An autonomously detectable underwater hull cleaning robot and method

Abstract

The application discloses an underwater ship body cleaning robot capable of autonomous detection and a method thereof. The underwater ship body cleaning robot comprises a rack, at least two mechanical arms, a propeller, a cleaning module, a camera and a control module arranged on the rack, a vacuum pump connected with a cavitation jet nozzle through a water pipe, an electromagnetic adsorption device arranged at the tail end of the mechanical arm, a control driving module connected with the camera, the mechanical arm, the propeller and the electromagnetic adsorption device, and the cleaning module comprising the cavitation jet nozzle. The application can be flexibly adsorbed and climbed on the ship body, can be applied to uneven ship body surfaces, ensures the smooth progress of the cleaning process, has high flexibility, has no absolute requirement for the environment where the ship body is located, increases the application range, and can check whether important components are adsorbed and attached and clean in time during a long voyage.

Description

Technical Field

[0001] This invention specifically relates to an underwater hull cleaning robot and method capable of autonomous detection. Background Technology

[0002] Currently, with the continuous development of domestic technology, most underwater hull cleaning methods in China have evolved from simple dock cleaning to more advanced methods such as manual underwater cleaning, cleaning with large-scale cleaning equipment, or cleaning with manually controlled underwater robots. These methods are better suited to the complexities and harsh environments of underwater operations and have improved work efficiency. However, these methods still suffer from problems such as low efficiency, high labor costs, incomplete cleaning, demanding cleaning conditions, and inconvenience. Therefore, to address these issues, this invention provides a highly efficient, comprehensive, and autonomously detectable underwater hull cleaning robot. Summary of the Invention

[0003] The purpose of this invention is to provide an underwater hull cleaning robot and method that can autonomously detect and climb onto the hull. It is applicable to uneven hull surfaces, ensuring the smooth progress of the cleaning process. It has high flexibility and no absolute requirements on the environment of the hull, thus increasing its applicability. It can also check whether important components are adsorbed and cleaned in a timely manner during long voyages.

[0004] The technical solution adopted in this invention is:

[0005] An autonomous underwater hull cleaning robot includes a frame, and at least two robotic arms, a propeller, a cleaning module, a camera, and a control module mounted on the frame. The ends of the robotic arms are equipped with electromagnetic adsorption devices. The control and drive module is connected to the cleaning module, the camera, the robotic arms, the propeller, and the electromagnetic adsorption devices. The cleaning module includes cavitation jet nozzles.

[0006] Preferably, the frame includes two side plates and a bottom plate. The two side plates are arranged side by side, and the two ends of the bottom plate are respectively connected to the inner sides of the two side plates. A support rod is also connected between the two side plates.

[0007] Preferably, a cylindrical compartment is provided on the bottom plate, and the control module is located inside the cylindrical compartment.

[0008] Preferably, there are two robotic arms, which are arranged on both sides of the frame.

[0009] Preferably, there are 3 to 5 propellers.

[0010] Preferably, the cleaning module further includes a vacuum pump and a water pipe, the cavitation jet nozzle is mounted on the frame, and the vacuum pump is connected to the cavitation jet nozzle through the water pipe.

[0011] Preferably, the control module includes a main controller and a slave controller. The main controller is set on the water surface and connected to the slave controller via a cable. The slave controller is set in a cylindrical compartment on the frame. The main controller is the host computer and the slave controller is the slave computer. The slave controller is connected to the robotic arm, the propeller thruster, the cleaning module, and the camera.

[0012] Preferably, the rack is also equipped with lighting equipment and a light vision computer, which is connected to a camera.

[0013] A method for cleaning underwater hull attachments using an autonomous underwater hull cleaning robot as described above involves manipulating the underwater hull cleaning robot to move to one side of the underwater hull to be cleaned area and capturing images of the underwater hull attachments in the area to be cleaned using a camera.

[0014] By using a trained neural network model based on a fusion attention mechanism and U-Net architecture, the underwater hull images collected by the robot are preprocessed and the features of the underwater hull attachments are extracted. After analysis and processing, the location of the underwater hull attachments is identified.

[0015] The cavitation jet nozzles of the cleaning module are used to gradually remove the deposits on the underwater hull, starting from the edge and root of the deposits.

