Underwater or under-ice detection robot control system and control method for narrow space detection

Abstract

This invention provides a control system and method for an underwater or sub-ice exploration robot applied to confined space exploration. One aspect of the control system includes a load-bearing umbilical cable. One end of the umbilical cable is connected to an ROV (Remotely Operated Vehicle), and the other end is wound around a winch drum. A magnetic ring is fixed to the end of the umbilical cable connected to the ROV, forming a linkage structure with the ROV to generate analog feedback to the control unit. A spring-loaded structure for resetting the magnetic ring is provided between the magnetic ring and the ROV. A tension sensor for monitoring the tension on the umbilical cable is fixed to one side of the winch, and the tension sensor is connected to the control unit. This invention can detect the umbilical cable status in real time and adaptively adjust the umbilical cable length, ensuring that the umbilical cable does not become taut due to contact with obstacles it avoids. This allows the ROV to autonomously return from complex terrain after completing observations in confined areas.

Description

Technical Field

[0001] This invention belongs to the field of detection technology, specifically relating to a control system and control method for an underwater or under-ice detection robot applied to detection in confined spaces. Background Technology

[0002] In the exploration of deep-sea biological and mineral resources, close-range observation is often required to achieve detailed imaging and 3D mapping of deep-sea organisms and ecosystems. This often necessitates close proximity to the terrain, and even traversing narrow and complex terrain for exploration. In the operation and maintenance of subsea oil and gas pipelines, for pipeline nodes with mesh protective covers, it is necessary to enter the grid to better detect leaks. In the exploration of subglacial lakes in polar regions or under ice on other planets, similar narrow and complex environments may exist beneath the ice. In such cases, underwater robots are required to have the ability to autonomously return after close-range exploration.

[0003] A Remotely Operated Vehicle (ROV) is an underwater work vehicle remotely operated by a surface operator using an umbilical cable. ROVs are widely used in deep-sea oil extraction, marine scientific research, military applications, and aquaculture. In an ROV, power and communication are transmitted via the umbilical cable, allowing it to operate without limitations on power and data processing capabilities and enabling real-time monitoring of the underwater environment. As a crucial component of the ROV, the stability of the umbilical cable directly affects its power, control, and other functions. The umbilical cable needs to be deployed and retracted in real-time based on the ROV's position. Excessive release can lead to entanglement and snagging, while delayed release can cause dragging, severely impacting movement and reducing operational efficiency. Monitoring the umbilical cable's status improves ROV operational efficiency, and real-time monitoring allows for a more accurate understanding of the ROV's working condition, enabling proactive prevention and handling of potential problems.

[0004] Existing identification methods include "a method for identifying and warning about the end status of ROV umbilical cable". This method only identifies and warns about the ROV side end of the umbilical cable, without making any actual processing. It can only remind the operator of the current status value and cannot judge the overall status of the umbilical cable.

[0005] The patented method and system for umbilical cable take-up and release control based on fiber optic grating stress sensing involves adding optical fibers to the umbilical cable and detecting strain by identifying the wavelength of the light reflected from the grating. However, this method only measures the stress at the connection between the ROV and the umbilical cable. When passing through narrow and complex areas, the umbilical cable may bend in multiple sections, which could have an unknown impact on the overall strain results, making it unsuitable for this application.

[0006] "A cable shape monitoring system for underwater collaborative operations" measures the relative frequency shift of Brillouin scattered light in an optical fiber using a distributed optical fiber temperature and strain monitoring system (Brillouin Optical Time Domain Analysis, BOTDA). This is combined with sensor fusion using location analysis, cable length calculation modules, and other sensors. While BOTDA can obtain information for every point on the optical fiber by measuring strain and temperature, it is more susceptible to interference and has high operating costs. Summary of the Invention

[0007] To address the problems described in the background section, the present invention aims to provide a control system and method for an underwater or under-ice exploration robot applied to confined space exploration. This system monitors the umbilical cable status in real time and adaptively adjusts the cable length to ensure that the cable does not become taut due to contact with obstacles, thus preventing ROV movement. It also allows the ROV to autonomously return from complex terrain after completing its exploration in a confined area, without the cable becoming tangled due to excessive length or stretched due to insufficient length. Furthermore, it provides auxiliary functions for ROV control, enabling the ROV to autonomously return after completing its task in a confined space.

