A magnetic coupling driving-based internal-external collaborative pipeline inspection robot system

Abstract

This application discloses a magnetically coupled collaborative pipeline inspection robot system, relating to the field of pipeline inspection equipment technology. The system includes an internal robot and an external robot. The internal robot is located inside the pipeline, while the external robot is located outside the pipeline at a corresponding position, and the two are magnetically coupled together. During the inspection process, angular velocity, acceleration, and distance to the pipeline wall are collected to control the movement of the external robot. A camera on the internal robot captures images of the pipeline interior, which are analyzed by a vision processor to determine defects inside the pipeline. By employing a magnetically coupled dual-drive system and a Halbach array design, this application achieves contactless power transmission between the external and internal robots, eliminating the risks of entanglement, wear, and leakage explosion that may occur in high-risk environments such as oil and gas and power plants caused by cable-driven power supply structures, thus improving the safety of inspection operations and the stability of system operation.

Description

Technical Field

[0001] This application relates to the field of pipeline inspection equipment technology, and in particular to a robotic system for inspecting pipelines both inside and outside via magnetic coupling. Background Technology

[0002] Pipeline systems are widely used in oil and gas transportation, power generation, municipal water supply and drainage, and chemical production, serving as a crucial component for ensuring the safe operation of energy and urban infrastructure. With increasing service life, pipelines are prone to problems such as cracks, corrosion, scale buildup, weld defects, and structural aging, necessitating regular inspection and evaluation to promptly identify potential hazards. Due to the confined space and enclosed environment within pipelines, coupled with their complex structures including bends, diameter changes, branches, and deposits, manual inspection is challenging and carries high safety risks. Therefore, pipeline inspection technology based on robotic platforms is gradually becoming an important development direction.

[0003] Existing pipeline inspection robots mostly operate by moving inside pipelines and using cables for power and communication. This type of cable-driven structure has significant structural shortcomings in practical applications: as the inspection distance increases, the cable is prone to friction, entanglement, knotting, and wear on the pipeline wall and at bends, affecting not only the robot's stability but also significantly increasing maintenance costs. In flammable and explosive environments such as oil and gas or chemical plants, cable damage can also lead to leakage or spark risks, posing significant safety hazards. Furthermore, the traction and inertia of the cable continuously interfere with the robot's movement, easily causing interruptions in the inspection process over long distances or in multi-bend pipelines, making continuous and stable inspection operations difficult.

[0004] To break free from cable constraints, some existing technologies attempt to achieve cableless operation using built-in batteries or single-unit drive structures. However, such solutions are typically limited by the internal space of pipelines, making it difficult to balance battery capacity and drive power, resulting in limited endurance. Furthermore, the high integration of detection, drive, and power supply modules into a single unit complicates the system structure and concentrates the load, leading to poor maneuverability in complex bends or pipe sections with varying diameters, and difficulty adapting to pipe environments with different materials and diameters. In addition, single-unit rigid structures are prone to instability, jamming, or slippage when passing through welds or areas of abrupt curvature changes, affecting detection reliability.

[0005] In terms of intelligent and remote operation and maintenance, existing pipeline inspection systems mostly rely on manual real-time monitoring or post-event data analysis, lacking stable support conditions for the inspection process and making it difficult to continuously acquire high-quality inspection data under complex operating conditions. At the same time, due to limitations in power supply methods and structural layout, existing technologies still have significant shortcomings in terms of driving capability, inspection stability, and system coordination, making it difficult to simultaneously meet the engineering application requirements of safety, adaptability, and continuous operation.

[0006] Therefore, how to achieve stable and reliable cableless drive and detection operation in complex pipeline environments while avoiding the inherent defects of cable-driven structures remains a key issue that needs to be addressed in existing pipeline inspection technologies. Summary of the Invention

[0007] This application provides a magnetically coupled, collaborative pipeline inspection robot system to address the aforementioned problems in the prior art.

