An underwater UUV laser cutting actuator and method of execution

Abstract

The application provides an underwater UUV laser cutting execution mechanism and an execution method, comprising a UUV platform, a mechanical arm module, an execution module, a compensation control module and a positioning unit; the UUV platform is used for providing underwater navigation power, energy supply and a pressure-resistant carrying space for accommodating multiple modules for collaborative operation; the mechanical arm module is installed on a support frame of the UUV platform and extends outwards through a waterproof hatch when receiving a work instruction; the execution module is installed at a terminal flange of the mechanical arm module, and the execution module comprises a positioning unit and an underwater laser cutting unit; the compensation control module is in communication connection with the UUV platform, the mechanical arm module and the execution module through a double-redundancy industrial field bus, and the UUV platform is positioned on the surface of a metal structure by using the positioning unit.

Description

Technical Field

[0001] This invention relates to the field of underwater laser cutting technology, and in particular to an underwater UUV laser cutting actuator and method. Background Technology

[0002] With the rapid development of marine resource development and underwater engineering construction, the demand for maintenance, dismantling, and emergency rescue operations of underwater metal structures is increasing. Traditional underwater cutting operations mainly rely on divers carrying handheld cutting tools or using tethered remotely operated vehicles (ROVs) equipped with mechanical cutting equipment. These methods face severe challenges in deep waters, complex sea conditions, or hazardous environments containing radioactive, flammable, or explosive materials. Unmanned underwater vehicles (UUVs), as autonomous and intelligent underwater operation platforms, possess advantages such as untethered operation, maneuverability, and the ability to carry various sensors and actuators, and are gradually becoming an important equipment direction for deep-sea engineering operations.

[0003] When UUVs are equipped with cutting equipment for underwater operations, they face numerous technical bottlenecks: Regarding positioning stability, the underwater environment is subject to continuous current disturbances, wave forces, and buoyancy changes, making it difficult for the UUV platform and robotic arm to maintain stable hold on the target metal surface, leading to deviations in the cutting starting point or interruptions in the cutting process. Regarding trajectory accuracy, the robotic arm's movement underwater is affected by fluid damping effects, generating additional mass forces and resistance, causing jitter and lag in the trajectory, making it difficult to guarantee the geometric accuracy of the cutting path. Regarding cutting technology, seawater has a strong absorption and scattering effect on laser energy, and the metal structure surface is often covered with marine organisms or corrosion layers, resulting in unstable cutting energy coupling efficiency and difficulty in controlling cutting quality. Furthermore, existing underwater cutting robots typically employ open-loop control or simple PID feedback control, lacking the ability to predict and actively compensate for complex underwater dynamic disturbances, making it difficult to adapt to the high-precision cutting requirements of the complex deep-sea environment. Summary of the Invention

[0004] The technical problem solved by this invention is the dynamic positioning stability and trajectory accuracy control problem in underwater laser cutting operations.

[0005] To address the aforementioned technical problems, this invention provides an underwater UUV laser cutting actuator, comprising a UUV platform, a robotic arm module, an execution module, a compensation control module, and a positioning unit. The UUV platform provides underwater propulsion, energy supply, and a pressure-resistant mounting space for multiple modules to work collaboratively. The robotic arm module is mounted on the support frame of the UUV platform and extends outward through a waterproof hatch upon receiving a work command. The execution module is mounted on the end flange of the robotic arm module and includes a positioning unit and an underwater laser cutting unit. The compensation control module is connected to the UUV platform, robotic arm module, and execution module via a dual-redundant industrial fieldbus. The UUV platform utilizes the positioning unit for positioning on the surface of a metal structure.

[0006] Furthermore, the execution module includes multiple finger-like structures, and at least three independently operating composite adsorption feet are arranged in an array on the inner side of the finger-like structures. Each composite adsorption foot integrates a flexible negative pressure suction cup and a magnetic array.

[0007] As a preferred embodiment of the underwater UUV laser cutting actuator of the present invention, wherein: the total normal holding force of the positioning unit on the target metal surface The mathematical expression is: ; in, This represents the number of working units in the magnetic array. For the first The effective magnetic induction intensity generated by each magnetic attraction unit at the contact surface For the first The effective adsorption area of ​​each magnetic unit. The permeability of free space, For the first The angle between the axis of each magnetic unit and the normal to the surface of the metal being attracted. The number of activated flexible negative pressure suction cups. The absolute hydrostatic pressure of the external environment at the current water depth. For the first The residual absolute pressure inside the cavity of a flexible negative pressure suction cup For the first The effective projected area after the flexible negative pressure suction cup is attached.

