specifically, the chassis motion system has an adjustable gauge structure, including symmetrically arranged traveling seat units, and a gauge adjustment screw drive module is arranged in the fixed-side traveling seat unit.

[0115] Furthermore, the traveling seat unit that cooperates with the track is the main seat of the traveling component 903.

[0116] Traveling components 903 cooperating with the track are provided at both ends of the optical axis 902. The traveling component 903 on one side is fixed to one end of the optical axis 902, and the traveling component 903 and the optical axis 902 are movably connected on the other side. The traveling component 903 fixed to the optical axis 902 is the fixed side.

[0117] Specifically, the scanning detection system is a three-dimensional laser scanning device, including a high-strength shell and an internal integrated equipment cabin; the internal integrated equipment cabin has a scanning unit and a control unit arranged in layers;

[0118] The scanning unit includes laser radar, industrial camera, and infrared camera;

[0119] The control unit includes a data router and controller and a system power supply;

[0120] Each unit is fixed by a modular bracket. The bracket body adopts a connection column and distributed connection point design and is connected to the inspection vehicle module through a damping hinge.

[0121] In one embodiment of the present invention, the travel unit utilizes a dual-axis support structure at the end, with a powered roller mounted on the main shaft and an adjustable auxiliary wheel set on the auxiliary shaft. The power system is housed within the travel unit's housing, and a multi-stage transmission mechanism achieves synchronous drive of the dual-side wheels. The motion of the powered roller follows the formula F = T·R, where F is the roller driving force, T is the torque of the transmission mechanism, and R is the roller radius.

[0122] Furthermore, a screw drive module is integrated within the fixed travel seat, transmitting motor power to the pitch-adjusting screw via a synchronous pulley mechanism. The screw's threaded segments form a precision helical pair with the movable base within the movable travel seat, achieving millimeter-level pitch adjustment accuracy. The kinematic relationship for the screw drive is as follows: x = n·p, where x is the linear displacement of the screw, n is the number of screw revolutions, and p is the pitch.

[0123] Furthermore, a polished rod positioning assembly is provided at the bottom of the connection platform, comprising a connection seat structure with a built-in guide shaft. The guide shaft adopts a differentiated connection method, with one end rigidly connected to the fixed side travel seat body and the other end dynamically connected to the movable side travel seat body via a sliding bearing. The polished rod positioning adopts the following form: Where θ is the deflection angle of the guide shaft; F is the external force; and k is the rigidity coefficient of the shaft.

[0124] In summary, the present invention integrates robotics, visual measurement, sensor technology and embedded system technology to design a track inspection robot based on the SLAM algorithm. Through the organic combination of the chassis motion system 12, the scanning control system 3, the SLAM synchronous positioning and map construction algorithm model module 4, the path planning module 2, the path navigation and control system 5, the intelligent control module 10, the network communication module 8 and the host computer module 7, the track operation path of the inspection vehicle module 9 is planned by the A* algorithm and Dijkstra algorithm of the SLAM synchronous positioning and map construction algorithm model module 4 and the path navigation and control system 5. The planned path includes the optimal driving path algorithm and the shortest driving path algorithm, which can realize the coordinated cooperation of manual control, local path planning and global path planning. It realizes high-precision perception of the track environment, real-time positioning and map construction, efficient path planning and autonomous driving, dynamic obstacle avoidance and intelligent control, as well as remote monitoring and management, and has the beneficial effect of stably and reliably completing automated inspection tasks under complex track conditions.

[0125] The beneficial effects of the present invention are described in detail in conjunction with the above technical solutions, as follows:

[0126] (1) The rail inspection robot of the present invention comprehensively enhances the intelligent level of rail inspection through a multi-layered software system architecture consisting of a central controller and a host computer. The central controller is responsible for core tasks such as chassis motion control, camera photography, laser ranging, and communication, while the host computer provides a user interface for issuing inspection tasks, checking inspection vehicle status, and displaying operation and inspection maps.

[0127] (2) The present invention utilizes the ROS framework to improve the versatility and reusability of the system while reducing development costs. The open source nature and extensive community support of ROS provide flexibility and scalability for the implementation of robot functions.

[0128] (3) The present invention optimizes positioning information by analyzing lidar and ultra-wideband (UWB) data and applying filtering algorithms to ensure the accuracy and stability of robot positioning.

[0129] (4) The present invention adopts SLAM technology, combined with A* algorithm and Dijkstra algorithm to achieve autonomous navigation, and supports the coordinated cooperation of manual control, local path planning and global path planning.

[0130] (5) After receiving an inspection task, the inspection vehicle of the present invention automatically plans an inspection route based on a multi-sensor system. It is capable of executing complex inspection tasks, thus improving the accuracy and safety of inspections.

[0131] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0132] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A track inspection robot based on SLAM algorithm, characterized in that: include: Inspection vehicle module, host computer module and network communication module. The inspection vehicle module and the host computer module are connected through the network communication module. The host computer module is used to publish inspection tasks, check the status of the inspection vehicle and display the inspection map; the inspection vehicle module receives the inspection tasks issued by the host computer module and plans the optimal inspection path based on the SLAM algorithm.