[0008] The technical solution adopted in this invention is:

[0009] A control system for an underwater or sub-ice exploration robot used for exploration in confined spaces includes a control unit, a winch, an ROV (Remotely Operated Vehicle), and a load-bearing umbilical cable. One end of the load-bearing umbilical cable is connected to the ROV, and the other end is wound around the winch drum. A magnetic ring is fixed to the end of the load-bearing umbilical cable connected to the ROV, and the magnetic ring and the ROV form a linkage structure to generate analog feedback to the control unit. A spring structure for resetting the magnetic ring is provided between the magnetic ring and the ROV. A tension sensor for monitoring the tension on the load-bearing umbilical cable is fixed to one side of the winch, and the tension sensor is connected to the control unit.

[0010] Furthermore, the load-bearing umbilical cable is a zero-buoyancy load-bearing cable containing an internal battery core, which can transmit electrical energy and signals simultaneously.

[0011] Furthermore, the ROV is equipped with a waterproof and pressure-resistant Hall sensor array, an underwater camera, and an attitude sensor. All of these components are connected to the control unit. The Hall sensor array is arranged in a circular structure and is fixed in a ring-like embedded manner at the port where the ROV connects to the load-bearing umbilical cable. It interacts with the magnetic ring on the load-bearing umbilical cable to provide real-time feedback on the angle and orientation of the cable connection. The underwater camera is arranged parallel to the port where the ROV connects to the load-bearing umbilical cable and is used to capture images of the ROV's tail and the load-bearing umbilical cable. In this invention, when the magnetic ring approaches the outer sensors, the values ​​of nearby Hall sensors increase, while the values ​​of Hall sensors farther from the magnetic ring decrease. This allows the data obtained from the sensor array to infer changes in the spatial magnetic field. Specifically, the distance between the Hall sensor array and the magnetic ring can help determine the direction of the umbilical cable and the lateral tension near the ROV end, thus preventing cable entanglement caused by prolonged angular deviation during ROV deployment and retrieval due to turning.

[0012] Furthermore, the ROV is equipped with a positioning sensor capable of positioning, and the positioning sensor is connected to the control unit. For example, a DVL or USBL sensor.

[0013] A control method for underwater or sub-ice exploration robots used for exploration in confined spaces, the specific steps of which are as follows:

[0014] S1, the ROV detaches from the docking station and begins exploration. The underwater camera captures images of the ROV's tail, and the control unit performs VSLAM on the tail images to acquire feature points and record the path.

[0015] S2, the winch works in conjunction with the ROV's thruster to release the load-bearing umbilical cable, and autonomously judges and controls the state adjustment of the load-bearing umbilical cable by using the vision of the underwater camera, the Hall sensor array, and the parameters of the tension sensor.

[0016] S3, determine whether to recycle ROV. If not, continue with steps S1 and S2. If yes, proceed to step S4.

[0017] S4, the ROV begins to return to the dock. Using the recorded path calculated by VSLAM as a reference, the ROV's return path is planned, and the thrusters control the ROV to return to the dock along the planned path.

[0018] The S5 winch, in conjunction with the ROV's thruster control, retrieves the load-bearing umbilical cable and autonomously adjusts the cable's condition based on underwater camera vision, Hall sensor array, and tension sensor parameters.

[0019] Furthermore, VSLAM is performed in step S1, with the specific steps as follows:

[0020] S11 performs pre-processing for defogging and sharpening of images captured by the camera;

[0021] S12: Feature points in the image are collected using the ORB-SLAM algorithm, the real-time position of the ROV is analyzed, and the movement route is given. The images along the route are processed and maps are constructed to simulate the real-time movement trajectory of the ROV.

[0022] S13, semantic segmentation of the load-bearing umbilical cable is performed through visual recognition, and the state of the visible part of the load-bearing umbilical cable at the tail is identified separately while eliminating the interference of the load-bearing umbilical cable on path recognition.

