Sea cucumber suction-catching robot and suction-catching device

Abstract

The invention relates to a sea cucumber suction catching robot and a sea cucumber suction catching device, and solves the technical problems of how to improve the sea cucumber catching efficiency and reduce the damage to sea cucumbers. The device comprises an underwater vehicle and a suction catching device, the suction catching device is installed on the underwater vehicle, and the suction catching device comprises a first Y-axis direction linear module, a second Y-axis direction linear module, an X-axis direction linear module, a rear hose, a front end pipeline, a collecting cabin, a left underwater propeller, a right underwater propeller, a left filter screen, a right filter screen and a depth camera. The hanging type suction catching device sucks and collects the sea cucumbers, actions are soft, the structure is compact, and the sea cucumber suction catching device is suitable for completing catching under various terrains and postures.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sea cucumber fishing, in particular to a sea cucumber suction fishing robot and a suction fishing device. BACKGROUND

[0002] The ocean accounts for 70.8% of the total area of the earth, and contains rich mineral resources and marine biological resources, which are important wealth for the sustainable development of human society.

[0003] Sea cucumber is a kind of high-nutrition marine product, and sea cucumber feeds on seaweed and plankton in the offshore water and is usually artificially bred. At present, sea cucumber fishing is mainly carried out manually by using simple tools through artificial diving, which has the problems of low efficiency, high cost and high risk. The existing technology also has a fishing method of using a mechanical hand to grab sea cucumber, but the mechanical hand is easy to damage sea cucumber and has low operation efficiency. Referring to the invention patent application with the application publication number CN118489638A and the invention application with the application publication number CN112544577A, in addition, there is also a method of using an underwater fishing robot with a walking mechanism, but such a robot is easy to damage seaweed and the growing environment of sea cucumber during walking in the sea cucumber growing area on the seabed, and is easy to crush and kill part of the sea cucumber; for the scene with rocks and silt on the seabed, the walking mechanism is difficult to reliably and stably operate, thereby affecting the normal operation of the robot.

[0004] Therefore, how to improve the efficiency of sea cucumber fishing and reduce the damage to sea cucumber is a technical problem to be solved by those skilled in the art. SUMMARY

[0005] The present application is to solve the technical problem of how to improve the efficiency of sea cucumber fishing and reduce the damage to sea cucumber, and provides a sea cucumber suction fishing robot and a suction fishing device.

[0006] The present application provides a sea cucumber suction fishing robot, which comprises an underwater vehicle and a suction fishing device, the underwater vehicle comprising a frame, a propeller connected to the frame, and a camera connected to the frame.

[0007] The trapping device comprises a first Y-axis direction linear module, a second Y-axis direction linear module, an X-axis direction linear module, a rear hose, a front end pipeline, a collection cabin, a left underwater thruster, a right underwater thruster, a left filter screen, a right filter screen and a depth camera, the first Y-axis direction linear module and the second Y-axis direction linear module are arranged side by side, the first Y-axis direction linear module is provided with a first sliding block, the second Y-axis direction linear module is provided with a second sliding block, one end of the X-axis direction linear module is connected with the first sliding block of the first Y-axis direction linear module, the other end of the X-axis direction linear module is connected with the second sliding block of the second Y-axis direction linear module, the rear end of the rear hose is connected with the front end of the collection cabin, the front end pipeline is connected with the front end of the rear hose, and the depth camera is connected with the front end pipeline; the collection cabin is provided with a detachable rear cover, the left underwater thruster is connected with the left side plate of the collection cabin, the right underwater thruster is connected with the right side plate of the collection cabin, the left underwater thruster and the right underwater thruster are oppositely arranged, the left filter screen is connected with the inlet end of the left underwater thruster, and the right filter screen is connected with the inlet end of the right underwater thruster.

[0008] The first Y-axis direction linear module and the second Y-axis direction linear module are respectively connected at the bottom of the frame of the underwater vehicle, and the top of the collection cabin is connected with the bottom of the frame of the underwater vehicle.

[0009] Preferably, the front end pipeline is a front end hose or a front end rigid pipe.