[0008] Furthermore, the underwater laser cutting unit includes a semiconductor-pumped solid-state laser generator, an underwater laser collimation and focusing system, and a coaxial auxiliary high-pressure air blowing and slag removal head; the semiconductor-pumped solid-state laser generator is used to output a green light band laser with a wavelength of 532nm to reduce seawater absorption attenuation; the coaxial auxiliary high-pressure air blowing and slag removal head is connected to a high-pressure gas cylinder, and the coaxial auxiliary high-pressure air blowing and slag removal head is used to continuously spray inert gas at the cutting focal point.

[0009] Furthermore, each rotary joint of the robotic arm module is coaxially integrated with a bidirectional torque sensor, a high-resolution photoelectric or magnetic absolute encoder, and a harmonic reducer; the rotary joint is covered with an oil-filled, pressure-resistant, compensated outer shell.

[0010] Furthermore, the compensation control module includes a high-frequency inertial measurement unit, a six-dimensional torque sensor, and an acousto-optic composite sensing network. The high-frequency inertial measurement unit includes a three-axis microelectromechanical gyroscope and a three-axis accelerometer. The six-dimensional torque sensor is installed between the end flange and the end effector of the robotic arm module. The acousto-optic composite sensing network includes a multi-beam forward-looking sonar, a three-dimensional structured light camera, and an underwater low-light camera. The acousto-optic composite sensing network is used to acquire multi-source heterogeneous topographic data of the target surface in waters with extremely low visibility or high turbidity. The compensation control module adopts a heterogeneous computing architecture based on a field-programmable gate array (FPGA) and a digital signal processor (DSP). The FPGA is used to perform parallel acquisition and hardware-level Kalman filtering noise reduction of the IMU high-frequency signal and multi-axis encoder signal at microsecond-level clock cycles. The DSP is used to execute inverse kinematics calculation, trajectory planning, and multi-source data fusion algorithms.

[0011] The present invention also provides an underwater UUV laser cutting execution method, comprising: Step S1: The UUV platform uses a multi-beam forward-looking sonar and a three-dimensional structured light camera to perceive the three-dimensional boundary of the target metal structure, solves the thrust distribution matrix of the thruster in the airborne DSP computing unit, navigates to the target surface within a set threshold range, and establishes the initial transformation matrix of the geodetic coordinate system, the UUV body coordinate system and the target working surface coordinate system. Step S2: The robotic arm module guides the execution module to approach the target surface, activates the magnetic array to provide normal preload based on the surface roughness, starts the negative pressure pump system, and dynamically adjusts the suction flow rate using the proportional-integral-derivative (PID) control algorithm until the anti-slip safety threshold is reached, thus completing the physical locking of the target area. Step S3: After physical locking is completed, the acousto-optic composite sensing network performs a fine scan of the cutting area, uses the Iterative Closest Point (ICP) algorithm to perform local point cloud registration, and generates a smooth NURBS laser cutting spatiotemporal trajectory equation. In step S4, the underwater laser cutting unit starts to vibrate and outputs a continuous laser beam according to the set trajectory. At the same time, the compensation control module uses the nonlinear disturbance observer (NDO) to estimate the transient ocean current disturbance torque in real time and calculate the inverse solution of the Jacobian matrix to perform reverse micro-displacement compensation on each joint of the high-frequency driven robotic arm module. Step S5: During the entire cutting operation, the high-frequency solution of the end focus tracking error is used to feed the tracking error back to the main control circuit. If the error exceeds the preset tolerance or a sudden loss of pressure occurs, the laser power supply is cut off and the attitude holding or emergency escape logic is executed.

[0012] Furthermore, in step S2, the suction process of the negative pressure pump system employs a closed-loop PID control algorithm to maintain absolute pressure, and the control voltage of the digital micro electromagnetic proportional valve... The mathematical expression is: ; in, This is the gain coefficient for the controller proportional to the input. The gain coefficient for integration. The gain coefficient is the derivative. For pressure tracking error, The target safe negative pressure value is set based on the current water depth and the calculated tangential friction force requirements. This refers to the residual absolute pressure that is fed back in real time inside the cavity.

[0013] Furthermore, in step S3, in order to eliminate the trajectory jitter of the robotic arm module in the underwater actuator due to the water resistance effect, the smooth NURBS laser cutting spatiotemporal trajectory equation is... The mathematical expression is: ; in, This is an array of control points for the cutting shape extracted from the point cloud model. These are the weighting factors for the corresponding control points. for B-spline basis functions These are the normalized time and arc length node parameters.