[0016] Preferably, the specific process of preprocessing the underwater hull images collected by the robot is as follows: the underwater images collected by the robot are converted to grayscale and enhanced; the images are segmented using digital image processing and detection technology; the images are segmented using threshold-based methods and edge detection methods; the images are segmented using the Otsu algorithm with morphological processing to filter out noise and extract image information.

[0017] The beneficial effects of this invention are:

[0018] This invention utilizes a robotic arm and its end-effector electromagnetic adsorption device to flexibly adhere to and climb onto the hull, making it suitable for uneven hull surfaces. This provides better stability for the underwater hull cleaning robot and ensures a smooth cleaning process. The cleaning module's cavitation jet nozzles effectively remove deposits from the hull, offering high flexibility. Powered by a propeller, the underwater hull cleaning robot can perform cleaning while the ship is moored, without strict requirements on the hull's environment, thus expanding its applicability. Even during long voyages, it can inspect important components for deposits and remove them promptly. It can autonomously perform inspection and cleaning tasks with only simple manual guidance, demonstrating a high level of intelligence. Attached Figure Description

[0019] Figure 1 This is a perspective view of the robot in an embodiment of the present invention;

[0020] Figure 2This is a cross-sectional view of the robot in an embodiment of the present invention;

[0021] Figure 3 This is a perspective view of the underwater hull cleaning robot capable of autonomous detection in an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of the process for cleaning attachments on a ship hull using the autonomous underwater hull cleaning robot described in this embodiment of the invention.

[0023] Figure 5 This is a schematic diagram of the movement process of the underwater hull cleaning robot in an embodiment of the present invention.

[0024] In the diagram: 1.1-Electromagnetic adsorption device, 1.2-Robotic arm; 2.1-Frame, 2.2-Cavitation jet nozzle, 2.3-Crossbar, 2.4-Support rod, 2.5-Cylindrical chamber, 2.6-Connecting rod, 2.7-Base plate, 2.8-Propeller; 3.1-Vacuum pump, 3.2-Water pipe, 3.3-Robot. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0026] In the description of this invention, it should be understood that if terms such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise" are used to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, they 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, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" 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 communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0028] Example 1

[0029] An underwater hull cleaning robot capable of autonomous detection, such as Figures 1-3 As shown, it includes a frame 2.1, and at least two robotic arms 1.2, a propeller 2.8, a cleaning module, a camera, and a control module mounted on the frame 2.1. The end of the robotic arm 1.2 is provided with an electromagnetic adsorption device 1.1. The control drive module is connected to the camera, the robotic arm 1.2, the propeller 2.8, and the electromagnetic adsorption device 1.1.

[0030] Furthermore, the frame 2.1 includes two side plates and a bottom plate 2.7. The two side plates are arranged side by side, and the two ends of the bottom plate 2.7 are respectively connected to the inner sides of the two side plates. A support rod is also connected between the two side plates.

[0031] Furthermore, strut 2.4 is a round rod.

[0032] Furthermore, a cylindrical compartment 2.5 is provided on the base plate 2.7, and the control module is located inside the cylindrical compartment 2.5.

[0033] Multiple connecting rods 2.6 are provided below the base plate 2.7, and the two ends of the connecting rods 2.6 are connected to the two side plates respectively.

[0034] Furthermore, a crossbar 2.3 is connected between the two side plates, and a cavitation jet nozzle 2.2 is disposed on the crossbar 2.3.

[0035] Furthermore, there are two robotic arms 1.2, which are arranged on both sides of the frame 2.1.

[0036] Furthermore, the symmetrically mounted robotic arms 1.2 serve as adsorption modules, used to attach an electromagnetic adsorption device 1.1 with controllable adsorption force to the surface of the ship by symmetrically mounting the robotic arms 1.2 on both sides of the frame 2.1.

[0037] Furthermore, the number of propeller thrusters 2.8 is 3 to 5.

[0038] Furthermore, the cleaning module includes a cavitation jet nozzle 2.2, a vacuum pump 3.1, and a water pipe 3.2. The cavitation jet nozzle 2.2 is mounted on a crossbar 2.3 on the frame 2.1, and the vacuum pump 3.1 is connected to the cavitation jet nozzle 2.2 through the water pipe 3.2. The cleaning module is used to clean the area to be cleaned by means of cavitation jet principle.

[0039] Furthermore, the control module includes a main controller and a slave controller. The main controller is set on the water surface and connected to the slave controller via a cable. The slave controller is set inside the cylindrical compartment 2.5 on the frame 2.1. The main controller is the host computer and the slave controller is the slave computer. The slave controller is connected to the robotic arm 1.2, the propeller thruster 2.8, the cleaning module, and the camera.

