An underwater robot capable of autonomously replacing a working tool and a method thereof

Abstract

The application discloses an underwater robot capable of autonomously replacing a working tool and a method thereof, and belongs to the field of underwater robots. The robot main body is vertically stacked from bottom to top by a driving cabin, a battery cabin and a main control cabin, can be spliced with a tool cabin through the top features of the main control cabin; the driving cabin is used for providing power, the battery cabin provides energy support, the control cabin makes decision control, and the light signal beacon on the tool cabin can be identified and pose estimated through the visual perception system of the multiple cameras. The application captures the beacon on the tool through the camera, realizes underwater positioning by combining a pose settlement method, identifies the tool type by reverse projection using the pose information, guides the magnetic attraction module to cooperate in the close-range docking stage, realizes the splicing and autonomous replacement of the tool cabin, and overcomes the single limitation of the traditional robot carrying tools through modular design, realizes the autonomous replacement of the working tool, and significantly improves the operation ability of the underwater robot in complex environments.

Description

Technical Field

[0001] This invention belongs to the field of underwater robots, specifically relating to an underwater robot and its method that can autonomously change working tools. Background Technology

[0002] Underwater robots are intelligent devices that can work independently in underwater environments. They can replace or assist humans in performing tasks in dangerous or difficult-to-access underwater environments, and are therefore widely used in fields such as marine resource exploration, environmental monitoring, and underwater facility maintenance.

[0003] In existing technologies, traditional underwater robots are typically equipped with fixed tools and can only perform simple operations for a single task. However, actual underwater tasks are often complex and involve multiple steps, making it difficult for traditional underwater robots to meet the demands of real-world underwater operations. Furthermore, the tool-changing process for traditional underwater robots is complex and lacks automation, exhibiting a lack of autonomous tool identification and replacement capabilities. This limitation severely restricts the operational flexibility and task coverage of underwater robots.

[0004] Therefore, there is an urgent need for an underwater robot that can autonomously identify tasks, plan routes, and change operating tools according to actual needs, overcoming the limitations of traditional underwater robots and significantly improving the adaptability and operational efficiency of underwater robots in complex multi-task environments. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an underwater robot capable of autonomously changing its working tools. The underwater robot adopts a modular design, with its main body consisting of a drive compartment, a battery compartment, and a main control compartment stacked vertically. The tool compartment is autonomously identified and replaced through a multi-camera vision system on the top of the main control compartment and a magnetic docking mechanism.

[0006] The technical solution adopted in this invention is as follows:

[0007] In a first aspect, the present invention proposes an underwater robot capable of autonomously changing its working tools, comprising a robot body and a tool compartment;

[0008] The robot body consists of a drive compartment, a battery compartment, and a main control compartment arranged vertically from bottom to top;

[0009] The drive cabin is used to provide underwater propulsion and is equipped with a drive control system and an inertial measurement unit.

[0010] The battery compartment is used to provide power to the robot;

[0011] The main control cabin has a built-in main control board and is equipped with a front-side camera, a top camera, a first assembly guide groove, an electromagnet, an electromagnetic signal source, and a Hall sensor; the first assembly guide groove and the electromagnet are located on the top of the main control cabin, and the electromagnet is embedded in the first assembly guide groove.

[0012] The bottom of the tool compartment is equipped with a second assembly guide groove, an iron block, a magnet, an LED array beacon, and a tool compartment signal sensor. The LED array beacon consists of four asymmetrically arranged anchor LEDs and a central 5×3 area of ​​digital dot LEDs. The iron block is embedded in the second assembly guide groove, which is curved to fit the first assembly guide groove to form a mechanical guiding structure. The iron block and the electromagnet are positioned correspondingly. By controlling the on and off of the electromagnet, the tool compartment and the robot body can be connected and separated by adsorption. When the connection is successful, the magnet transmits a magnetic signal to the Hall sensor to provide feedback on the docking status. The tool compartment signal sensor is used to receive the magnetic field signal from the electromagnetic signal source to control the tool in a non-contact manner.