[0014] Furthermore, in step S4, the compensation control module adopts a feedforward-feedback composite dynamic control law based on a nonlinear disturbance observer (NDO), and the joint output torque of the robotic arm module... Issued according to the following dynamic model: ; in, These are the actual angles of each joint in the robotic arm module. The angular velocities of each joint in the robotic arm module are... The desired angular velocity is calculated by inverse solution of the NURBS trajectory equation. The desired angular acceleration is calculated using the inverse solution of the NURBS trajectory equation. For position tracking error, The inertia matrix, The Coriolis force matrix, This is the underwater gravity and buoyancy compensation term. It is a positive definite proportional matrix. The differential gain matrix is... Nonlinear disturbance torque of water flow collected by the observer; The underwater laser cutting unit obtains the local thickness of the metal structure based on the visual model scan. and the real-time displacement speed of the end effector of the robotic arm module Dynamically adjust the output power of the laser generator This results in an effective linear energy density acting on the cutting surface. Maintaining within a constant interval, its mathematical expression is: ; in, This represents the transmission efficiency coefficient of green laser light in a localized low-pressure cavitation channel. The diameter of the laser focal spot. For the target metal density, Specific heat capacity of the metal This refers to the temperature difference required for a metallic material to rise from ambient water temperature to its molten state.

[0015] The beneficial effects of the present invention are as follows: (1) The present invention achieves autonomous execution of underwater laser cutting operations through the coordinated cooperation of UUV platform, robotic arm module, execution module and compensation control module. The execution module adopts a combination of magnetic array and flexible negative pressure suction cup for holding, which can achieve stable and reliable physical locking on the surface of complex curved metal structure, solving the problem of unstable positioning in underwater environment. The compensation control module adopts FPGA and DSP heterogeneous computing architecture to realize microsecond-level signal acquisition and real-time calculation, providing a computing basis for high-precision trajectory tracking. (2) The present invention uses NURBS trajectory planning method to generate smooth and continuous cutting path, effectively eliminating the trajectory jitter caused by water resistance effect when the robotic arm moves underwater. The feedforward-feedback composite dynamic control law based on nonlinear interference observer (NDO) can estimate and actively compensate for ocean current disturbance torque in real time, significantly improving trajectory tracking accuracy. The adaptive linear energy density control logic can dynamically adjust the laser power according to the changes in metal thickness and cutting speed to ensure the consistency of cutting quality. (3) This invention employs a 532nm green laser and coaxial high-pressure air blowing slag removal technology, which helps reduce the absorption and attenuation of laser energy by seawater and creates a local dry cavitation environment in the cutting area, significantly improving the efficiency and quality of underwater laser cutting. It provides an efficient, precise, and reliable intelligent technical means for the maintenance, dismantling, and emergency rescue operations of deep-sea metal structures. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0017] The underwater UUV laser cutting actuator in this embodiment is deployed on an operational unmanned underwater vehicle (UUV) platform. The hardware environment includes: an industrial control computer, DSP digital signal processor board, FPGA field-programmable gate array board, laser drive power supply, negative pressure pump driver, and motor driver integrated inside the pressure tank; and a multi-beam forward-looking sonar, 3D structured light camera, underwater low-light camera, inertial measurement unit (IMU), robotic arm joint motors and encoders, a six-dimensional force sensor at the end effector, a semiconductor-pumped solid-state laser generator, and an optical focusing system outside the pressure tank. The software environment includes: main control software based on the VxWorks real-time operating system, a perception and planning algorithm module based on the ROS (Robot Operating System), and hardware signal processing logic embedded in the FPGA.

[0018] Example 1: This example provides an underwater UUV laser cutting actuator and method, including a UUV platform, a robotic arm module, an execution module, a compensation control module, and a positioning unit.

[0019] The UUV platform provides underwater propulsion, energy supply, and a pressure-resistant mounting space for multi-module collaborative operation. The UUV platform has an internal pressure-resistant mounting space for underwater propulsion and energy supply. It has automatically opening and closing waterproof doors on its sides or belly. Inside the doors is a high-rigidity support frame to support the robotic arm module. Furthermore, the UUV platform's front end integrates an acoustic-optical composite sensing network, including a multi-beam forward-looking sonar, a 3D structured light camera, and an underwater low-light camera, used to acquire multi-source heterogeneous morphological data of target metal structures in deep-water environments with extremely low visibility or high turbidity.

[0020] The robotic arm module's base is mounted on the UUV platform's support frame via shock-absorbing damping components, and extends outward through the waterproof hatch upon receiving a work command. The robotic arm module possesses at least six degrees of freedom. Normally, the robotic arm module is stored inside the UUV, extending outward through the waterproof hatch during operation. The shock-absorbing damping components effectively isolate the module from vibrations. Each rotary joint of the robotic arm module coaxially integrates a bidirectional torque sensor, a high-resolution absolute encoder, and a harmonic reducer. The rotary joints are externally encased in an oil-filled, pressure-resistant shell, utilizing internally filled insulating oil and external pressure compensation bladders to achieve dynamic balance between the internal cavity and the external deep-sea water pressure. The robotic arm module's foremost end flange is used for docking with the work module.