[0040] The control module is used to control the adsorption module, drive module, motion module, detection module, and cleaning module via a microcontroller.

[0041] The master-slave control model, consisting of a host computer and a slave controller, is the most common control method currently adopted for ROVs. The host controller, or upper-level computer, primarily controls the motion of the thrusters, and also handles emergency response and logical operations. The slave controller, or lower-level computer, is responsible for executing commands from the host computer and providing feedback on the status of the underwater thrusters and other equipment. Cables maintain the communication between the host and slave controllers. The control system mainly controls the ROV's attitude, movement path, and actuators in the water, including the surface support, control unit, data acquisition device, communication device, power management module, and propulsion system module. Given the unpredictable nature of the marine environment, ROVs need a reliable system capable of handling various situations. This invention utilizes a computer as the host computer, primarily for command transmission, data reception, data storage, and early warning functions. The host computer is designated as the server, and the robot is connected via cables. The thrusters are controlled via input devices such as handles or the main controller. Onshore touchscreens and displays provide real-time monitoring of the ROV's motion status, allowing for adjustments to the thrusters to achieve optimal underwater robot operation.

[0042] The underwater robot, acting as the lower-level controller, primarily performs data acquisition, storage, and transmission functions. Upon receiving instructions from the upper-level computer, it changes its speed and direction according to a pre-programmed sequence. Simultaneously, the camera transmits video data to the upper-level computer for real-time monitoring.

[0043] Furthermore, the rack 2.1 is also equipped with lighting equipment and a light vision computer, which is connected to a camera.

[0044] Example 2

[0045] A method for cleaning debris from a ship's hull using an autonomous underwater hull cleaning robot as described above, such as... Figures 4-5 As shown, the underwater hull cleaning robot is manipulated to move to one side of the underwater hull to be cleaned area, and the underwater hull attachments in the area to be cleaned are captured by the camera.

[0046] The underwater hull images collected by the robot are preprocessed using a trained neural network model based on a fusion attention mechanism and U-Net architecture. Features of underwater hull attachments are extracted and analyzed to determine the location of the attachments. Once the location is determined, the robot adjusts its posture using the adsorption robotic arm 1.2 to align the cavitation jet nozzle with the root of the attachment's edge.

[0047] The cavitation jet nozzle 2.2 of the cleaning module cleans the underwater hull from the edge root to remove the deposits.

[0048] Furthermore, the specific preprocessing process for the underwater hull images acquired by the robot is as follows: The underwater images acquired by the robot undergo image grayscale conversion and image enhancement; digital image processing and detection techniques are used for image segmentation, employing threshold-based and edge detection segmentation methods; morphological processing utilizes the Otsu algorithm to segment the image, filter noise, and extract image information; image enhancement, combined with threshold segmentation, further improves the ability to recognize barnacles and other similar objects; this involves several steps including image grayscale conversion, linear transformation for image enhancement, and morphological processing. Image grayscale conversion reduces the memory footprint of the acquired underwater images, improving computational efficiency; image enhancement improves image clarity and contrast; morphological processing uses the Otsu algorithm to segment the image, filter noise, and extract image information.

[0049] Connection method: The main body of this device is connected by two identical frames 2.1 with rods; four propellers 2.8 that control the forward and backward movement of the machine are installed on the frames 2.1 respectively; the adsorption module is symmetrically placed outside the two frames 2.1; a plate is fixed by the rods inside the frames 2.1, and the control module and detection module are placed on the top; a propeller 2.8 that controls the left and right movement of the machine is installed below the frames 2.1; the cavitation jet device nozzle is fixed above the control module by another rod, and the hose at the rear is connected to the pump body.

[0050] Image acquisition and processing

[0051] The equipment required for image acquisition includes: an underwater camera, lighting equipment, a visual computer (for underwater image acquisition, image enhancement, and recognition), and a control computer (for making decisions and planning based on the observation results, and controlling the robot's next actions).

[0052] This paper proposes an automatic identification technique based on deep learning, focusing on underwater attachments (mainly barnacles) to ship hulls. The technique constructs a neural network model that integrates an attention mechanism and a U-Net architecture. By capturing subtle features of underwater attachment images collected by a robot, it first performs specialized edge enhancement on the acquired images to detect barnacle edges, and then processes and analyzes attachments of different sizes and shapes. Compared with existing fully convolutional networks and U-Net models, this approach significantly improves recognition accuracy and can effectively enhance the detection capability of attachments to ship hulls.