[0010] Preferably, the depth camera collects sea cucumber images and transmits the images to an upper computer, the upper computer identifies target sea cucumbers through an improved YOLOv8 target detection model, and the improved YOLOv8 target detection model is provided with a full attention feature enhancement module SCC based on a multi-branch hollow convolution structure.

[0011] Preferably, the motion of the sea cucumber trapping robot is controlled through a sliding mode control, and a longitudinal motion equation can be modeled as:

[0012]

[0013] Wherein, m is the mass of the robot, w is the vertical velocity; is the derivative of w with respect to time; mg is the gravity, B is the buoyancy, D(w) is the hydrodynamic resistance function, and Tz is the thrust provided by the longitudinal thruster; the target depth is set as zd, the current depth is measured as z(t), and the depth error is:

[0014] e(t)=z d -z(t)

[0015] The sliding mode surface is defined as:

[0016]

[0017] And a control law is designed:

[0018] T z (t)=T eq -K·sgn(s)

[0019] where T eq is the equivalent control and sgn(s) is a function.

[0020] The application also provides a suction device, comprising a first Y-axis linear module, a second Y-axis linear module, an X-axis linear module, a rear hose, a front pipeline, a collection cabin, a left underwater propeller and a right underwater propeller, the first Y-axis linear module and the second Y-axis linear module are arranged side by side, the first Y-axis linear module is provided with a first sliding block, the second Y-axis linear module is provided with a second sliding block, one end of the X-axis linear module is connected with the first sliding block of the first Y-axis linear module, the other end of the X-axis linear module is connected with the second sliding block of the second Y-axis linear module, the rear end of the rear hose is connected with the front end of the collection cabin, and the front pipeline is connected with the front end of the rear hose; the collection cabin is provided with a detachable rear cover, the left underwater propeller is connected with a left side plate of the collection cabin, the right underwater propeller is connected with a right side plate of the collection cabin, and the left underwater propeller and the right underwater propeller are oppositely arranged.

[0021] Preferably, the suction device further comprises a left filter screen and a right filter screen, the left filter screen is connected with an inlet end of the left underwater propeller, and the right filter screen is connected with an inlet end of the right underwater propeller.

[0022] Preferably, the front pipeline is a front hose or a front rigid pipe.

[0023] Preferably, the left underwater propeller is located at a middle upper portion of the left side plate of the collection cabin, and the right underwater propeller is located at a middle upper portion of the right side plate of the collection cabin.

[0024] The application also provides a suction device, comprising a rear hose, a front pipeline, a collection cabin, a left underwater propeller and a right underwater propeller, the rear end of the rear hose is connected with the front end of the collection cabin, and the front pipeline is connected with the front end of the rear hose; the collection cabin is provided with a detachable rear cover, the left underwater propeller is connected with a left side plate of the collection cabin, the right underwater propeller is connected with a right side plate of the collection cabin, and the left underwater propeller and the right underwater propeller are oppositely arranged.

[0025] Preferably, the left underwater propeller is located at a middle upper portion of the left side plate of the collection cabin, and the right underwater propeller is located at a middle upper portion of the right side plate of the collection cabin.

[0026] The beneficial effects of the present application are that the innovative trapping method improves the trapping effect, and is flexible, convenient and efficient to operate.

[0027] The image recognition part is based on the RKNN accelerated deployment of a lightweight target detection model, and combines a multi-core parallel thread pool design to realize high-frame-rate target recognition, target center extraction and positioning.

[0028] The motion of the robot is controlled by an improved sliding mode control, which realizes better robustness, improves stability and real-time performance, ensures high-precision attitude stability, and is suitable for fine work or stable observation.

[0029] Further features and aspects of the present application will be described in the following detailed description of specific embodiments, with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is an isometric view of a sea cucumber trapping robot;

[0031] Figure 2 is a front view of the sea cucumber trapping robot shown in Figure 1

[0032] Figure 3 is a left view of the sea cucumber trapping robot shown in Figure 1

[0033] Figure 4 is a right view of the sea cucumber trapping robot shown in Figure 1

[0034] Figure 5 is a top view of the sea cucumber trapping robot shown in Figure 1

[0035] Figure 6 is a bottom view of the sea cucumber trapping robot shown in Figure 1

[0036] Figure 7 is a rear view of the sea cucumber trapping robot shown in Figure 1