[0021] The execution module, mounted at the end flange, includes a clamping structure and an underwater laser cutting unit. The execution module comprises multiple finger-like structures, with at least three independently operating composite suction feet arrayed on the inner side of each finger structure. Each composite suction foot integrates a flexible negative pressure suction cup and a magnetic array. The magnetic array can be a controlled electromagnet array or a Heilbeck permanent magnet array; when using a controlled electromagnet array, the UUV platform has an internal drive control module that automatically adjusts the excitation current based on the magnetic permeability and thickness of the target metal material.

[0022] The UUV platform is equipped with a bidirectional plunger-type negative pressure pump system. This system connects independently to the flexible negative pressure suction cups of each composite adsorption foot via armored pressure-resistant hoses, and each hose branch is equipped with a digital miniature electromagnetic proportional valve. The skirts of the flexible negative pressure suction cups are made of seawater-resistant nitrile rubber or fluorosilicone rubber, with an embedded metal support frame inside. The underwater laser cutting unit includes a semiconductor-pumped solid-state laser generator with an output wavelength of 532nm green light and is equipped with a coaxial auxiliary high-pressure air blowing head for slag removal.

[0023] Total normal holding force of the positioning unit on the target metal surface The mathematical expression is: ; in, This represents the number of working units in the magnetic array. For the first The effective magnetic induction intensity generated by each magnetic attraction unit at the contact surface For the first The effective adsorption area of ​​each magnetic unit. The permeability of free space, For the first The angle between the axis of each magnetic unit and the normal to the surface of the metal being attracted. The number of activated flexible negative pressure suction cups. The absolute hydrostatic pressure of the external environment at the current water depth. For the first The residual absolute pressure inside the cavity of a flexible negative pressure suction cup For the first The effective projected area after the flexible negative pressure suction cup is attached.

[0024] The underwater laser cutting unit includes a semiconductor-pumped solid-state laser generator, an underwater laser collimation and focusing system, and a coaxial auxiliary high-pressure air blowing head for slag removal. The semiconductor-pumped solid-state laser generator outputs a 532nm green light wavelength laser to reduce absorption attenuation in seawater. The coaxial auxiliary high-pressure air blowing head is connected to a high-pressure gas cylinder and continuously injects inert gas at the cutting focal point. This displaces the liquid water around the cutting focal point, creating a localized low-pressure cavitation channel, ensuring that the high-energy laser beam directly irradiates the target metal surface in a waterless environment.

[0025] The compensation control module includes a high-frequency inertial measurement unit (IMU), a six-dimensional torque sensor, and an acousto-optic composite sensing network. The IMU comprises a three-axis microelectromechanical gyroscope and a three-axis accelerometer, while the six-dimensional torque sensor is mounted between the end flange of the robotic arm module and the end effector. The compensation control module employs a heterogeneous computing architecture based on a field-programmable gate array (FPGA) and a digital signal processor (DSP). The FPGA is used for parallel acquisition of IMU high-frequency signals and multi-axis encoder signals at microsecond clock cycles, along with hardware-level Kalman filtering for noise reduction. The DSP is used to execute inverse kinematics calculations, trajectory planning, and multi-source data fusion algorithms.

[0026] The compensation control module is connected to the UUV platform, robotic arm module, and execution module via a dual-redundant industrial fieldbus. The UUV platform uses a positioning unit to position itself on the metal structure surface. Employing a dual mechanism of negative pressure adsorption and magnetic attraction, the UUV achieves stable adhesion and precise positioning across all attitudes on metal structures with complex flow field interference and attached marine organisms.

[0027] Example 2, refer to Figure 1 This embodiment, based on the previous embodiment, provides an underwater UUV laser cutting execution method, including: Step S1: The UUV platform uses a multi-beam forward-looking sonar and a three-dimensional structured light camera to perceive the three-dimensional boundary of the target metal structure. The thrust distribution matrix of the thruster is solved in the airborne DSP computing unit. The platform navigates to the target surface within a set threshold range and establishes the initial transformation matrix between the geodetic coordinate system, the UUV body coordinate system and the target working surface coordinate system.

[0028] In this embodiment, the UUV platform utilizes an acousto-optic composite sensing network to scan the target submarine cable. The onboard DSP computing unit solves the thrust distribution matrix of the propeller, guiding the UUV to hover within a set threshold range of the submarine cable and establishing an initial transformation matrix. Subsequently, the waterproof hatch is opened, and a multi-joint robotic arm module is extended. Through multi-sensor fusion sensing, the limitations of single underwater sensors due to environmental constraints are overcome, achieving accurate reconstruction of the three-dimensional boundary of the target metal structure. The multi-source heterogeneous sensing data fusion mechanism is a technical process that performs spatiotemporal registration and feature-level fusion of target information acquired by different types of sensors to construct a complete three-dimensional model of the target structure. In this embodiment, the multi-source heterogeneous sensing data fusion mechanism receives long-range contour data from a multi-beam forward-looking sonar and close-range high-precision topographic data from a three-dimensional structured light camera. After time synchronization and coordinate unification, the data is transmitted to the DSP computing unit for fusion processing and works in conjunction with the inertial navigation system to achieve autonomous positioning and navigation of the UUV platform.