[0008] This application provides an embodiment of a magnetically coupled, collaborative pipeline inspection robot system, including: The internal robot includes a camera and a vision processor, which are connected by a central axis, on which an internal magnet disk and wheeled support legs are mounted. The external robot includes a main body and external tires. The bottom of the main body is equipped with an external magnet disk, and the external tires are connected to the bottom of the main body through adjustable suspension support legs. The external tires are driven by a geared motor, and the main body is equipped with an inertial measurement unit and a laser ranging module. When the internal robot is inside the pipe and the external robot is outside the pipe at a position corresponding to the internal robot, the magnetic force between the inner and outer magnetic disks couples the two robots together. The inertial measurement unit (IMU) collects the angular velocity and acceleration of the external robot during its movement, performs attitude estimation based on the angular velocity and acceleration, and controls the geared motors based on the estimated attitude. The laser ranging module collects the distances between the left and right sides of the external robot and the outer wall of the pipe, and controls the rotation speed of the geared motors on the left and right sides based on the deviation between the two distances. When the internal robot moves synchronously with the external robot, the camera captures images of the inside of the pipe, which are analyzed by the vision processor to determine defects inside the pipe.

[0009] The magnetically coupled, collaborative pipeline inspection robot system described in this application has the following advantages: 1. By adopting a magnetically coupled dual-drive system and Halbach array design, contactless power transmission between the outdoor and indoor units is achieved. This fundamentally eliminates the risks of entanglement, wear, leakage, and explosion that may occur in high-risk environments such as oil and gas or electricity supply systems, significantly improving the safety of testing operations and the stability of system operation. The system adopts an "external drive, internal sensing" dual-machine collaborative architecture, completely decoupling the power module and the detection module. This effectively avoids the problem of detection signal distortion caused by power interference, making the structural layout more reasonable and the operation more reliable.

[0010] 2. At the structural level, this application introduces a rigid-flexible hybrid connection mechanism, a universal joint connection structure, and a retractable support leg design, enabling the robot to operate flexibly in pipes of different diameters, materials, and complex curvatures. This structure endows the robot with excellent cornering ability and posture self-adaptation, allowing it to maintain stable wall-hugging movement and image acquisition even in complex working conditions such as welds, diameter changes, and curved pipe sections.

[0011] 3. Regarding motion control, this application combines the MPU6050 inertial measurement unit, Kalman filtering algorithm, and PID (proportional-integral-derivative) control method to construct an attitude angle fusion and dynamic adjustment mechanism. This enables the outdoor unit to correct attitude deviations in real time during operation, ensuring magnetic coupling synchronization and motion balance between the indoor and outdoor units, thereby effectively preventing slippage, loss of synchronization, or magnetic coupling disconnection problems. The laser ranging module and path-keeping algorithm equipped on the outdoor unit can sense the position of the outer wall in real time and autonomously adjust the trajectory, ensuring high-precision cruise and low-energy operation of the system in complex environments.

[0012] 4. In terms of detection capabilities, the internal unit integrates a Raspberry Pi edge computing platform and a lightweight YOLO-Fast algorithm, enabling real-time defect identification and analysis in low-light, corroded, or dirty environments. The system can intelligently detect and automatically label various types of defects such as cracks, corrosion, and flaking, significantly improving detection efficiency and accuracy compared to traditional manual inspections, while reducing data latency and manual intervention.

[0013] 5. By combining the Qt host computer visualization platform with the VNC (Virtual Network Control) remote interaction protocol, a cross-platform, traceable remote operation and maintenance system was built. Users can remotely control robot operation, view inspection screens, and record defect data in real time through multiple terminals, realizing a fully automated closed loop from data collection to analysis report generation, improving the standardization and traceability of inspections.

[0014] In summary, this application achieves the organic integration of cableless drive, highly adaptable structural design, intelligent detection and analysis, and remote operation and maintenance, forming a safe, reliable, intelligent, efficient, and scalable new pipeline inspection system with broad engineering application value and industrial promotion potential. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the internal pipeline robot provided in an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of the structure of the external pipeline robot provided in an embodiment of this application.

[0018] Figure 3 This is a schematic diagram of the Halbach array magnet in an embodiment of this application.

[0019] Figure 4 A schematic diagram of the leg structure of an internal pipeline robot provided in an embodiment of this application.

[0020] Figure 5 A schematic diagram of the leg structure of the external pipeline robot provided in this application embodiment.

[0021] Figure 6 This is a flowchart illustrating the attitude control and path maintenance in an external pipeline robot provided in an embodiment of this application.

[0022] Figure 7 A flowchart illustrating the visual inspection and remote visualization process of an internal pipeline robot provided in this application embodiment.