[0053] (I) Dataset

[0054] A dataset was established to determine the criteria for identifying attachments. Experimental data, namely photographs of ship hull attachments in underwater environments, was acquired through online and real-world methods. To ensure data quality, images with excessive noise or irregular shapes were first excluded, resulting in a dataset containing 300 images for each of various environmental conditions. All images were adjusted to a uniform resolution to ensure model compatibility and efficient training.

[0055] The overall dataset undergoes data augmentation followed by random shuffling to reduce potential bias or sequence effects. It is then split into a 6:2:2 ratio. This ratio is widely adopted in dataset splitting practices across multiple domains because it balances the needs of model training with the validation of generalization ability. Specifically, 60% of the data is used as the training set to learn model parameters; 20% is used as the validation set for adjusting and optimizing hyperparameters to avoid overfitting; and the remaining 20% ​​is used as the test set to evaluate the model's generalization ability on unknown data, thus simulating performance in real-world applications. Despite this meticulous approach, it's important to note that the dataset comes from a public repository and may contain inherent biases.

[0056] (II) Enhancing extraction capabilities based on deep learning and other methods

[0057] This system utilizes a deep neural network structure to automatically extract complex features from images, employing a deep feature extractor to identify details such as barnacle edges, width, position, and brightness in underwater images that are difficult to capture directly. Deep learning algorithms are used for the automatic identification of ship attachments; multi-scale deep convolutional feature fusion improves the efficiency of attachment feature extraction, and supervised learning methods adjust the sample ratio to address data imbalance. Furthermore, by incorporating residual connections into the barnacle model on the ship hull, the system enhances its ability to extract barnacle features.

[0058] After extraction is completed, the accuracy of the constructed model, i.e. the trained neural network model, is judged based on the accuracy rate.

[0059] For classification performance, accuracy is used as the most direct metric, defined as the ratio of correctly classified attachment samples to the total size of the dataset. Furthermore, a confusion matrix is ​​used to calculate TP, TN, FP, and FN to gain a more detailed understanding of the model's classification performance. The confusion matrix is ​​a 2x2 matrix where rows represent the actual class and columns represent the predicted class. According to the definition of a confusion matrix:

[0060] TP (True Positive): The number of samples correctly identified as positive.

[0061] TN (True Negative): The number of samples correctly identified as negative.

[0062] FP (False Positive): The number of samples that were incorrectly identified as positive.

[0063] FN (False Negative): The number of samples that were incorrectly identified as negative.

[0064] Accuracy formula

[0065]

[0066] By calculating these metrics, the model's performance can be evaluated more comprehensively, thereby determining the accuracy of the extracted information on ship hull attachments. When the accuracy reaches 0.9 or higher, the model can be considered to have completed training.

[0067] (III) Feature Extraction-Based Attachment Detection Method

[0068] The feature extraction-based method applies traditional digital image processing techniques to preprocess underwater images of the hull acquired by the robot and extract features of attachments such as barnacles. The processed results are then analyzed to determine the location of the barnacles. Digital image processing detection techniques, including image segmentation, utilize thresholding, edge detection, and mathematical morphology methods. Image enhancement is applied to the underwater images acquired by the robot, combined with thresholding segmentation, to further improve the recognition ability of barnacles, etc. This involves several steps: image grayscale conversion, linear transformation for image enhancement, and morphological processing. Image grayscale conversion reduces the memory footprint of the acquired underwater images, improving computational efficiency; image enhancement improves image clarity and contrast; and morphological processing utilizes the Otsu algorithm to segment the image, filter noise, and extract image information.

[0069] Image conversion from robot-acquired images: Colored images of hull attachments are acquired by an underwater hull cleaning robot. These images contain various information, and a weighted average method is used to convert the hull attachment images to grayscale.

[0070] Assuming the acquired crack image is Image(i,j), then the grayscale value VGray(ij) of the image is:

[0071]

[0072] In the formula: R(ij), G(i,j), and B(i,j) are the values ​​of the color image point (ij), where R, G, and BR represent red, G represents green, and B represents blue.