[0023] Furthermore, in step S12, ORB-SLAM, based on the matching of ORB feature points in the segmented image and combined with data provided by the attitude sensor and positioning sensor, pre-assumes the rotation and displacement of the ROV, and estimates pixel points based on the current camera pose transformation to form a visual odometry for SLAM. The path obtained through VSLAM enhancement is more accurate than the path directly obtained by traditional DVL. Simultaneously, the map information obtained from SLAM mapping can not only serve as a reference factor for obstacle avoidance but also provide data on terrain surveying along the movement path.

[0024] Furthermore, the process of adjusting the state of the load-bearing umbilical cable in step S2 includes:

[0025] The tension of the load-bearing umbilical cable is judged in real time by Hall sensor array and tension sensor. When the deflection angle is too large or the tension value is too large, the winch is automatically controlled to accelerate the release of the load-bearing umbilical cable. When the magnetic ring rebounds to the threshold area or the tension is lower than the threshold, the normal cable release speed is restored.

[0026] If the tension of the load-bearing umbilical cable is too low or the umbilical cable at the tail of the visual recognition device is too long, the load-bearing umbilical cable will be pulled back faster until the visual recognition is normal or the analog value of the tension sensor returns to the threshold range, and the normal cable release speed will be restored.

[0027] Furthermore, step S4 also includes:

[0028] If there are no turning points in the map obtained by VSLAM, the judgment is made directly according to the straight line, and the umbilical cable is retrieved directly while driving; if there is no better regression solution, at this time, while the ROV moves in the reverse direction according to the original calculated path, the underwater camera performs machine vision recognition on the load-bearing umbilical cable to determine the orientation of the load-bearing umbilical cable and the situation of obstacles, so as to achieve a multi-directional ROV regression effect.

[0029] As the ROV approaches the docking station, the camera identifies visual features to determine its location and uses a docking algorithm to guide it back to dock, achieving more precise recovery.

[0030] Furthermore, the process of adjusting the state of the load-bearing umbilical cable in step S5 includes:

[0031] The tension of the load-bearing umbilical cable is judged in real time by Hall sensor array and tension sensor. When the deflection angle is too large or the tension value is too large, the winch is automatically controlled to release part of the load-bearing umbilical cable. When the magnetic ring rebounds to the threshold area or the tension is lower than the threshold, the normal cable winding speed is restored.

[0032] When the tension of the load-bearing umbilical cable is too low or the umbilical cable at the tail of the visual recognition is too long, the load-bearing umbilical cable is pulled back faster until the visual recognition is normal or the analog value of the tension sensor returns to the threshold range, and the normal cable pulling speed is restored.

[0033] Compared with the prior art, the significant advantages of this invention include:

[0034] 1. Suitable for controlling small ROVs in confined spaces, and provides an autonomous recovery method, which is more complete, specific, and intelligent compared to other solutions. It not only solves the problem of judging the umbilical cable status from three dimensions, avoiding the impact of excessively long or short umbilical cables on the ROV, but also enhances operator control assistance to a certain extent, achieving low-cost implementation and improving the accuracy of path mapping.

[0035] 2. It is less expensive than using fiber Bragg gratings, reuses commonly used sensors in marine exploration, and can better utilize existing sensors.

[0036] 3. Combining the functions of attitude sensors to determine its own state and DVL or USBL to determine relative position, the orientation and path of the ROV are determined and identified. Then, through the parameters of machine vision, Hall sensor array, and tension sensor, autonomous discrimination and control are performed to form a multi-sensor perception fusion and application. Finally, through the execution control of winch and thruster, the auxiliary detection and autonomous return functions in confined spaces are realized. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the winch structure of the present invention.

[0038] Figure 2 This is a schematic diagram of the ROV tail section of the present invention.

[0039] Figure 3 This is the ROV control flowchart of the present invention.

[0040] Figure 4 This is a system execution block diagram of the present invention. Detailed Implementation

[0041] The present invention will be further described below with reference to specific embodiments, but the invention is not limited to these specific embodiments. Those skilled in the art should recognize that the present invention covers all alternatives, improvements and equivalents that may be included within the scope of the claims.