[0023] The following are the reference numerals: 1. Camera; 2. Universal joint; 3. Vision processor; 4. Inner magnet plate; 5. Outer magnet plate; 6. Outer tire; 7. Gear motor; 8. Steering device; 9. Laser rangefinder module; 10. Flange; 11. Link 1; 12. Inner tire; 13. Link 2. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] Figure 1-5 This is a schematic diagram of a magnetically coupled, collaborative pipeline inspection robot system for internal and external pipeline inspection provided in an embodiment of this application. The embodiment of this application provides a magnetically coupled, collaborative pipeline inspection robot system for internal and external pipeline inspection, including: The internal robot includes a camera 1 and a vision processor 3, which are connected by a central axis. An internal magnetic disk 4 and wheeled support legs are mounted on the central axis. The external robot includes a main body and external tires 6. An external magnet disk 5 is provided at the bottom of the main body. The external tires 6 are connected to the bottom of the main body through adjustable suspension support legs. The external tires 6 are driven by a geared motor 7. An inertial measurement unit and a laser ranging module 9 are provided on the main body. When the internal robot is inside the pipe and the external robot is outside the pipe at a position corresponding to the internal robot, the magnetic force between the inner magnetic disk 4 and the outer magnetic disk 5 couples the internal and external robots together. The inertial measurement unit collects the angular velocity and acceleration of the external robot during its movement, performs attitude estimation based on the angular velocity and acceleration, and controls the geared motor 7 based on the estimated attitude. The laser ranging module 9 collects the distance between the left and right sides of the external robot and the outer wall of the pipe, and controls the rotation speed of the geared motors 7 on the left and right sides based on the deviation between the two distances. When the internal robot moves synchronously with the external robot, the camera 1 collects images of the inside of the pipe, which are analyzed by the vision processor 3 to determine the defects inside the pipe.

[0026] For example, the magnet blocks in the inner magnet disk 4 and the outer magnet disk 5 are arranged in a Halbach array.

[0027] Specifically, both the inner magnet disk 4 and the outer magnet disk 5 are composed of multiple N52 neodymium iron boron magnets. These N52 neodymium iron boron magnets are arranged in a Halbach array, so that the magnets in the outer magnet disk 5 are arranged with high directivity to form a unilateral enhanced magnetic field, while the magnets in the inner magnet disk 4 form a corresponding magnetic flux path on the other side of the pipe wall. Through the superposition effect of the directional magnetic fields of the Halbach array, a synchronous magnetic field torque can be generated when the outer robot rotates, enabling the inner robot to achieve completely synchronized movement.

[0028] Compared to traditional toroidal magnets, the Halbach array exhibits stronger magnetic field concentration and higher flux utilization, enabling the achievement of greater torque transmission in a smaller space and reducing flux leakage, thereby improving drive efficiency and energy transfer stability. Magnetic field simulations have verified that this structure maintains stable magnetic coupling in pipes made of different materials, including steel and PVC (polyvinyl chloride), providing a reliable guarantee for continuous system propulsion.

[0029] Furthermore, a universal joint 2 is provided on the central shaft, dividing the central shaft into two parts. The two parts of the central shaft are respectively connected to the camera 1 and the vision processor 3. Each part of the central shaft is equipped with an inner magnet disk 4 and a set of wheel-type support legs. For example... Figure 4As shown, the wheeled support leg includes a flange 10, a first connecting rod 11, a second connecting rod 13 for hinged connection, and an internal tire 12. Both flanges 10 are mounted on a central shaft, and fastening bolts are installed on the flanges 10. One end of the first connecting rod 11 and the second connecting rod 13 is hinged to a fixed point on the central shaft, and the other end is hinged to one of the flanges 10. The internal tire 12 is mounted on the outer end of the second connecting rod 13. By adjusting the tightness of the fastening bolts on the flanges 10, the flanges 10 can slide along the central shaft, thereby changing the angle between the first connecting rod 11 and the second connecting rod 13, allowing the internal tire 12 to expand or retract radially, adapting to pipes of different inner diameters. After adjustment, tightening the bolts locks the flanges 10 in the corresponding position on the central shaft. When the wheeled support leg is expanded, the internal tire 12 contacts the inner wall of the pipe, providing radial support and centering for the internal robot. Simultaneously, the internal tire 12 can rotate freely without hindering the internal robot's axial movement along the pipe.