[0037] Figure 8 is an isometric view of another view of the sea cucumber trapping robot;

[0038] Figure 9 is an isometric view of another view of the sea cucumber trapping robot;

[0039] Figure 10 ​​​​​​is another perspective view of the sea cucumber suction robot;

[0040] Figure 11 is a perspective view of the suction device;

[0041] Figure 12 is Figure 11 is a front view of the suction device shown in

[0042] Figure 13 is Figure 11 is a rear view of the suction device shown in

[0043] Figure 14 is Figure 11 is a left view of the suction device shown in

[0044] Figure 15 is Figure 11 is a right view of the suction device shown in

[0045] Figure 16 is Figure 11 is a top view of the suction device shown in

[0046] Figure 17 is Figure 11 is a bottom view of the suction device shown in

[0047] Figure 18 is another perspective view of the suction device;

[0048] Figure 19 is a filter screen installation structure diagram in the suction device;

[0049] Figure 20 is a state diagram after the front hose is extended forward;

[0050] Figure 21 is an architecture diagram of the improved YOLOv8 target detection model;

[0051] Figure 22 is a calculation process diagram of the improved YOLOv8 target detection model.

[0052] Explanation of symbols in the figure:

[0053] 100. underwater vehicle, 101. thruster, 102. bottom of frame, 103. front camera, 104. right camera, 105. left camera; 200. suction device, 201. first Y-axis linear module, 202. second Y-axis linear module, 203. X-axis linear module, 204. rear hose, 205. front hose, 206. collection cabin, 206-1. rear cover, 207. left underwater thruster, 208. right underwater thruster, 209. left filter screen, 210. right filter screen, 211. depth camera. Detailed Implementation

[0054] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] The specific embodiments described below are merely preferred embodiments of this application, and the scope of protection of this application is not limited thereto. Those skilled in the art can make modifications or variations based on the principles, concepts, and spirit of this application, and the resulting technical solutions should all be covered within the scope of protection of this application.

[0056] like Figures 1-19 As shown, the sea cucumber suction robot includes an underwater vehicle 100 and a suction device 200. The underwater vehicle 100 includes a frame and thrusters. Six thrusters 101 are mounted on the frame, providing six degrees of freedom for maneuvering and attitude adjustment. It can move underwater in six degrees of freedom: forward / backward, ascend / descend, lateral, yaw, pitch, and roll. Forward / backward movement is moving forward or backward, ascend / descend is floating up and down, and lateral movement is moving left and right laterally, making the robot flexible, stable, and adaptable to complex underwater environments. The underwater vehicle 100 includes a front camera 103, a right camera 104, and a left camera 105. The front camera 103 is mounted at the front end of the bottom 102 of the frame, the right camera 104 is mounted on the right side of the frame, and the left camera 105 is mounted on the left side of the frame. The front camera 103, right camera 104, and left camera 105 serve as underwater environmental visual perception cameras, providing the robot with a global field of view. The frame is equipped with sensors such as a depth gauge, IMU, barometer, and optical flow module, which are electrically connected to the Orange Pi. The lower-level system uses the Orange Pi as the main control core, forming a collaborative control structure with an STM32 microcontroller. The Orange Pi and STM32 microcontroller communicate via serial communication for master-slave interaction. The Orange Pi is responsible for image recognition and task decision-making, while the STM32 microcontroller handles attitude control, the six thrusters, and the suction device's movements. The depth gauge, IMU, barometer, and optical flow module enable real-time sensing of data such as depth, attitude, and displacement.