[0042] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "clockwise," and "counterclockwise," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more, unless otherwise expressly defined.

[0043] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0044] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0045] Terminology Explanation:

[0046] ROV: Remotely Operated Vehicle

[0047] DVL: Doppler Velocity Log

[0048] USBL: Ultra-Short Baseline (acoustic positioning system)

[0049] VSLAM: Visual Simultaneous Localization and Mapping

[0050] ORB-SLAM: Oriented FAST and Rotated BRIEF SLAM

[0051] TMS: Tether Management System

[0052] Example 1

[0053] See Figure 1 , Figure 2 This embodiment provides a control system for an underwater or under-ice exploration robot applied to exploration in confined spaces, including a control unit, a winch 1, an ROV 2, and a load-bearing umbilical cable 3. One end of the load-bearing umbilical cable 3 is connected to the ROV 2, and the other end is wound around the winch drum 11. A magnetic ring 4 is fixed on the end of the load-bearing umbilical cable 3 connected to the ROV 2. The magnetic ring 4 and the ROV 2 form a linkage structure, generating analog feedback to the control unit. A spring-loaded structure 5 is provided between the magnetic ring 4 and the ROV 2 to reset the magnetic ring 4. The spring-loaded structure 5 can be a wound spring. A tension sensor 6 for monitoring the tension on the load-bearing umbilical cable 3 is fixed on one side of the winch 1. The tension sensor 6 is connected to the control unit.

[0054] Specifically, the winch drum 11 is fixed to the winch slip ring, and an underwater motor is fixed next to the winch drum to control its clockwise and counterclockwise rotation. A tension sensor 6 is fixed to one side of the winch 1 to monitor the tension on the umbilical cable, and its data is fed back to the control system in real time in analog form. The tension sensor 6 has a three-pulley structure, including two guide pulleys and one load-bearing pulley. The guide pulleys are fixedly connected to the side plates, and their central axis is slightly lower than that of the load-bearing pulley to facilitate the passage of the umbilical cable. The tension detection sensor is connected to the load-bearing pulley, and the three pulleys convert the parallel force on the umbilical cable into a vertical force acting on the sensor, thereby detecting the tension on the umbilical cable. The load-bearing umbilical cable 3 is a zero-buoyancy load-bearing cable containing an internal battery core, which can transmit electrical energy and signals simultaneously. Its load-bearing capacity facilitates the deployment and retrieval of the ROV2. The spring-loaded structure 5 ensures that the ROV proximal magnetic ring of the umbilical cable returns to the center position when there is no tension.

[0055] In this embodiment, the ROV2 is equipped with a waterproof and pressure-resistant Hall sensor array 7, an underwater camera 8, and an attitude sensor. All of these components are connected to a control unit. The Hall sensor array 7 is arranged in a circular structure and is fixed in a ring-like embedded manner at the port where the ROV2 connects to the load-bearing umbilical cable 3. It interacts with the magnetic ring 4 on the load-bearing umbilical cable 3 to provide real-time feedback on the angle and orientation of the connection point. The underwater camera 8 is arranged parallel to the port where the ROV2 connects to the load-bearing umbilical cable 3, and is used to capture images of the ROV2's tail and the load-bearing umbilical cable 3. When the magnetic ring 4 approaches the peripheral sensors, the values ​​of nearby Hall sensors increase, while the values ​​of Hall sensors farther from the magnetic ring 4 decrease. This allows the data obtained from the sensor array to infer changes in the spatial magnetic field. Specifically, the distance between the Hall sensor array 7 and the magnetic ring 4 can help determine the direction of the load-bearing umbilical cable 3 and the lateral tension near the ROV end, thus preventing cable entanglement caused by prolonged angular deviation during ROV deployment and retrieval due to turning. The ROV2 is equipped with a positioning sensor capable of positioning, which is connected to the control unit. For example, a DVL or USBL sensor can be used.