[0073] Image enhancement is performed on the images acquired by the robot to improve image quality. Linear grayscale transformation is used for image enhancement. Morphological processing is then applied, using the Otsu algorithm to calculate the maximum inter-class variance, selecting a threshold, segmenting the acquired barnacle images, extracting barnacle information, and using morphological principles to filter out noise and highlight the information in the barnacle images.

[0074] Both feature extraction and feature selection aim to identify the most effective features (invariance to similar samples, discriminative power against different samples, and robustness to noise) from the original features. The main feature extraction and discrimination methods are:

[0075] The shell is conical with a serrated rim and a smooth surface, lacking longitudinal ribs; the aperture is roughly pentagonal. These main features are extracted from the image after detection. Feature description or feature vectorization is then performed.

[0076] Feature extraction: The original features are transformed into a set of features with obvious physical meaning (Gabor, geometric features [corners, invariants], texture [LBP HOG]) or statistical meaning or kernel. After feature extraction, they are imported into the deep learning-based discriminative model. The extraction method is feasible when the accuracy is greater than 95%.

[0077] The workflow of this invention is as follows: First, images of barnacles awaiting removal are manually collected as samples. These images are then analyzed by a computer to obtain sample data (sample images). The following steps are then performed: First, the hull area to be cleaned is selected. With the aid of lighting equipment, an underwater camera captures images of this area. The captured image information is then transmitted to a visual computer for image enhancement (using convolutional neural networks, which can rely on modern AI technology) to improve image clarity and better observe the marine life to be cleaned. The enhanced results are then compared with the sample data (images) to determine the type and location of the organisms to be cleaned. This information is transmitted to a control computer for decision analysis (whether cleaning is necessary; if not, return to step one and select the next area for operation). The machine is then controlled to clean the area being tested. Finally, the above process is repeated.

[0078] Operation mode: During operation, the underwater components to be cleaned are imported into a model (i.e., the hull model is imported); a map model is constructed using a grid method, while a camera collects real-time information; an algorithm is used to plan the path; the control system controls the robot to run along the planned path, such as... Figure 5 As shown.

[0079] The robot collects attitude information through sensors and transmits it to the main controller. The host computer sends path information to the main controller, which controls the rotation speed of the propeller 2.8 by controlling the PWM output, thereby realizing the robot's functions of balancing, floating, moving forward and backward, and turning left and right.

[0080] Based on its current location, the camera controls the robotic arm 1.2 to move to the hull and activates the electromagnetic adsorption device 1.1 to adhere to the hull. The camera analyzes the barnacles' location to control the direction of the cavitation jet nozzles, thereby achieving the purpose of cleaning the barnacles.

[0081] During modeling, the ship's hull information is imported, and the work area is transformed into interconnected but non-overlapping grid cells according to a certain division method. By transforming the complex ship bottom environment information into discrete grid information, the robot's movement path is simplified.

[0082] In summary: 1. The underwater hull cleaning robot of this invention possesses high flexibility, exhibiting significant advantages in cleaning complex structures such as propellers. 2. The cavitation jet nozzle separated from the pump body in this invention reduces the robot's structural load, lowers energy consumption, and improves its flexibility, effectively ensuring its agile operation in water. 3. This invention employs electromagnetic adsorption technology, which, compared to other adsorption methods, is suitable for uneven hull surfaces, better stabilizing the underwater cleaning robot and ensuring smooth cleaning operations. 4. This invention's underwater hull cleaning robot can perform cleaning while the ship is moored, without absolute requirements on the ship's environment, expanding its applicability. Even during long voyages, it can inspect important components (such as propellers) for adhesion to barnacles and remove them promptly. 5. The underwater hull cleaning robot of this invention has higher practicality. Compared with commonly used wall-climbing robots, the underwater hull cleaning robot of this invention has higher cleaning safety and efficiency, can target key areas for cleaning, takes less time, and can skip areas that do not need cleaning. 6. The underwater hull cleaning robot of this invention can perform detection and cleaning work autonomously, requiring only simple manual guidance, and has a high degree of intelligence.