[0057] The suction device 200 includes a first Y-axis linear module 201, a second Y-axis linear module 202, an X-axis linear module 203, a rear hose 204, a front hose 205, a collection chamber 206, a left underwater thruster 207, a right underwater thruster 208, a left filter 209, a right filter 210, and a depth camera 211. The first Y-axis linear module 201 and the second Y-axis linear module 202 are arranged side by side. The first Y-axis linear module 201 is provided with a first slider, and the second Y-axis linear module 202 is provided with a second slider. One end of the X-axis linear module 203 is connected to the first slider of the first Y-axis linear module 201. The other end of 03 is connected to the second slider of the second Y-axis linear module 202. The rear end of the rear hose 204 is connected to the front end of the collection chamber 206. The front end hose 205 is connected to the front end of the rear hose 204. The depth camera 211 is connected to the front end hose 205. The collection chamber 206 is provided with a detachable rear cover 206-1. The left underwater thruster 207 is installed on the left side plate of the collection chamber 206. The right underwater thruster 208 is installed on the right side plate of the collection chamber 206. The left underwater thruster 207 and the right underwater thruster 208 are arranged opposite to each other. The left filter 209 is connected to the inlet end of the left underwater thruster 207. The right filter 210 is connected to the inlet end of the right underwater thruster 208. When the first Y-axis linear module 201 and the second Y-axis linear module 202 operate synchronously, they can drive the entire X-axis linear module 203 to move along the Y-axis, thereby driving the front-end hose 205 to move along the Y-axis. When the X-axis linear module 203 operates, it can drive the front-end hose 205 to move along the X-axis. With the X-axis linear module 203 and the two Y-axis linear modules working in coordination, the position of the front-end hose 205 in the XY plane is adjusted. (Reference) Figure 20 The state shown is in Figure 1 Based on the state shown, the first Y-axis linear module 201 and the second Y-axis linear module 202 operate to extend the front hose 205 forward a certain distance, resulting in the following state diagram. When the inlet of the front hose 205 is directly facing the sea cucumber, the left underwater thruster 207 and the right underwater thruster 208 are activated simultaneously. The rotation speed of the left underwater thruster 207 is equal to that of the right underwater thruster 208. The left underwater thruster 207 pumps water out of the collection chamber 206, and the right underwater thruster 208 pumps water out of the collection chamber 206 in opposite directions. This creates suction in the rear hose 204, which is located in the middle position. Water from the external environment continuously enters from the front hose 205, passes through the rear hose 204, and flows into the collection chamber 206.

[0058] The first Y-axis linear module 201 and the second Y-axis linear module 202 are fixedly installed on the bottom 102 of the frame (which may be detachably connected), and the top of the collection chamber 206 is fixedly connected to the bottom 102 of the frame (which may be detachably connected). Figure 9 and Figure 10 Rear hose 204 is not shown.

[0059] The suction device 200 works in conjunction with the underwater vehicle 100. When the underwater vehicle 100 descends to the sea cucumber area on the seabed, it adjusts its position and hovers over the sea cucumber area. The suction device 200 then operates, with the first Y-axis linear module 201, the second Y-axis linear module 202, and the X-axis linear module 203 working in tandem to align the inlet of the front hose 205 with the target sea cucumber. Next, the left underwater thruster 207 and the right underwater thruster 208 work synchronously, generating suction in the pipeline formed by the rear hose 204 and the front hose 205. The sea cucumber is then sucked in through the inlet of the front hose 205 and transported to the collection chamber 206. Normally, the sea cucumber will concentrate in the lower part of the collection chamber 206. The left filter 209 and the right filter 210 prevent the sea cucumber from being sucked away by the left underwater thruster 207 and the right underwater thruster 208, and also prevent damage to the sea cucumber. The left underwater thruster 207 and the right underwater thruster 208 are preferably installed in the upper middle part of the side plate of the collection chamber 206. After the suction operation is completed, a certain number of sea cucumbers are collected in the collection chamber 206. After the machine is stopped, the rear cover 206-1 of the collection chamber 206 is opened and the sea cucumbers are taken out.

[0060] To accurately identify the target sea cucumber, a sea cucumber target recognition model is deployed on the host computer. A depth camera on the suction device 200 captures images of sea cucumbers in the seabed area, and the target sea cucumber is identified through the sea cucumber target recognition model. This makes it easier to align the inlet of the front-end flexible tube 205 with the target sea cucumber. To ensure even more accurate alignment of the inlet of the front-end flexible tube 205 with the target sea cucumber, visual servo control is employed. The position coordinates of the target sea cucumber are determined based on the acquired images. Then, based on the target sea cucumber's position coordinates, the first Y-axis linear module 201, the second Y-axis linear module 202, and the X-axis linear module 203 are linked, thereby precisely bringing the inlet of the front-end flexible tube 205 close to the target sea cucumber's position coordinates.