[0029] Furthermore, the multi-source heterogeneous sensing data fusion mechanism specifically includes sonar sensing data acquisition, optical sensing data acquisition, and data fusion processing. Sonar sensing data acquisition utilizes a multi-beam forward-looking sonar to emit sound waves and receive echoes, generating two-dimensional contours or three-dimensional point cloud data of the target structure, used to obtain the approximate location of the target in turbid waters or at long distances. Optical sensing data acquisition uses a three-dimensional structured light camera to project coded gratings and acquire reflected images, calculating high-precision three-dimensional coordinates of the target surface using triangulation principles, used to obtain detailed features of the target in clear waters at close range. Data fusion processing employs extended Kalman filtering (EKF) or graph optimization algorithms to register and fuse sonar data with optical data, generating a complete three-dimensional boundary model of the target metal structure.

[0030] The thrust distribution matrix solution is a control algorithm that calculates the thrust required by each thruster based on the desired motion state of the UUV platform, and is used to achieve precise maneuvering and hovering of the UUV platform.

[0031] In this invention, the thrust distribution matrix solver receives the desired position and attitude commands from the navigation system, calculates the required resultant force and resultant torque in combination with the hydrodynamic model of the UUV platform, outputs the thrust distribution results to each thruster driver, and works with the attitude controller to achieve stable hovering of the UUV platform.

[0032] Solving the thrust distribution matrix for the thrusters specifically includes calculating motion state errors, determining control force requirements, and optimizing thrust allocation. Motion state error calculation involves comparing the desired position and attitude with actual measurements to obtain position and attitude deviations. Control force requirement determination uses PID or sliding mode control laws to convert motion state errors into the resultant force and torque required by the UUV platform. Optimizing thrust allocation uses the thruster layout matrix and a pseudo-inverse method or quadratic programming algorithm to solve for the thrust commands of each thruster, minimizing thruster energy consumption or thrust saturation. The multi-coordinate system transformation matrix is ​​a technical method for constructing the mathematical mapping relationship between the geodetic coordinate system, the UUV body coordinate system, and the target working surface coordinate system, used to describe the relative pose of the UUV platform and the target structure.

[0033] In this embodiment, the multi-coordinate system transformation matrix receives target pose information from the fusion perception system, calculates the rotation matrix and translation vector between each coordinate system in the DSP computing unit, outputs the transformation matrix to the trajectory planning module, and works with the motion control module to achieve precise control of the robotic arm.

[0034] The establishment of the multi-coordinate system transformation matrix specifically includes defining the geodetic coordinate system, the UUV body coordinate system, the target working surface coordinate system, and calculating the transformation matrix. The geodetic coordinate system is an inertial reference system with a fixed point in the work area as its origin, used to describe the absolute position of the UUV platform. The UUV body coordinate system is a body reference system with the UUV platform's center of buoyancy as its origin, used to describe the UUV platform's attitude and the robotic arm's installation position. The target working surface coordinate system is a local reference system with a feature point on the target metal structure surface as its origin, used to describe the relative position of the cutting trajectory. The transformation matrix calculation is based on sensor measurement data, using the singular value decomposition (SVD) algorithm to solve for the optimal rigid body transformation matrix between the coordinate systems.

[0035] During the approach to the target, the UUV platform first uses multi-beam forward-looking sonar to perform a long-range scan, obtaining a rough outline of the target's metal structure and guiding the UUV to the vicinity of the target. When the distance is less than the effective working distance of the 3D structured light camera, the structured light camera is activated to perform a fine scan, acquiring high-precision point cloud data of the target surface. The DSP computing unit fuses the data from both types of sensors in real time to construct a 3D model of the target and calculates the optimal working pose of the UUV platform relative to the target, controlling the thrusters to achieve fixed-point hovering and ensuring that the robotic arm can reach the target cutting area.

[0036] In step S2, the robotic arm module guides the execution module close to the target surface. Based on the surface roughness, the magnetic array is activated to provide normal preload. The negative pressure pump system is started, and the suction flow rate is dynamically adjusted using a proportional-integral-derivative (PID) control algorithm until the anti-slip safety threshold is reached, completing the physical locking of the target area. Step S2 achieves stable fixation of the execution module on the target metal surface. Through the dual action of magnetic attraction and negative pressure, it ensures that the end effector remains relatively stationary with respect to the target surface during the cutting process, providing a stable working platform for high-precision cutting.