[0083] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0084] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. An underwater hull cleaning robot capable of autonomous detection, characterized in that: It includes a frame, and at least two robotic arms, a propeller, a cleaning module, a camera, and a control module mounted on the frame. The ends of the robotic arms are equipped with electromagnetic adsorption devices. The control and drive module is connected to the cleaning module, the camera, the robotic arms, the propellers, and the electromagnetic adsorption devices. The cleaning module includes cavitation jet nozzles. The method for cleaning hull attachments using the autonomous underwater hull cleaning robot is as follows: the underwater hull cleaning robot is moved to one side of the underwater hull area to be cleaned, and images of the underwater hull attachments in the area to be cleaned are captured by a camera. The trained neural network model is used to preprocess the underwater hull images collected by the robot and extract the features of the underwater hull attachments. After analysis and processing, the location of the attachments is obtained. After the location is determined, the robot adjusts its posture by adsorption robotic arm so that the cavitation jet nozzle is aimed at the edge root of the attachment. The cavitation jet nozzles of the cleaning module clean the underwater hull from the edge and root. The specific process of preprocessing underwater hull images collected by the robot is as follows: the underwater images collected by the robot are converted to grayscale and enhanced. The images are segmented using digital image processing and detection technology, using threshold-based methods, edge detection segmentation methods, and morphological processing methods to segment the images, filter out noise, and extract image information. By capturing subtle features of underwater barnacle attachment images collected by the robot, the images are first enhanced with specialized edge enhancement for barnacles, and then processed and analyzed for attachments of different sizes and shapes. By adding residual connections to the barnacle model on the hull, the ability to extract barnacle features is enhanced. The underwater hull images collected by the robot are preprocessed and features of attachments such as barnacles are extracted. The processed results are then analyzed to determine the location of the barnacles. Digital image processing detection technology image segmentation, including threshold-based, edge detection segmentation methods, and mathematical morphology methods; Image enhancement is performed on underwater images collected by the robot, combined with threshold segmentation, to further improve the ability to identify barnacles and other similar organisms. The process involves several steps, including image grayscale conversion, linear transformation for image enhancement, and morphological processing. Image grayscale conversion reduces the memory footprint of the collected underwater images, improving computational efficiency. Image enhancement improves image clarity and contrast. Morphological processing utilizes the Otsu algorithm to segment the image, filter out noise, and extract image information. The images collected by the robot are converted to grayscale: Color images of ship hull attachments are collected by the underwater ship hull cleaning robot. The images contain various information. The weighted average method is used to process the ship hull attachment images into grayscale. Image enhancement is performed on the images acquired by the robot to improve image quality. Linear grayscale transformation is used for image enhancement. Morphological processing is performed by using the Otsu algorithm to calculate the maximum inter-class variance, selecting a threshold, segmenting the acquired barnacle images, extracting barnacle information, and using morphological principles to filter out noise and highlight the information of the barnacle images. Feature extraction and feature selection both aim to find the most effective features from the original features. The main features extracted and discriminated are: conical shell; serrated shell aperture; smooth surface without longitudinal ribs; and a slightly pentagonal shell aperture. After detecting the main features mentioned above, extract them from the image and perform feature description or feature vectorization. Feature extraction: The original features are converted into a set of features with obvious physical or statistical significance or kernel significance. After feature extraction, they are imported into the discriminant model after deep learning.

2. The autonomous underwater hull cleaning robot as described in claim 1, characterized in that: The frame includes two side plates and a bottom plate. The two side plates are arranged side by side, and the two ends of the bottom plate are connected to the inner sides of the two side plates respectively. A support rod is also connected between the two side plates.

3. The autonomous underwater hull cleaning robot as described in claim 2, characterized in that: A cylindrical compartment is located on the bottom plate, and the control module is located inside the cylindrical compartment.

4. The underwater hull cleaning robot capable of autonomous detection as described in claim 1, characterized in that: There are two robotic arms, one on each side of the frame.

5. The autonomous underwater hull cleaning robot as described in claim 1, characterized in that: The number of propellers is 3 to 5.

6. The autonomous underwater hull cleaning robot as described in claim 1, characterized in that: The cleaning module also includes a vacuum pump and water pipes. The cavitation jet nozzle is mounted on the frame, and the vacuum pump is connected to the cavitation jet nozzle via the water pipes.

7. The autonomous underwater hull cleaning robot as described in claim 1, characterized in that: The control module includes a main controller and a slave controller. The main controller is set on the water surface and connected to the slave controller via a cable. The slave controller is set on the frame. The main controller is the host computer and the slave controller is the slave computer. The slave controller is connected to the robotic arm, propeller thruster, cleaning module and camera.

8. The autonomous underwater hull cleaning robot as described in claim 1, characterized in that: The rack is also equipped with lighting equipment and a light vision computer, which is connected to a camera.

Images