[0061] Sea cucumber target recognition models can employ target detection models, and more specifically, YOLOv8 target detection models. Further improvements can be made to the conventional YOLOv8 target detection model by designing a fully attention-based feature enhancement module (SCC) based on a multi-branch dilated convolution structure. The specific process is as follows:

[0062] Step (1): The input image is adapted to the model size by proportional scaling and edge padding (LetterBox);

[0063] Step (2): After obtaining the actual image, obtain the feature representation map of the intermediate layer from the backbone network;

[0064] Step (3): Input the feature map into a multi-branch dilated convolutional structure for computation. One branch with dilated convolution is a residual structure used to preserve key information of small targets. The other three branches perform cascaded standard convolution operations, with dilated convolutional layers added to the two middle branches to expand the receptive field and preserve more contextual information. The computation formula for this part is as follows:

[0065]

[0066] in, and These represent standard convolution operations with kernel sizes of 1×1, 1×3, 3×1, and 3×3, respectively. Cat(·) represents the dilated convolution operation; Cat(·) represents the feature map concatenation operation. The input feature map is F; W1, W2, and W3 represent the output feature maps of the first three branches after standard convolution and dilated convolution, respectively; and Y is the output feature map of FEM.

[0067] Step (4): Input the output feature map Y into the two parallel paths at the bottom and perform pooling respectively. Then, slice the feature map along the H and W dimensions respectively, and merge the two sets of slices. Finally, capture the complete attention A through the Affinity operation. The Affinity operation is calculated as follows:

[0068]

[0069] Where A ij This represents the correlation between the i-th and j-th channels at a specific spatial location. The results are then reshaped into two sets, and these two sets are summed to form long-range context information. Finally, this context information is multiplied by a scaling parameter and added element-wise to the input features to obtain the final output.

[0070] Step (5): Integrate features at different scales to achieve multi-scale target information integration.

[0071] Step (6): Predict the target bounding box by decoding the multi-scale feature map through distributed regression and calculate the class confidence.

[0072] Step (7): Use non-maximum suppression to remove overlapping prediction boxes and finally generate target category and location detection results, which can be efficiently inferred on the RK3588 multi-core NPU through the RKNN framework.

[0073] As can be seen, the full attention feature enhancement module SCC based on a multi-branch dilated convolutional structure increases the receptive field while capturing contextual information at different scales through multiple parallel convolutional paths, enabling it to capture multi-scale information from local to global perspectives. Furthermore, because it can adjust the dilation rate within the same convolutional layer to adapt to different receptive fields, it can extract features at different scales, effectively improving fine-grained feature learning and significantly enhancing the recognition of small targets. Moreover, by expanding the receptive field, the network can simultaneously capture target details and background information over a larger region. The full attention mechanism integrates horizontal and vertical contextual information, overcoming the limitation of conventional convolutional operations that only process neighboring pixels, effectively modeling global spatial relationships, and offering advantages for complex visual tasks.

[0074] The trained and improved YOLOv8 object detection model is deployed on the host computer.

[0075] The underwater vehicle can specifically be an unmanned remotely operated vehicle (ROV).

[0076] The control system does not use the traditional PID control method, but instead employs an improved sliding-mode control (SMC). For the depth loop of the control depth, the error e is defined. z =z des -z, sliding surface The control law uses an improved saturation function:

[0077]

[0078] In the formula, T0 is the approximate gravity and buoyancy compensation term, k is the gain, and φ is the boundary layer thickness, used to reduce the chattering caused by the classical sign function. For roll or pitch control, the error e is defined as... θ =θ des -θ, input the torque distribution matrix to obtain the PWM of each thruster. des This ensures that the rotation around the Z-axis is at a constant rate.