[0013] Furthermore, the LED array beacon is a 5×5 array, and the anchor LEDs and digital dot LEDs are marked with colors of different wavelengths.

[0014] Furthermore, the front-side camera is a binocular RGB wide-angle camera, used for mid-to-long-range identification of the LED array beacon and autonomous obstacle avoidance; the top camera is a monocular RGB wide-angle camera, used to provide vertical top-down pose deviation feedback during close-range docking.

[0015] Furthermore, the drive control system of the drive bay includes multiple horizontally and vertically arranged motors, an electronic speed controller, and a PWM signal generator board; the motors are connected to the electronic speed controller via aviation connectors; the electronic speed controller is connected to the power interface of the battery compartment via an ESC power plug; the PWM signal generator board 17 is connected to the electronic speed controller via a signal line, and is used to receive control commands issued by the main control board and generate PWM signals, thereby adjusting the speed and direction of each motor through the electronic speed controller.

[0016] Furthermore, the main control board inside the main control cabin is also equipped with a GPU-accelerated image processing unit, which is used to perform HSV color space conversion and extract specific color pixels from the images captured by the camera.

[0017] Furthermore, each sealed compartment of the robot body is sealed by a waterproof groove protruding from the edge of the compartment shell and a groove with a silicone waterproof strip. The drive compartment is also equipped with a counterweight for adjusting the robot's center of gravity.

[0018] Furthermore, a power switch is also provided on the top of the main control cabin to control the start-up and power-off of the entire robot system.

[0019] Secondly, this invention proposes a splicing method for autonomously changing operational tools based on the aforementioned underwater robot, comprising:

[0020] Step 1: Capture images of the LED array beacons on the tool bay using the front-side camera of the main control cabin, and extract the pixel centroid coordinates of the anchor LED and the digital dot LED;

[0021] Step 2: Based on the pixel centroid coordinates of the anchor LEDs, calculate the relative pose of the robot body with respect to the tool compartment using the PnP algorithm;

[0022] Step 3: Locate the digital region in the image using the centroid coordinates of the LED pixels, and inversely project the preset digital sampling grid onto the location region using the relative pose. Extract the tool cabin ID through brightness sampling and binarization to confirm the tool type.

[0023] Step 4: As the robot continues to approach the tool compartment based on the relative pose obtained in Step 2, when the distance information in the relative pose is lower than the preset threshold, the visual control is switched from the front camera to the top camera, and the robot pose is finely adjusted based on the pose data provided by the top camera to continuously guide the alignment of the assembly guide slot. When the pose deviation meets the preset conditions, the electromagnet is energized to attract the iron block and complete the splicing.

[0024] Step 5: After successful assembly, the magnetic signal is transmitted from the magnet in the tool compartment to the Hall sensor in the main control compartment to provide feedback on the docking status, and control commands are sent from the electromagnetic signal source in the main control compartment to the tool compartment signal sensor in the tool compartment.

[0025] Furthermore, the process of extracting the pixel centroid coordinates of the anchor LED and the digital LED in step 1 includes: first, performing HSV color space conversion on the captured image, and then extracting LED pixels of specific colors by calculating the hue components and combining them with the saturation threshold, thereby obtaining the pixel centroid coordinates of the anchor LED and the digital LED respectively.

[0026] Furthermore, step 2 also includes: inputting the calculated relative pose data into a first-order linear prediction filter, predicting the pose of the next frame by recording and compensating for the instantaneous rate of change of the pose, and compensating for instantaneous frame loss caused by light flickering or occlusion.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] (1) The underwater robot of the present invention can autonomously change its working tools and adopts a modular tool compartment design, which overcomes the limitation of traditional robots that can only carry a single tool; it can autonomously change the mounted tools according to different task requirements, which significantly improves the working range and efficiency in complex water environments.

[0029] (2) The underwater robot of the present invention has a high degree of automation. By combining visual recognition and magnetic locking, it realizes the precise and autonomous replacement of the tool compartment. Compared with traditional mechanical buckles and manual intervention, the present invention is more automated and faster and more stable, greatly enhancing its reliability in the underwater environment. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the external structure of an underwater robot capable of autonomously changing its operating tools.