[0056] This invention is suitable for controlling small ROVs in confined spaces and provides a method for autonomous recovery, which is more complete, specific, and intelligent compared to other solutions. It not only solves the problem of judging the umbilical cable status from three dimensions, avoiding the impact of excessively long or short umbilical cables on the ROV, but also enhances control assistance for operators to a certain extent, achieving low-cost implementation and improving the accuracy of path mapping. It is also lower in cost than using fiber Bragg gratings and reuses commonly used sensors in marine exploration, making better use of existing sensors.

[0057] Example 2

[0058] Reference Figure 3 , Figure 4 This embodiment provides a control method for an underwater or under-ice exploration robot applied to exploration in confined spaces, the specific steps of which are as follows:

[0059] S1, the ROV detaches from the docking station and begins exploration. The underwater camera captures images of the ROV's stern, and the control unit performs VSLAM on the stern images to acquire feature points and record the path. The specific steps of performing VSLAM are as follows:

[0060] S11 performs pre-processing dehazing and sharpening on images captured by the camera to enhance the accuracy of feature recognition.

[0061] S12: Feature points in the image are collected by ORB-SLAM algorithm, the position of ROV in real time is analyzed and the movement route is given. The images along the route are processed and maps are built to simulate the real-time movement trajectory of ROV. The obtained trajectory route can be used as a reference for manual control and also provides a route basis for ROV autonomous regression.

[0062] ORB-SLAM, based on the matching of ORB feature points in the segmented image and combined with data from the attitude and localization sensors, pre-assumes the rotation and displacement of the ROV and estimates pixel points based on the current camera pose change, forming a visual odometry for SLAM. The path obtained through VSLAM enhancement is more accurate than the path directly obtained by traditional DVL. Furthermore, the map information obtained from SLAM mapping can not only serve as a reference factor for obstacle avoidance but also provide data on terrain surveying along the movement path.

[0063] The key feature of ORB-SLAM is an improved FAST corner detection method, which uses the difference in grayscale between a local pixel and the brightness of its neighboring pixels as a feature for corner detection. Pixel p is selected, and its brightness is set to I. p Set the threshold to I p 20% of the value, and select 16 pixels on a circle with a radius of 3 centered at p. When the brightness of N consecutive points on the circle is greater than I... p 120% or less than I p When 80% of the image is sampled, it is considered a feature point. Based on this, image pyramids of different resolutions can be constructed by downsampling the image at different levels, achieving scale invariance; the moment of image block B is defined as m using the gray-level centroid method. pq =∑ x,y∈B x p y q I(x,y),p,q={0,1}, find the center of gray value weights of the image patch, that is, the centroid of the image gray values. The direction vector connecting the geometric center O and centroid C of image patch B Then the direction of the feature point θ = arctan(m) 01 / m 01 This process achieves rotation invariance. After extracting keypoints, ORB uses an improved BRIEF feature description to randomly select and compare feature points, resulting in an n-dimensional vector composed of 0s and 1s, thus obtaining ORB feature points. Finally, feature matching is used to achieve data association in the graph.

[0064] S13: Semantic segmentation of the load-bearing umbilical cable is performed using visual recognition. This allows for the separate identification of the visible tail portion of the load-bearing umbilical cable while eliminating interference from the cable's role in path recognition. Since the umbilical cable is typically a yellow, zero-buoyancy cable, the yellow portion is separated from the acquired image using grayscale thresholding. Edge detection is performed on the segmented sections, and based on pre-set settings and training, the system determines if the umbilical cable is becoming too long, thereby controlling the winch to retrieve it. ORB-SLAM is then used to generate the remaining portion of the image, enabling the separation of the umbilical cable portion that moves with the ROV and eliminating specified interference.

[0065] S2, the winch works in conjunction with the ROV's thruster to release the load-bearing umbilical cable, and autonomously judges and controls the state adjustment of the load-bearing umbilical cable by using the vision of the underwater camera, the Hall sensor array, and the parameters of the tension sensor.

[0066] The process of adjusting the condition of the load-bearing umbilical cable includes:

[0067] The tension of the load-bearing umbilical cable is judged in real time by Hall sensor array and tension sensor. When the deflection angle is too large or the tension value is too large, the winch is automatically controlled to accelerate the release of the load-bearing umbilical cable. When the magnetic ring rebounds to the threshold area or the tension is lower than the threshold, the normal cable release speed is restored.