[0079] The robot can be manually controlled via remote control. Roll and pitch are controlled by angle; the joystick input directly maps to the desired roll and pitch angles, and sliding mode control (SMC) achieves high-precision attitude stabilization, allowing the ROV to maintain any tilt angle precisely, suitable for delicate operations or stable observation. Roll and pitch are controlled by angular velocity; the joystick input corresponds to the desired rotational angular velocity around the vertical axis (Yaw axis), and sliding mode control again achieves constant-speed rotation, facilitating smooth heading adjustments by the operator. For XY plane motion (forward / backward & left / right movement): acceleration control is used; the joystick input directly maps to acceleration in the X and Y directions, enabling smooth acceleration and deceleration and improving maneuverability. For depth control, the joystick input directly corresponds to the desired depth, achieving overshoot-free and fast-converging depth adjustment, ensuring stable hovering of the ROV even in complex water flow environments.

[0080] Maintaining a stable longitudinal altitude is crucial for ensuring observation accuracy and the safety of mechanical operations during underwater operations. Firstly, based on the vertical dynamics model of the ROV robot, factors such as gravity, buoyancy, hydrodynamic drag, and propulsion are comprehensively considered. Its longitudinal motion equations can be modeled as follows:

[0081]

[0082] Where m is the mass of the robot, and w is the vertical velocity; is the derivative of w with respect to time, the rate of change of vertical velocity with time, i.e., vertical acceleration; mg is gravity, B is buoyancy, D(w) is the hydrodynamic drag function, and Tz is the thrust provided by the longitudinal thruster. If the target depth is set to zd, and the current depth is measured in real time as z(t), then the depth error is:

[0083] e(t) = z d -z(t)

[0084] The controller needs to dynamically adjust the thruster thrust based on the error. The conventional method is PID control.

[0085]

[0086] Where Kp, Ki, and Kd are the proportional, integral, and derivative gains, respectively.

[0087] In complex sea conditions, a single PID controller is easily affected by external disturbances (such as water flow disturbances and density gradient changes). To improve robustness, sliding mode control (SMC) can be introduced. For example, a sliding surface can be defined as follows:

[0088]

[0089] And design the control law:

[0090] T z (t)=T eq -K·sgn(s)

[0091] Where T eq For equivalent control, sgn(s) is the sign function. This method ensures that the ROV maintains high-precision altitude-keeping cruise even under external disturbances and parameter uncertainties.

[0092] Improved sliding mode control achieves better robustness while enhancing stability and real-time performance. It ensures high-precision attitude stabilization, making it suitable for delicate operations or stable observations.

[0093] The host computer communicates with the Orange Pi via the UDP communication protocol, and the remote control communicates with the host computer.

[0094] The host computer is equipped with a graphical user interface (GUI) for human-computer interaction. The GUI integrates functions such as video display, status monitoring, and control command interaction, providing an intuitive and easy-to-operate host computer visualization platform.

[0095] One working process of the aforementioned sea cucumber suction robot is as follows: The operator controls the robot's movement on the seabed via a remote control, observing environmental video perceived by the front camera 103, right camera 104, and left camera 105 on a graphical user interface. Under sliding mode control, the robot's posture is adjusted for stable observation. Once a sea cucumber area is detected, the robot hovers in a specific posture according to the actual working conditions. Next, the depth camera 211 captures images of the sea cucumber and transmits them to a host computer. The host computer identifies the target sea cucumber using a sea cucumber target detection model. Then, through visual servoing, based on the target sea cucumber's position coordinates, the first Y-axis linear module 201, the second Y-axis linear module 202, and the X-axis linear module 203 are linked, allowing the inlet of the front flexible hose 205 to precisely approach the target sea cucumber. Finally, the left underwater thruster 207 and the right underwater thruster 208 operate, sucking the target sea cucumber into the collection chamber 206.

[0096] It should be noted that the specific structure of the left underwater thruster 207 and the right underwater thruster 208 can be either a structure in which the propeller is connected to the rotating shaft, or a structure without a rotating shaft.

[0097] It should be noted that a rigid pipe can be used instead of a flexible hose 205 at the front end. Using a rigid pipe is a specific implementation method for the front-end pipeline.