[0031] Figure 2 This is a schematic diagram of the underwater robot's drive cabin.

[0032] Figure 3 This is a schematic diagram of the underwater robot's drive cabin from another perspective;

[0033] Figure 4 This is a structural diagram of the underwater robot's battery compartment;

[0034] Figure 5 This is a structural schematic diagram of the main control cabin of the underwater robot;

[0035] Figure 6 This is a structural diagram of the tool compartment carried by the underwater robot;

[0036] In the diagram, 1-Main control cabin shell, 2-Power switch, 3-Top camera, 4-Battery compartment shell, 5-Propeller protective cover, 6-Drive cabin shell, 7-Motor, 8-Front-side camera, 9-Assembly guide groove, 10-Electromagnetic signal source, 11-Hall sensor, 12-Electromagnet, 13-Cable management hole, 14-Aerospace connector recess, 15-Electronic speed controller, 16-Electronic speed controller power plug, 17-PWM signal generator board, 18-Counterweight, 19-Inertial measurement unit, 20-Drive cabin signal interface, 21-Drive cabin power interface, 22-Battery, 23-Main control board, 24-Iron block, 25-Tool power supply battery, 26-Tool side assembly guide groove, 27-Tool cabin signal sensor, 28-Magnet. Detailed Implementation

[0037] The present invention will be further described and illustrated below with reference to specific embodiments. The embodiments described are merely examples of the content of this disclosure and do not limit the scope of the invention. The technical features of each embodiment in the present invention can be combined accordingly, provided that there is no mutual conflict.

[0038] like Figure 1As shown, this invention provides an underwater robot capable of autonomously changing its working tools. The robot's shell is hexagonal prism-shaped. The main body of the robot is composed of sealed compartments stacked vertically, from bottom to top: a drive compartment, a battery compartment, and a main control compartment. The top layer of the robot body is the main control compartment, covered by a main control compartment shell 1. A power switch 2 and a top camera 3 are installed on its top. The power switch 2 is used for system start-up and power-off control, and the top camera 3 is used for monitoring the water above and guiding close-range docking. The middle layer is the battery compartment, protected by a battery compartment shell 4, which houses batteries 22 to provide energy support for the entire robot. The bottom layer is the drive compartment, sealed by a drive compartment shell 6, with a propeller protective cover 5 and a motor 7 on the outside. The drive compartment is responsible for providing power for underwater movement. This underwater robot can be spliced ​​with the tool compartment through the top feature structure of the main control compartment to expand its operational capabilities.

[0039] In one specific embodiment of the present invention, the sealed compartment is waterproofed by the raised waterproof grooves and recesses on the edge of the compartment shell. The recesses are provided with silicone waterproof strips that can fit into the waterproof grooves on the edge of the compartment to achieve a waterproof effect and ensure the stability of each compartment in the underwater environment.

[0040] like Figure 2 and Figure 3 The figures shown are schematic diagrams of the underwater robot's propulsion cabin from different perspectives. The propulsion cabin is sealed by an outer shell 6 and houses the propulsion control system. The outer side of the propulsion cabin is enclosed by an integrated propeller protective cover 5 to protect the internal components from impacts. Figure 3 As shown, an inertial measurement unit 19 is installed at the bottom of the drive cabin to monitor the robot's attitude and motion data in real time. Two horizontal and four vertical motors 7 are also arranged at the bottom, forming a multi-degree-of-freedom power array. Figure 2 As shown, the drive compartment includes cable management holes 13, an aviation connector recess 14, an electronic speed controller 15, an ESC power plug 16, a PWM signal generator board 17, and a counterweight 18. The aviation connector recess 14 is located on the inner wall or a recessed area of ​​the drive compartment shell 6, and is used to install a waterproof aviation connector. The motor 7 connects to the electronic speed controller 15 through the aviation connector installed in the aviation connector recess 14. The electronic speed controller 15 is installed adjacent to the aviation connector recess 14 and receives power from the battery compartment through the ESC power plug 16. The PWM signal generator board 17 is fixed to the inner wall of the drive compartment and connected to the electronic speed controller 15 via a signal line. It is responsible for converting the commands from the main control board 23 of the autonomous control cabin into PWM signals. The electronic speed controller 15 receives the PWM control signals provided by the PWM signal generator board 17 and precisely adjusts the motor speed. The cable management holes 13 are distributed on the inner wall or partition of the compartment, used to centrally organize motor cables and power cords to prevent tangling. The counterweight 18 is fixed at the bottom of the drive compartment near the inertial measurement unit 19, and is used to dynamically adjust the robot's center of gravity to improve the stability of underwater movement.