[0037] The physical locking mechanism is a technical means by which the execution module establishes a rigid connection on the target metal surface through a combination of magnetic adsorption and negative pressure adsorption, in order to resist cutting forces and external disturbances during the cutting process.

[0038] In this invention, the physical locking mechanism receives a locking command from the main control system. The robotic arm guides the execution module to approach the target surface, sequentially activating the magnetic array and the negative pressure pump system, feeding back the locking state to the main control system, and working in conjunction with the six-dimensional force sensor to monitor changes in the holding force.

[0039] The physical locking mechanism specifically includes contact approach control, magnetic pre-tensioning force application, and negative pressure adsorption force maintenance. Contact approach control involves the robotic arm adjusting its end effector posture based on the target surface normal vector, controlling the execution module to approach the target surface at a preset speed and contact force. Magnetic pre-tensioning force application involves selectively activating some magnetic units based on the target surface roughness assessment results, providing an initial normal pre-tensioning force to prevent the execution module from slipping. Negative pressure adsorption force maintenance involves activating the negative pressure pump system to establish a vacuum environment inside the flexible negative pressure suction cup, using the internal and external pressure difference to generate adsorption force until the total holding force reaches the anti-slip safety threshold.

[0040] The total normal holding force calculation model is a mathematical formula used to quantify the combined effect of magnetic attraction and negative pressure adsorption, and is used to precisely control the magnitude of the holding force.

[0041] In this embodiment, the total normal holding force calculation model receives magnetic suction unit parameters, water depth environment parameters, and negative pressure chamber pressure parameters, calculates the total holding force in real time in the DSP calculation unit, outputs the calculation results to the control loop to determine whether the safety threshold has been reached, and works with the safety monitoring module to ensure operational safety.

[0042] In this embodiment, the multi-joint robotic arm module guides the execution module close to the submarine cable, and the clamping structure closes and adheres to the cable surface. First, the magnetic array inside the composite adsorption foot is activated to provide normal pre-tightening force; then, the negative pressure pump system is started, and the suction flow rate is dynamically adjusted using a PID control algorithm. The PID closed-loop control of the negative pressure pump system receives the real-time pressure value inside the cavity from the pressure sensor, compares it with the target safe negative pressure value, calculates the control voltage and outputs it to the electromagnetic proportional valve to adjust the suction flow rate until the pressure error converges to the allowable range.

[0043] The suction process of the negative pressure pump system uses a closed-loop PID control algorithm to maintain absolute pressure, and the control voltage of the digital miniature electromagnetic proportional valve is... The mathematical expression is: in, This is the gain coefficient for the controller proportional to the input. The gain coefficient for integration. The gain coefficient is the derivative. For pressure tracking error, The target safe negative pressure value is set based on the current water depth and the calculated tangential friction force requirements. This involves real-time feedback of residual absolute pressure within the cavity. Ultimately, this pressure is brought to the anti-slip safety threshold, achieving absolute rigidity and locking of the submarine cable.

[0044] A six-dimensional force sensor mounted at the end of the robotic arm monitors the contact force in real time. Once stable contact between the execution module and the target surface is detected, the magnetic array is activated first. The main control system determines the surface flatness based on pre-scanned surface roughness data: if the surface is relatively flat, all magnetic units are activated; if there are local unevennesses, only those magnetic units that can make good contact are selectively activated. Subsequently, the negative pressure pump system is started, and the DSP calculation unit uses a PID algorithm to adjust the electromagnetic proportional valve in real time according to the target safe negative pressure value, so that a vacuum environment is quickly established inside the flexible negative pressure suction cup. When the calculated value of the total normal holding force exceeds the anti-slip safety threshold (e.g., more than 3 times the tangential cutting force), the physical locking is determined to be complete, and the system enters the ready-to-cut state.

[0045] In step S3, after physical locking is completed, the acousto-optic composite sensing network performs a fine scan of the cutting area, uses the Iterative Closest Point (ICP) algorithm for local point cloud registration, and generates a smooth NURBS laser cutting spatiotemporal trajectory equation. Step S3 achieves precise planning of the cutting path. Through high-precision scanning and smooth trajectory generation, it ensures that the laser focus can accurately track the target cutting line and eliminates jitter during the robotic arm's movement.

[0046] Local point cloud registration technology is an algorithm that aligns the point cloud of the cutting region acquired through real-time scanning with a pre-stored target model to determine the precise position of the cutting path in the current coordinate system. Local point cloud registration technology receives point cloud data of the cutting region from an acousto-optic composite sensing network, uses the ICP algorithm to calculate the optimal rigid body transformation between the point cloud and the model, outputs the registration result to the trajectory generation module, and collaborates with the cutting path planning module to determine the actual cutting trajectory.