[0030] Specifically, the internal robot can be divided into a head and a tail section via the universal joint 2. The camera 1 is located in the head, while the vision processor 3 is located in the tail. The head and tail sections are connected by the universal joint 2, giving the internal robot both rigidity and a certain degree of flexibility. This allows the internal robot to move smoothly in scenarios requiring turning, such as when turning or branching in pipes, without getting stuck due to the length of the internal robot.

[0031] Furthermore, the main body also has two parts, which are connected by a steering device 8. The two main bodies are used to install the battery and the MPU6050 inertial measurement unit, respectively. Each part has an outer magnet disk 5 at the bottom, and the positions of the two outer magnet disks 5 correspond one-to-one with the two inner magnet disks 4.

[0032] Specifically, the steering device 8 is a rigid-flexible hybrid series structure used to connect the front and rear sections of the external robot. This structure consists of multiple cross-shaped node units connected in series. Each node unit comprises a central sphere and connecting rods extending in four directions (up, down, left, and right), and is integrally formed using 3D printing. The connection method between adjacent node units varies depending on the direction: the connecting rods on the left and right sides are connected by flexible silicone parts, giving the steering device flexibility in the horizontal direction, allowing the front and rear sections to deflect left and right around the vertical axis to adapt to pipe bends, while also allowing a certain degree of torsion around the longitudinal axis to adapt to local surface changes on the pipe's outer wall; the connecting rods in the vertical direction are connected by 3D-printed rigid connectors, maintaining rigid constraint in the vertical direction, preventing relative folding of the front and rear sections in the direction of gravity, and ensuring structural stability. This rigid-flexible hybrid design allows the external robot to conform to the pipe's curvature in a controlled manner when passing through bends, while maintaining sufficient structural strength to bear its own weight and driving loads.

[0033] In the embodiments of this application, the external robot adopts a multi-degree-of-freedom adjustable suspension structure and a high-torque wheel set design, which can adjust the wheel set angle and posture according to the curvature of the outer wall of the pipe to maintain good fit and stability. Specifically, each side suspension support leg of the external robot adopts a multi-stage series hinge structure, consisting of a main body connecting plate, a rocker arm connecting rod, and a wheel set bracket. Each stage is connected by hinge points, and each hinge point is equipped with an adjusting bolt. By manually turning the bolt, the relative angle between adjacent stages can be changed and locked. Before use, the operator adjusts the angle of each hinge point according to the outer diameter of the target pipe, changing the unfolding range of the wheel set bracket relative to the main body, so that the external tire 6 can fit the outer wall curvature of pipes with different diameters. A high-reduction ratio geared motor 7 is installed at the end of the wheel set bracket. The geared motor 7 directly drives the external tire 6, and the large reduction ratio obtains sufficient output torque to ensure that the external robot travels stably on the outer wall of the pipe. The adjustable suspension structure allows the external robot to adapt to pipes of different diameters, while the multi-stage articulation design can absorb local unevenness such as welds and protrusions on the outer wall of the pipe, maintaining a good fit between the external tires 6 and the pipe wall.

[0034] Furthermore, the inertial measurement unit is model MPU6050. During attitude estimation, a Kalman filter algorithm is used to suppress noise accumulation errors.

[0035] Specifically, the inertial measurement unit (IMU) collects the angular velocity and acceleration data of the external robot in real time to achieve attitude angle fusion estimation. The angle fusion process can be expressed as follows:

[0036] In the formula, and The first Time and the The estimated attitude angle at time t. For Kalman gain, The accelerometer readings obtained by the accelerometer in the inertial measurement unit are the first... From the perspective of time, The first result obtained by integrating the gyroscope in the inertial measurement unit. The predicted angle at any given time. Kalman filtering can suppress accumulated noise errors, thus achieving high-precision attitude estimation, such as... Figure 6 As shown.

[0037] Furthermore, when controlling the geared motor 7 based on the estimated posture, a PID control algorithm is used to control the geared motor 7 in order to achieve dynamic correction of the posture error of the external robot.