[0041] like Figure 4 The diagram shows the structure of the underwater robot's battery compartment. Located below the main control compartment and sealed by a battery compartment shell 4, the battery compartment houses two sets of batteries 22 arranged side-by-side. The batteries 22 supply power to the main control compartment upwards and to the drive compartment downwards via a drive compartment power interface 21 located at the bottom of the compartment. Additionally, a drive compartment signal interface 20 is provided in the battery compartment, allowing the main control board 23 in the main control compartment to connect to the inertial measurement unit 19 located in the drive compartment via the drive compartment signal interface 20. The battery compartment's design ensures compactness and safety, supporting extended underwater operations.

[0042] like Figure 5 The diagram shows the structure of the underwater robot's main control cabin. The main control cabin is sealed by a main control cabin shell 1. The top of the main control cabin houses a power switch 2, a top camera 3, an assembly guide groove 9, an electromagnetic signal source 10, and a Hall sensor 11. An electromagnet 12 is embedded in the assembly guide groove 9 to connect to the tool cabin via magnetic attraction. The electromagnetic signal source 10 transmits control commands to the tool cabin by emitting magnetic field signals (controlling the start and stop of the tools carried in the tool cabin). The Hall sensor 11 receives magnetic signals from the magnet 28 inside the tool cabin, providing feedback on successful assembly. The top camera 3 is a monocular RGB wide-angle camera used for monitoring the water above and providing pose feedback during close-range docking. A front-side camera 8, a binocular RGB wide-angle camera, is located at the front of the main control cabin and is used for the robot to identify LED array beacons on the tool cabin at medium to long distances and for autonomous obstacle avoidance. The main control cabin houses a main control board 23, which connects to the inertial measurement unit 19 of the drive cabin via a drive cabin signal interface 20 to acquire real-time attitude data and processes the images captured by the camera for pose calculation.

[0043] like Figure 6The diagram shows the structure of the tool compartment carried by the underwater robot. The tool compartment is a modular component of the underwater robot. This diagram illustrates the essential features of the tool compartment but does not show specific operating tools. An iron block 24 is located at the bottom of the tool compartment for connecting to an electromagnet 12 mounted on the top of the main control compartment. The assembly and separation of the tool compartment and the robot are controlled by energizing and de-energizing the electromagnet 12. An assembly guide groove 26 is located at the bottom of the tool compartment, which conforms to the curvature of the assembly guide groove 9 on the top of the main control compartment, forming a mechanical guiding structure to guide the robot and the tool compartment during assembly, ensuring precise alignment. A magnet 28 is located inside the tool compartment. When the tool compartment and the main control compartment are successfully assembled, the magnet 28 transmits a magnetic signal to a Hall sensor 11 inside the main control compartment to confirm the docking status. A tool compartment signal sensor 27 is also located inside the tool compartment to receive magnetic field signals from an electromagnetic signal source 10 inside the main control compartment, enabling non-contact control of the start and stop of the tools carried in the tool compartment. The tools carried in the tool compartment are powered by a tool power supply battery 25. In this embodiment, a specific 5×5 LED array beacon is provided at the bottom of the tool compartment. The beacon consists of four asymmetrically arranged anchor point LEDs and digital dot LEDs in the central 5×3 area, and the two types of LEDs are distinguished by different wavelength colors.