[0068] If the tension of the load-bearing umbilical cable is too low or the umbilical cable at the tail of the visual recognition device is too long, the load-bearing umbilical cable will be pulled back faster until the visual recognition is normal or the analog value of the tension sensor returns to the threshold range, and the normal cable release speed will be restored.

[0069] S3, determine whether to recycle ROV. If not, continue with steps S1 and S2. If yes, proceed to step S4.

[0070] S4, the ROV begins to return to the dock. Using the recorded path calculated by VSLAM as a reference, the ROV's return path is planned, and the thrusters control the ROV to return to the dock along the planned path.

[0071] This also includes:

[0072] If there are no turning points in the map obtained by VSLAM, the judgment is made directly according to the straight line, and the umbilical cable is retrieved directly while driving; if there is no better regression solution, at this time, while the ROV moves in the reverse direction according to the original calculated path, the underwater camera performs machine vision recognition on the load-bearing umbilical cable to determine the orientation of the load-bearing umbilical cable and the situation of obstacles, so as to achieve a multi-directional ROV regression effect.

[0073] As the ROV approaches the docking station, it identifies visual features to determine its location, such as visually recognizing light beacons on the docking dock and using a docking algorithm to guide the ROV back to dock, achieving more precise recovery.

[0074] S5 uses a winch in conjunction with the ROV's thruster to retrieve the load-bearing umbilical cable, and autonomously adjusts the state of the load-bearing umbilical cable by using underwater camera vision, Hall sensor array, and tension sensor parameters.

[0075] The process of adjusting the condition of the load-bearing umbilical cable includes:

[0076] The tension of the load-bearing umbilical cable is judged in real time by Hall sensor array and tension sensor. When the deflection angle is too large or the tension value is too large, the winch is automatically controlled to release part of the load-bearing umbilical cable. When the magnetic ring rebounds to the threshold area or the tension is lower than the threshold, the normal cable winding speed is restored.

[0077] When the tension of the load-bearing umbilical cable is too low or the umbilical cable at the tail of the visual recognition is too long, the load-bearing umbilical cable is pulled back faster until the visual recognition is normal or the analog value of the tension sensor returns to the threshold range, and the normal cable pulling speed is restored.

[0078] This invention centers on camera recognition, analyzing the umbilical cable's status using a Hall sensor array, tension sensor, and camera. It analyzes the path using a camera, attitude sensor, and either DVL or USBL sensor. Finally, it performs visual recognition of light features upon docking, forming a multi-sensor perception fusion and application. Ultimately, through winch and thruster control, it achieves auxiliary detection and autonomous return functions in confined spaces. Specifically, it combines the attitude sensor's ability to determine its own status with the DVL or USBL's ability to determine relative position to identify the ROV's orientation and path. Then, it uses machine vision, Hall sensor array, and tension sensor parameters for autonomous discrimination and control, forming a multi-sensor perception fusion and application. Finally, through winch and thruster control, it achieves auxiliary detection and autonomous return functions in confined spaces.