[0038] Specifically, the discrete control law of the PID control algorithm is:

[0039] In the formula, For the first k The output control signal at any given time, i.e., the duty cycle of PWM (Pulse Width Modulation). For the first k The deviation between the target attitude at any given moment and the current attitude angle estimate. For the first k The deviation at time -1 for n Timing deviation, To control the period, i.e. the sampling interval, , , These are the proportional, integral, and derivative coefficients, respectively. The output control signal of the PID control algorithm. Used to control the speed of the geared motor 7. Specifically, the control signal... The PWM duty cycle of the corresponding geared motor 7 driver is adjusted to change the speed of the geared motor, thereby achieving dynamic correction of the external robot's posture deviation.

[0040] Furthermore, the laser ranging module 9, model VL53L0X, is used to collect in real time the distances between the left and right sides of the front end of the external robot and the outer wall of the pipe. d L and d R Calculate the normalized path deviation e Then, the motor speed difference is calculated using a PD (proportional-derivative) control law. The speeds of each reduction motor 7 are adjusted uniformly to correct the travel path. The path deviation is calculated based on the left and right distance measurement data, and its mathematical expression is:

[0041] In the formula, and These are the distance values ​​measured by the laser ranging modules 9 on the left and right sides, respectively. This is the normalized path deviation. Based on this deviation, the controller in the external robot adjusts the speed difference between the left and right reduction motors 7 in real time to achieve automatic path correction. The adjustment rule is as follows:

[0042] In the formula, For the first k Motor speed difference control quantity at any given time. and The first k and k Normalized path bias at time -1 This is the path deviation ratio coefficient. This is the path deviation rate coefficient. This control law can effectively suppress path deviation and improve the fitting accuracy of the pipe outer wall and the stability of the system.

[0043] Furthermore, during the movement of the external robot, an external robot attitude balance model is established. The output torque of the geared motor 7 is calculated using the external robot attitude balance model, and then the output torque of the geared motor 7 is controlled.

[0044] Specifically, in complex working conditions such as curves, welds, and cracks, the system in this embodiment can automatically correct the running path, reducing attitude fluctuations and magnetic coupling loss of synchronization. The relationship between the output torque of the geared motor 7 and the attitude angle change of the external robot can be expressed by the dynamic equation:

[0045] In the formula, This is the output torque of the geared motor 7. For the rotational inertia of the external robot, Angular acceleration, which is the second derivative of the attitude angle with respect to time. The current attitude angle, The damping coefficient is... It is the system equilibrium angle, i.e., the desired attitude angle. This represents the magnetic coupling recovery coefficient. This equation allows for the establishment of an outdoor unit attitude balance model, providing a theoretical basis for controller parameter adjustment.

[0046] Furthermore, the vision processor 3 is equipped with a YOLO-Fast defect recognition model for identifying defects inside the pipeline.

[0047] Specifically, the vision processor 3 on the internal robot uses a Raspberry Pi 4B, and this processing platform, along with camera 1, constructs a lightweight edge computing environment. Through a YOLO-Fast defect recognition model deployed on the NCNN (Neural Networking Matrix) inference framework, the vision processor 3 can achieve real-time identification and localization of defects such as cracks, corrosion, flaking, and scale buildup inside the pipe. The system in this application adopts the publicly available YOLO-Fast lightweight object detection network architecture. YOLO-Fast is a single-stage detection model optimized for embedded platforms, employing a simplified convolutional backbone network. While maintaining high detection accuracy, it significantly reduces computational load, making it suitable for real-time inference on edge devices with limited computing power. The model's input image size is 320×320 pixels. Detection target categories include common defects inside pipes such as cracks, corrosion, flaking, and scale buildup. During the training phase, a publicly available pipe defect image dataset is used to label and train the above defect categories. After training, the model is deployed on a Raspberry Pi 4B platform through the NCNN inference framework to achieve real-time identification and localization of defects inside the pipe. During training, the core loss function of the detection model can be expressed as:

[0048] In the formula, For loss function, For the coordinates of the predicted bounding box, For the actual coordinates, This is the confidence level prediction value. This is the true confidence level, where 1 is the value if the target actually exists and 0 is the value if it does not exist. and These are the weight coefficients for coordinate loss and background loss, respectively. For indicator functions, when the first i The first grid j A prediction box is set to 1 when it is responsible for detecting a target, and 0 otherwise. S This represents the number of grid divisions in the feature map, i.e., dividing the input image into an S×S grid. B The number of bounding boxes predicted for each grid cell. This loss function achieves rapid identification of defective targets by optimizing location accuracy and confidence prediction.