[0044] In response to the above-mentioned robot structure, the present invention provides a method for autonomously changing working tools for the underwater robot. The core of this method lies in recognizing and estimating the pose of the LED array beacon on the tool compartment with high precision through a multi-camera collaborative visual perception system.

[0045] The multi-camera collaborative visual perception system consists of a front-side camera 8 mounted on the front of the main control cabin shell 1 and a top-side camera 3 mounted on the top. The front-side camera 8 is a binocular RGB wide-angle camera, primarily responsible for capturing image signals of the LED array on the target tool cabin at medium to long distances. The top-side camera 3 is a monocular RGB wide-angle camera, capturing the light beacon features at the bottom of the tool cabin in real time during close-range docking, providing high-precision pose feedback for stitching. During image signal recognition, the main control board 23 performs real-time color space conversion on the acquired images, converting them from RGB to HSV space. By calculating the hue components and combining them with saturation thresholds, it accurately extracts LED pixels of specific colors, thereby effectively filtering stray light interference from underwater suspended particles and background environmental noise.

[0046] During the autonomous tool changing process, the robot first uses the front-side camera 8 to identify four asymmetrically distributed anchor points in the LED array to determine the spatial pose of the tool compartment. Because the physical arrangement of the anchor points has asymmetrical geometric features, the algorithm uniquely determines the correspondence between the array's two-dimensional features and the three-dimensional model through geometric topological sorting. The main control board 23 combines the known 3D model coordinates with the 2D pixel centroid coordinates in the image coordinate system and uses the Perspective-n-Point (PnP) algorithm to calculate the robot's rotation vector relative to the tool compartment. Translation vector It can be based on translation vectors. The robot obtains real-time distance information from the tool compartment. To ensure stability in the dynamic underwater environment, the pose data is processed by a first-order linear prediction filter. By recording and compensating for the instantaneous rate of change of rotation and translation, the pose of the next frame is predicted in real time, thus effectively dealing with instantaneous frame loss caused by light flicker or occlusion.

[0047] After establishing stable pose feedback, the main control module uses the calculated rotation vector Translation vector The system inversely projects a preset 5×3 digital sampling grid onto the current 2D image plane. Local brightness sampling is performed at the projected pixel coordinates, and adaptive binarization is applied based on the extreme brightness values ​​within the region to obtain a binary sequence representing the tool's identity. This sequence is then matched against a local digital template library to confirm the current toolbox's function type.

[0048] As the robot approaches the tool compartment and enters the area below it, the main control board 23 smoothly switches the visual perception weight from the front camera 8 to the top camera 3 according to a preset distance threshold. In this embodiment, the visual perception weight is a weight assigned to the front camera 8. When the distance d is greater than the first threshold, the weight is 1, and the pose data is entirely provided by the front camera 8. When d decreases to the switching range, the weight decreases linearly with the distance, and the system comprehensively calculates the outputs of the front and top cameras 3 through a weighted fusion algorithm (such as the EKF algorithm). When d is lower than the second threshold, the weight is 0, and the visual weight is completely switched to the top camera 3. The top camera 3 of this invention captures the LED array from a vertical top-down perspective, eliminating perspective distortion caused by large-angle tilt, and providing millimeter-level pose deviation feedback for final alignment. The main control board 23 generates motion control signals based on the high-frequency data collected by the top camera 3, driving the motor 7 to fine-tune the robot's horizontal coordinates and yaw angle, ensuring that the assembly guide groove 9 on the bottom of the robot and the tool-side assembly guide groove 26 are precisely aligned axially. When the distance and angle deviation measured by the top camera 3 decreases to within the preset threshold, the main control module determines that the physical docking conditions have been met. Then, it triggers the electromagnet 12 to generate a strong magnetic force, which attracts the iron block 24 on the tool compartment. Together with the assembly guide groove 9 and the tool-side assembly guide groove 26, the tool is stably spliced, and finally the autonomous replacement of the working tool is completed.