2. A track inspection robot based on SLAM algorithm according to claim 1, characterized in that: The inspection vehicle module includes a scanning detection system, a path navigation and control system, and a chassis motion system; The scanning detection system is used to obtain the motion path and the surrounding environment information of the motion path, and transmit the motion path and the surrounding environment information of the motion path to the path navigation and control system; The path navigation and control system performs path planning and decision-making based on the motion path and the surrounding environment information of the motion path and the inspection tasks issued by the host computer, and generates specific motion instructions; The chassis motion system is used to receive and execute motion instructions to drive the inspection vehicle module to move along the planned path.

3. A track inspection robot based on SLAM algorithm according to claim 2, characterized in that: The path navigation and control system includes a SLAM synchronous positioning and map construction algorithm model module, a path planning module, an operation control model module, a dynamic obstacle avoidance module, a data fusion module, a communication module and an intelligent control module; The SLAM synchronous positioning and mapping algorithm model module uses the ORB-SLAM3 algorithm to perform real-time map construction and positioning based on the surrounding environment information to obtain basic information of the motion trajectory; The path planning module uses the A* algorithm and the Dijkstra algorithm to plan the path according to the basic information of the motion trajectory and determine the driving path; The operation control model module uses the PID operation control model to control the robot to track the driving path; The dynamic obstacle avoidance module detects the position of obstacles in real time through sensors and dynamically adjusts the driving path in combination with path planning; The data fusion module fuses the various data from the above modules to improve the accuracy and reliability of sensor detection and optimize the accuracy of positioning and map construction; The communication module is used to receive the inspection tasks issued by the host computer module and the instruction information of each module in the path navigation and control system, and transmit them to the intelligent control module; The intelligent control module is used to integrate the command information of each module in the path navigation and control system, perform comprehensive analysis and decision-making, and output comprehensive control instructions to the chassis motion system, while feeding back the control status to the host computer module through the communication module.

4. A track inspection robot based on SLAM algorithm according to claim 3, characterized in that: The ORB-SLAM3 algorithm performs real-time map construction and positioning based on surrounding environment information. The formula for feature extraction and matching is as follows: f(x)=ORB(I t ) Among them, I t is the input image at the current time t; f(x) is the feature point matrix extracted from the image; The pose estimation formula is as follows: x t =PnP(f(x t-1 ),f(x t )) Among them, x t-1 ,x t are the previous state and current state of the robot respectively; The map update formula is as follows: Among them, M t The map at the current time consists of multiple key frames f k The feature points of U represent the integration of feature points.

5. The track inspection robot based on SLAM algorithm according to claim 3, characterized in that: The A* algorithm is used to plan the optimal driving path, and the Dijkstra algorithm is used to plan the shortest driving path. The cost function for planning the optimal driving path is as follows: f(n)=g(n)+h(n) Where f(n) is the total cost from the starting point to the target location through node n, including the path taken and the estimated path; g(n) is the actual cost from the starting point to node n, calculated based on the terrain or known map path; h(n) is the heuristic estimate from node n to the target location; The formula for calculating the shortest driving path cost is as follows: f′(n)=g′(n); Where f′(n) represents the total cost from the starting point to node n; g′(n) represents the actual cumulative cost from the starting point to node n; The path planning module generates the optimal path for the inspection vehicle to travel on the track according to the calculation results of the formula.

6. A track inspection robot based on SLAM algorithm according to claim 3, characterized in that: The calculation formula of the PID operation control model is as follows: Where e(t) is the error at the current moment, K p , K i , K d are the proportional coefficient, integral coefficient, and differential coefficient.

7. The track inspection robot based on SLAM algorithm according to claim 3, characterized in that: When the dynamic obstacle avoidance module detects an obstacle, it uses an obstacle avoidance algorithm. The obstacle avoidance formula is as follows: if d<d threshold ,then adjust direction toθ Wherein, d represents the distance to the obstacle, θ represents the adjusted direction, and the movement speed and driving direction are adjusted in real time in combination with the detection results of the scanning detection system to achieve obstacle avoidance.

8. The track inspection robot based on SLAM algorithm according to claim 3, characterized in that: The data fusion module adopts a data fusion algorithm, and the formula is as follows: X fusion =X predicted +K·(Z-HX predicted ) Among them, X fusion It combines the estimated and actually observed sensor information to provide real-time correction; K is the Kalman gain, which controls the correction amplitude; Z is the actual measurement data; H is the observation model, which is used to map the prediction to the actual observation and is used to fuse the detection data from the scanning detection system.

9. The track inspection robot based on SLAM algorithm according to claim 2, characterized in that: The chassis motion system has an adjustable gauge structure, including symmetrically arranged traveling seat units, and a gauge adjustment screw drive module is arranged in the fixed side traveling seat unit.

10. The track inspection robot based on SLAM algorithm according to claim 2, characterized in that: The scanning detection system is a three-dimensional laser scanning device, including a high-strength shell and an internal integrated equipment cabin; the internal integrated equipment cabin has a scanning unit and a control unit arranged in layers; The scanning unit includes laser radar, industrial camera, and infrared camera; The control unit includes a data router and controller and a system power supply; Each unit is fixed by a modular bracket. The bracket body adopts a connection column and distributed connection point design and is connected to the inspection vehicle module through a damping hinge.

Images