[0047] Local point cloud registration techniques specifically include feature point extraction, corresponding point matching, transformation matrix solving, and iterative optimization. Feature point extraction involves extracting geometrically significant feature points from the point cloud data, such as corner points and edge points. Corresponding point matching involves finding the closest point in the target model to the feature point as the corresponding point. Transformation matrix solving involves calculating the rotation matrix and translation vector using the least squares method based on the corresponding point pairs. Iterative optimization involves repeatedly performing the matching and solving process until the registration error is less than a set threshold or the maximum number of iterations is reached.

[0048] The NURBS laser cutting spatiotemporal trajectory equation generation method utilizes the Non-Uniform Rational B-Spline (NURBS) approach to construct a smooth and continuous mathematical model of the cutting trajectory, which guides the movement of the robotic arm. The NURBS laser cutting spatiotemporal trajectory equation generation receives control points from the registered cutting path, calculates the trajectory equation coefficients in the DSP computing unit, outputs the trajectory equation to the motion control module, and coordinates with the interpolation algorithm to generate joint motion commands for the robotic arm. The NURBS laser cutting spatiotemporal trajectory equation generation specifically includes control point array extraction, weight factor setting, basis function calculation, and trajectory equation synthesis. The control point array PiPi, composed of cutting topographic feature points extracted from the point cloud model, determines the geometry of the trajectory.

[0049] In step S3, in order to eliminate the trajectory jitter of the robotic arm module in the underwater actuator caused by the water resistance effect, a smooth NURBS laser cutting spatiotemporal trajectory equation is used. The mathematical expression is: ; in, This is an array of control points for the cutting shape extracted from the point cloud model. These are the weighting factors for the corresponding control points. for B-spline basis functions These are the normalized time and arc length node parameters.

[0050] In practice, after physical locking is completed, the 3D structured light camera on the execution module performs a close-range fine scan of the cutting area to acquire high-density point cloud data. The DSP computing unit uses the ICP algorithm to register the scan data with the pre-planned cutting path, correcting path deviations caused by positioning errors. Subsequently, key points on the cutting path are extracted as control points to generate a NURBS trajectory equation. This trajectory equation not only describes the geometry of the cutting path but also includes time parameters, specifying the velocity of the laser focus at each point on the path, thus achieving unified planning of the spatiotemporal trajectory.

[0051] In step S4, the underwater laser cutting unit oscillates according to the set trajectory and outputs a continuous laser beam. Simultaneously, the compensation control module uses a nonlinear disturbance observer (NDO) to estimate the transient ocean current disturbance torque in real time and calculate the inverse Jacobian matrix solution. This high-frequency drive compensates for the reverse micro-displacement of each joint of the robotic arm module, isolating the influence of the body's elastic vibration on the end-effector focus. Step S4 is the core execution step of the cutting operation. Through the synergistic effect of laser output and dynamic compensation, high-precision, high-quality cutting is achieved in an underwater environment.

[0052] Nonlinear Disturbance Observer (NDO) feedforward compensation technology is a control method that uses an observer to estimate external disturbances in real time and actively cancels the disturbances through a feedforward channel to improve the system's anti-interference capability. The NDO receives data from sensors of the robotic arm's joint position, velocity, and torque, estimates the ocean current disturbance torque based on a dynamic model, and outputs the estimated value to the control law calculation module, which works in conjunction with feedback control to achieve high-precision trajectory tracking.

[0053] The NDO (Non-Displacement Optimization) feedforward compensation technique specifically includes disturbance estimation model construction, observer gain design, and feedforward compensation implementation. The disturbance estimation model construction is based on the underwater dynamics equations of the robotic arm, using the unknown disturbance torque as an extended state variable to design a nonlinear observer equation. The observer gain design determines the observer gain matrix through pole placement or the Linear Matrix Inequality (LMI) method, ensuring rapid convergence of the estimation error. Feedforward compensation implementation directly superimposes the disturbance torque estimated by the observer into the control law, enabling the robotic arm to generate a reverse compensation motion instantly upon being disturbed, rather than waiting for trajectory deviations to occur before feedback adjustment. The feedforward-feedback composite dynamic control law is a control algorithm that combines feedforward compensation and feedback correction to achieve high-precision trajectory tracking of the robotic arm.