[0049] The above method for autonomously changing work tools can be summarized into the following specific steps:

[0050] Step 1: The main control board 23 acquires the binocular RGB image captured by the front camera 8 in real time, performs HSV color space conversion on the image, calculates the hue component and combines it with the saturation threshold, and extracts the pixel centroid coordinates of the anchor LED and digital LED in the LED array beacon from the complex underwater background, effectively filtering background stray light interference.

[0051] Step 2: Based on the asymmetric geometric distribution characteristics of the anchor LEDs, the main control module establishes a unique correspondence between 2D image points and 3D physical model points through topological sorting. The Perspective-n-Point (PnP) algorithm is used to calculate the robot's relative pose to the tool bay. The calculated pose data is input into a first-order linear prediction filter to compensate for momentary frame drops caused by light flickering or occlusion, ensuring the continuity and stability of the motion control signal.

[0052] Step 3: Using the pose information obtained in Step 2, the main control module inversely projects a preset 5×3 digital sampling grid onto the current 2D image plane. By performing brightness sampling and adaptive binarization on the projected point matrix, a binary sequence representing the tool's identity ID is extracted and matched with the local template library to confirm the current tool compartment's functional type. Simultaneously, the main control module compares the current pose calculated in Step 2 with the preset docking target pose, generating pose deviation data. Based on this deviation data, the main control module generates a preliminary PWM control signal, which is sent to the electronic speed controller 15 via the PWM signal generator board 17. This drives the motor 7 to adjust the robot's pose, performing preliminary approach and alignment towards the bottom of the tool compartment, preparing for close-range precision docking.

[0053] Step 4: Based on real-time distance feedback, the main control module smoothly switches the visual weight from the front camera 8 to the top camera 3. The top camera 3 provides millimeter-level pose deviation data from a vertical top-down perspective. The main control board 23 generates a PWM control signal based on the pose deviation data, which is sent to the electronic speed controller 15 via the PWM signal generator board 17 to drive the motor 7 to fine-tune the robot's pose (horizontal coordinates and yaw angle). During the fine-tuning process, the assembly guide groove 9 on the top of the main control compartment is continuously guided to precisely align with the tool-side assembly guide groove 26 at the bottom of the tool compartment in the axial direction. When the deviations measured by the top camera 3 drop to the preset threshold, the main control module determines that the docking conditions have been met, triggers the electromagnet 12 to be energized, generating magnetic force to attract the iron block 24 at the bottom of the tool compartment, and completes a stable splicing in conjunction with the mechanical guide structure.

[0054] When the assembly is successful, the magnet 28 inside the tool compartment generates a magnetic signal. The Hall sensor 11 on the top of the main control compartment detects the magnetic signal in real time and feeds back the assembly success status to the main control board 23.

[0055] Step 5: After confirming successful assembly, the main control cabin emits a magnetic field signal through the electromagnetic signal source 10 on top. The tool cabin signal sensor 27 inside the tool cabin receives the magnetic field signal and interprets the command to control the start or stop of the tool carried in the tool cabin.

[0056] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. An underwater robot capable of autonomously changing its working tools, characterized in that, Including the robot body and tool compartment; The robot body consists of a drive compartment, a battery compartment, and a main control compartment arranged vertically from bottom to top; The drive cabin is used to provide underwater propulsion and is equipped with a drive control system and an inertial measurement unit. The battery compartment is used to provide power to the robot; The main control cabin has a built-in main control board and is equipped with a front-side camera, a top camera, a first assembly guide groove, an electromagnet, an electromagnetic signal source, and a Hall sensor; the first assembly guide groove and the electromagnet are located on the top of the main control cabin, and the electromagnet is embedded in the first assembly guide groove. The bottom of the tool compartment is equipped with a second assembly guide groove, an iron block, a magnet, an LED array beacon, and a tool compartment signal sensor. The LED array beacon consists of four asymmetrically arranged anchor LEDs and a central 5×3 area of ​​digital dot LEDs. The iron block is embedded in the second assembly guide groove, which is curved to fit the first assembly guide groove to form a mechanical guiding structure. The iron block and the electromagnet are positioned correspondingly. By controlling the on and off of the electromagnet, the tool compartment and the robot body can be connected and separated by adsorption. When the connection is successful, the magnet transmits a magnetic signal to the Hall sensor to provide feedback on the docking status. The tool compartment signal sensor is used to receive the magnetic field signal from the electromagnetic signal source to control the tool in a non-contact manner.