Claims

1. A control system for an underwater or under-ice exploration robot used for exploration in confined spaces, comprising a control unit, a winch, an ROV, and a load-bearing umbilical cable, wherein one end of the load-bearing umbilical cable is connected to the ROV and the other end is wound around the winch drum, characterized in that: A magnetic ring is fixed to one end of the load-bearing umbilical cable connected to the ROV. The magnetic ring and the ROV form a linkage structure, generating analog feedback to the control unit. A spring-loaded structure is provided between the magnetic ring and the ROV to reset the magnetic ring. A tension sensor for monitoring the tension of the load-bearing umbilical cable is fixed to one side of the winch. The tension sensor is connected to the control unit. A waterproof and pressure-resistant Hall sensor array, an underwater camera, and an attitude sensor are installed on the ROV. The Hall sensor array, underwater camera, and attitude sensor are all connected to the control unit. The Hall sensor array is arranged in a circular structure and is fixed in a ring-shaped embedded position at the port where the ROV and the load-bearing umbilical cable connect. It is linked with the magnetic ring on the load-bearing umbilical cable to provide real-time feedback on the angle and orientation of the load-bearing umbilical cable connection. The underwater camera is arranged in parallel with the port connecting the ROV and the load-bearing umbilical cable, and is used to capture images of the tail of the ROV and the load-bearing umbilical cable. The control method for underwater or sub-ice exploration robots used for exploration in confined spaces comprises the following steps: S1, the ROV detaches from the docking station and begins exploration. The underwater camera captures images of the ROV's tail. The control unit performs VSLAM on the tail images to acquire feature points and record the path. S2, the winch works in conjunction with the ROV's thruster to release the load-bearing umbilical cable, and autonomously judges and controls the state adjustment of the load-bearing umbilical cable by using the vision of the underwater camera, the Hall sensor array, and the parameters of the tension sensor. The process of adjusting the condition of the load-bearing umbilical cable includes: The tension of the load-bearing umbilical cable is judged in real time by Hall sensor array and tension sensor. When the deflection angle is too large or the tension value is too large, the winch is automatically controlled to accelerate the release of the load-bearing umbilical cable. When the magnetic ring rebounds to the threshold area or the tension is lower than the threshold, the normal cable release speed is restored. When the tension of the load-bearing umbilical cable is too low or the visual recognition tail umbilical cable is too long, the load-bearing umbilical cable is pulled back faster until the visual recognition is normal or the analog value of the tension sensor returns to the threshold range, and the normal cable release speed is restored. S3, determine whether to recycle ROV. If not, continue with steps S1 and S2. If yes, proceed to step S4. S4, the ROV begins to return to the dock. Using the recorded path calculated by VSLAM as a reference, the ROV's return path is planned, and the thrusters control the ROV to return to the dock along the planned path. The S5 winch, in conjunction with the ROV's thruster control, retrieves the load-bearing umbilical cable and autonomously adjusts the cable's condition based on underwater camera vision, Hall sensor array, and tension sensor parameters.

2. The underwater or under-ice exploration robot control system for exploration in confined spaces according to claim 1, characterized in that: The load-bearing umbilical cable is a zero-buoyancy load-bearing cable and contains an internal battery core.

3. The underwater or under-ice exploration robot control system for exploration in confined spaces according to claim 1, characterized in that: The ROV is equipped with a positioning sensor that enables positioning, and the positioning sensor is connected to the control unit.

4. The underwater or under-ice exploration robot control system for exploration in confined spaces according to claim 1, characterized in that: VSLAM is performed in step S1, and the specific steps are as follows: S11 performs pre-processing for defogging and sharpening of images captured by the camera; S12: Feature points in the image are collected using the ORB-SLAM algorithm, the real-time position of the ROV is analyzed, and the movement route is given. The images along the route are processed and maps are constructed to simulate the real-time movement trajectory of the ROV. S13, semantic segmentation of the load-bearing umbilical cable is performed through visual recognition, and the state of the visible part of the load-bearing umbilical cable at the tail is identified separately while eliminating the interference of the load-bearing umbilical cable on path recognition.

5. A control system for an underwater or under-ice exploration robot applied to exploration in confined spaces according to claim 4, characterized in that: In step S12, ORB-SLAM matches ORB feature points from the segmented image, combines data from the attitude sensor and the positioning sensor, pre-assumes the rotation and displacement of the ROV, and estimates pixel points based on the current camera pose change to form a visual odometry for SLAM.

6. The underwater or under-ice exploration robot control system for exploration in confined spaces according to claim 1, characterized in that: Step S4 also includes: If there are no turning points in the map obtained by VSLAM, the judgment is made directly according to the straight line, and the umbilical cable is retrieved directly while driving; if there is no better regression solution, at this time, while the ROV moves in the reverse direction according to the original calculated path, the underwater camera performs machine vision recognition on the load-bearing umbilical cable to determine the orientation of the load-bearing umbilical cable and the situation of obstacles, so as to achieve a multi-directional ROV regression effect. As the ROV approaches the docking station, the camera identifies visual features to determine its location and uses a docking algorithm to guide it back to dock.

Images