[0049] Furthermore, the vision processor 3 sends the acquired images and defect detection results to the external robot. The external robot then sends the images, detection parameters, and defect detection results to the host computer. The host computer interacts with the data through a visual interface developed based on the Qt framework and performs remote monitoring across multiple terminals based on the VNC remote desktop protocol. Specifically, the vision processor sends the acquired images and defect detection results to the external robot via Bluetooth wireless communication. The vision processor in the internal robot has a built-in Bluetooth module, which establishes a wireless connection with the Bluetooth receiver module on the external robot, enabling data transmission between both sides of the pipe wall without the need for physical cables to pass through the pipe wall.

[0050] Specifically, the external robot interacts with the host computer via a visual interface developed in Qt, and enables remote monitoring across multiple terminals based on the VNC remote protocol. Operators can view the inspection screen, adjust inspection parameters, and record defect information in real time on a PC (personal computer) or mobile device, thereby achieving closed-loop control of the entire process from data acquisition to analysis and output.

[0051] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0052] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A collaborative pipeline inspection robot system based on magnetic coupling drive, characterized in that, include: The internal robot includes a camera (1) and a vision processor (3), the camera (1) and the vision processor (3) being connected by a central axis, on which an internal magnet disk (4) and wheeled support legs are provided; An external robot includes a main body and an external tire (6). The bottom of the main body is provided with an external magnet disk (5). The external tire (6) is connected to the bottom of the main body through adjustable suspension support legs. The external tire (6) is driven by a geared motor (7). An inertial measurement unit and a laser ranging module (9) are provided on the main body. When the internal robot is inside the pipe and the external robot is outside the pipe at a position corresponding to the internal robot, the magnetic force between the inner magnet disk (4) and the outer magnet disk (5) couples the internal robot and the external robot together; the inertial measurement unit collects the angular velocity and acceleration of the external robot during its movement, performs attitude estimation based on the angular velocity and acceleration, and controls the geared motor (7) based on the estimated attitude; the laser ranging module (9) collects the distance between the left and right sides of the external robot and the outer wall of the pipe, and controls the rotation speed of the geared motor (7) on the left and right sides based on the deviation of the two distances; when the internal robot moves synchronously with the external robot, the camera (1) collects images inside the pipe, and the vision processor (3) analyzes and determines the defects inside the pipe.

2. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, A universal joint (2) is provided on the central shaft, which divides the central shaft into two parts. The two parts of the central shaft are respectively connected to the camera (1) and the vision processor (3). Each part of the central shaft is provided with an inner magnet disk (4) and a set of wheel support legs.

3. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 2, characterized in that, The main body also has two parts, which are connected by a steering device (8). Each part of the main body has an outer magnet disk (5) at its bottom, and the positions of the two outer magnet disks (5) correspond one-to-one with the two inner magnet disks (4).

4. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 3, characterized in that, The steering device (8) is a rigid-flexible hybrid series structure.

5. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, The magnet blocks in the inner magnet disk (4) and the outer magnet disk (5) are arranged in a Halbach array.

6. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, In the attitude estimation process, the Kalman filter algorithm is used to suppress noise accumulation error.

7. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, When controlling the geared motor (7) based on the estimated posture, a PID control algorithm is used to control the geared motor (7) in order to achieve dynamic correction of the posture error of the external robot.

8. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, During the movement of the external robot, an external robot attitude balance model is established, and the output torque of the geared motor (7) is calculated using the external robot attitude balance model, thereby controlling the output torque of the geared motor (7).

9. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, The vision processor (3) is equipped with a YOLO-Fast defect recognition model, which is used to identify defects inside the pipeline.

10. The internal and external collaborative pipeline inspection robot system based on magnetic coupling drive according to claim 1, characterized in that, The vision processor (3) sends the acquired images and defect detection results to the external robot. The external robot sends the images, detection parameters and defect detection results to the host computer. The host computer interacts with the data through a visual interface developed based on the Qt framework, and performs remote monitoring of multiple terminals based on the VNC remote desktop protocol.

Images