2. The underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, The LED array beacon is a 5×5 array, and the anchor LEDs and digital dot LEDs are marked with different wavelengths of color.

3. The underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, The front-side camera is a binocular RGB wide-angle camera used for mid-to-long-range identification of the LED array beacon and autonomous obstacle avoidance; the top camera is a monocular RGB wide-angle camera used to provide vertical top-down pose deviation feedback during close-range docking.

4. The underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, The drive control system of the drive bay includes multiple horizontally and vertically arranged motors, an electronic speed controller, and a PWM signal generator board. The motors are connected to the electronic speed controller via aviation connectors. The electronic speed controller is connected to the power interface of the battery compartment via an ESC power plug. The PWM signal generator board is connected to the electronic speed controller via a signal line and is used to receive control commands from the main control board and generate PWM signals to adjust the speed and direction of each motor through the electronic speed controller.

5. An underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, The main control board inside the main control cabin is also equipped with a GPU-accelerated image processing unit, which is used to perform HSV color space conversion and extract specific color pixels from the images captured by the camera.

6. An underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, Each sealed compartment of the robot body is sealed by a waterproof groove protruding from the edge of the compartment shell and a groove with a silicone waterproof strip. The drive compartment is also equipped with a counterweight for adjusting the robot's center of gravity.

7. An underwater robot capable of autonomously changing its working tools according to claim 1, characterized in that, The top of the main control cabin is also equipped with a power switch, which is used to control the start-up and power-off of the entire robot system.

8. A splicing method for autonomously changing working tools of the underwater robot according to claim 1, characterized in that, include: Step 1: Capture images of the LED array beacons on the tool bay using the front-side camera of the main control cabin, and extract the pixel centroid coordinates of the anchor LED and the digital dot LED; Step 2: Based on the pixel centroid coordinates of the anchor LEDs, calculate the relative pose of the robot body with respect to the tool compartment using the PnP algorithm; Step 3: Locate the digital region in the image using the centroid coordinates of the LED pixels, and inversely project the preset digital sampling grid onto the location region using the relative pose. Extract the tool cabin ID through brightness sampling and binarization to confirm the tool type. Step 4: As the robot continues to approach the tool compartment based on the relative pose obtained in Step 2, when the distance information in the relative pose is lower than the preset threshold, the visual control is switched from the front camera to the top camera, and the robot pose is finely adjusted based on the pose data provided by the top camera to continuously guide the alignment of the assembly guide slot. When the pose deviation meets the preset conditions, the electromagnet is energized to attract the iron block and complete the splicing. Step 5: After successful assembly, the magnetic signal is transmitted from the magnet in the tool compartment to the Hall sensor in the main control compartment to provide feedback on the docking status, and control commands are sent from the electromagnetic signal source in the main control compartment to the tool compartment signal sensor in the tool compartment.

9. The splicing method for autonomously changing work tools according to claim 8, characterized in that, The process of extracting the pixel centroid coordinates of anchor LED and digital LED in step 1 includes: first, performing HSV color space conversion on the captured image, and then extracting LED pixels of specific colors by calculating the hue component and combining it with the saturation threshold, thereby obtaining the pixel centroid coordinates of anchor LED and digital LED respectively.

10. The splicing method for autonomously changing work tools according to claim 8, characterized in that, Step 2 also includes: inputting the calculated relative pose data into a first-order linear prediction filter, and predicting the pose of the next frame by recording and compensating for the instantaneous rate of change of pose, thus compensating for instantaneous frame loss caused by light flickering or occlusion.

Images