Claims

1. A sea cucumber suction robot, characterized in that, It includes an underwater vehicle and a suction device. The underwater vehicle includes a frame, a thruster connected to the frame, and a camera connected to the frame. The suction device includes a first Y-axis linear module, a second Y-axis linear module, an X-axis linear module, a rear hose, a front pipe, a collection chamber, a left underwater thruster, a right underwater thruster, a left filter, a right filter, and a depth camera. The first Y-axis linear module and the second Y-axis linear module are arranged side by side. The first Y-axis linear module has a first slider, and the second Y-axis linear module has a second slider. One end of the X-axis linear module is connected to the first slider of the first Y-axis linear module. The other end of the assembly is connected to the second slider of the second Y-axis linear module. The rear end of the rear hose is connected to the front end of the collection chamber. The front end pipe is connected to the front end of the rear hose. The depth camera is connected to the front end pipe. The collection chamber is equipped with a detachable rear cover. The left underwater thruster is connected to the left side plate of the collection chamber. The right underwater thruster is connected to the right side plate of the collection chamber. The left and right underwater thrusters are arranged opposite to each other. The left filter is connected to the inlet end of the left underwater thruster. The right filter is connected to the inlet end of the right underwater thruster. The first Y-axis linear module and the second Y-axis linear module are respectively connected to the bottom of the underwater vehicle's frame, and the top of the collection compartment is connected to the bottom of the underwater vehicle's frame.

2. The sea cucumber suction robot according to claim 1, characterized in that, The front-end pipe is a flexible front-end tube or a rigid front-end tube.

3. The sea cucumber suction robot according to claim 1, characterized in that, The depth camera acquires images of sea cucumbers and transmits them to the host computer. The host computer identifies the target sea cucumbers using the improved YOLOv8 target detection model, which incorporates a full attention feature enhancement module (SCC) based on a multi-branch dilated convolution structure.

4. The sea cucumber suction robot according to claim 3, characterized in that, The motion of the sea cucumber suction robot is achieved through sliding mode control, and the longitudinal motion equation can be modeled as follows: Where m is the mass of the robot, and w is the vertical velocity; Let be the derivative of w with respect to time; mg be gravity, B be buoyancy, D(w) be the hydrodynamic drag function, and Tz be the thrust provided by the longitudinal thruster; set the target depth as zd, and the current depth measured in real time as z(t), then the depth error is: e(t)=z d -z(t) Define the sliding surface: And design the control law: T z (t)=T eq -K·sgn(s) Where T eq For equivalent control, sgn(s) is a function.

5. A suction device, characterized in that, The system includes a first Y-axis linear module, a second Y-axis linear module, an X-axis linear module, a rear hose, a front pipe, a collection chamber, a left underwater thruster, and a right underwater thruster. The first and second Y-axis linear modules are arranged side by side. The first Y-axis linear module has a first slider, and the second Y-axis linear module has a second slider. One end of the X-axis linear module is connected to the first slider of the first Y-axis linear module, and the other end of the X-axis linear module is connected to the second slider of the second Y-axis linear module. The rear end of the rear hose is connected to the front end of the collection chamber, and the front pipe is connected to the front end of the rear hose. The collection chamber has a removable rear cover. The left underwater thruster is connected to the left side plate of the collection chamber, and the right underwater thruster is connected to the right side plate of the collection chamber. The left and right underwater thrusters are arranged opposite to each other.

6. The suction device according to claim 5, characterized in that, The suction device also includes a left filter and a right filter. The left filter is connected to the inlet end of the left underwater thruster, and the right filter is connected to the inlet end of the right underwater thruster.

7. The suction device according to claim 5, characterized in that, The front-end pipe is a flexible front-end tube or a rigid front-end tube.

8. The suction device according to claim 5, characterized in that, The left underwater thruster is located in the upper middle part of the left side plate of the collection compartment, and the right underwater thruster is located in the upper middle part of the right side plate of the collection compartment.

9. A suction device, characterized in that, It includes a rear hose, a front pipe, a collection chamber, a left underwater thruster, and a right underwater thruster. The rear end of the rear hose is connected to the front end of the collection chamber, and the front pipe is connected to the front end of the rear hose. The collection chamber is equipped with a removable rear cover. The left underwater thruster is connected to the left side plate of the collection chamber, and the right underwater thruster is connected to the right side plate of the collection chamber. The left and right underwater thrusters are arranged opposite to each other.

10. The suction device according to claim 9, characterized in that, The left underwater thruster is located in the upper middle part of the left side plate of the collection compartment, and the right underwater thruster is located in the upper middle part of the right side plate of the collection compartment.

Images