[0054] The compensation control module employs a feedforward-feedback composite dynamic control law based on a nonlinear disturbance observer (NDO) to control the joint output torque of the robotic arm module. Compensation will be performed, and the following dynamic model will be used as the basis for the distribution: ; in, These are the actual angles of each joint in the robotic arm module. The angular velocities of each joint in the robotic arm module are... The desired angular velocity is calculated by inverse solution of the NURBS trajectory equation. The desired angular acceleration is calculated using the inverse solution of the NURBS trajectory equation. For position tracking error, The inertia matrix, The Coriolis force matrix, This is the underwater gravity and buoyancy compensation term. It is a positive definite proportional matrix. The differential gain matrix is... The nonlinear disturbance torque of the water flow collected by the observer is used; finally, the small displacement compensation of each rotating joint is performed by high-frequency driving to isolate the influence of the elastic vibration of the body on the end focus.

[0055] The underwater laser cutting unit incorporates adaptive linear energy density control logic, based on the local thickness of the metal structure obtained through visual model scanning. and the real-time displacement speed of the end effector of the robotic arm module Dynamically adjust the output power of the laser generator This results in an effective linear energy density acting on the cutting surface. Maintaining within a constant interval, its mathematical expression is: ; in, This represents the transmission efficiency coefficient of green laser light in a localized low-pressure cavitation channel. The diameter of the laser focal spot. For the target metal density, Specific heat capacity of the metal This refers to the temperature difference required for a metallic material to rise from ambient water temperature to its molten state.

[0056] After the cutting operation begins, a semiconductor-pumped solid-state laser generator outputs a 532nm green laser beam, which is focused onto the metal surface by a collimation and focusing system. A coaxial auxiliary high-pressure air-blowing slag removal head injects inert gas, creating localized cavitation in the cutting area to remove molten slag and reduce seawater absorption of the laser. Simultaneously, the NDO (Neural Doppler) observer monitors the robotic arm's status in real time. Once it detects a torque change caused by ocean current disturbance, it immediately calculates a compensation torque and drives the robotic arm joints to perform reverse micro-motions to maintain a stable laser focus position. When the vision sensor detects a change in metal thickness, the main control system automatically adjusts the laser power to ensure complete cut penetration.

[0057] Step S5: During the entire cutting operation, the high-frequency solution of the end focus tracking error is used to feed back the deviation status to the main control circuit. If the deviation exceeds the preset tolerance or a sudden loss of pressure occurs, the laser power supply is cut off and the attitude holding or emergency escape logic is executed.

[0058] Step S5 also includes a safety self-locking feedback process. The safety self-locking feedback includes physical detachment warning and optical path cutoff logic; if the high-frequency normal vibration variance detected by the six-dimensional torque sensor exceeds the material yield threshold, or the negative pressure monitoring circuit detects a pressure drop rate... ( (For leakage tolerance constant), the adhesion surface will be determined to be about to fail; at this time, the FPGA chip locks the laser generator power supply through hardware interrupt within 10 microseconds, and at the same time drives the thrust distribution network of the UUV platform to generate reverse escape thrust to ensure the safe recovery of deep-water equipment.

[0059] Step S5 implements safety monitoring and anomaly handling during the operation, ensuring the safety of the equipment and the workpiece. The full-process safety monitoring mechanism is a technical process that monitors and makes logical judgments on key status parameters during the cutting operation in real time, used to promptly detect and handle abnormal situations. The full-process safety monitoring mechanism receives real-time data from encoders, force sensors, and pressure sensors, performs high-frequency calculations and threshold comparisons in the FPGA, outputs the judgment results to the main control logic, and coordinates with safety relays to execute emergency shutdown operations.

[0060] The comprehensive safety monitoring mechanism includes focus tracking error monitoring, adsorption pressure monitoring, and anomaly handling logic. Focus tracking error monitoring calculates the deviation between the actual laser focus position and the planned trajectory using high-frequency calculations. If the deviation exceeds a preset tolerance (e.g., ±0.5mm), trajectory tracking is considered a failure. Adsorption pressure monitoring monitors the internal pressure of the negative pressure chamber in real time. If the pressure suddenly rises to near ambient pressure, a sudden pressure loss is considered (e.g., suction cup detachment or seal failure). Anomaly handling logic includes: immediately cutting off the laser power supply to prevent accidental injury; maintaining the current posture and awaiting manual intervention; or initiating the robotic arm retraction program for emergency disengagement.

[0061] The FPGA acquires data from each sensor at microsecond intervals and calculates the end-focus error in parallel. During normal operation, the error remains within the tolerance range, allowing for continuous operation. If a sudden strong ocean current causes the robotic arm to vibrate violently, and the focus error momentarily exceeds the limit, the FPGA immediately triggers an interrupt signal, cutting off the laser output and locking the robotic arm's current posture to prevent secondary damage. If the negative pressure sensor detects a sudden pressure increase, indicating adsorption failure, the system immediately initiates an emergency detachment procedure, controlling the robotic arm to retract and locking the waterproof hatch, and the UUV platform to retreat to a safe distance.

[0062] Example 3: The actuator and execution method of Example 1 are also applicable to the repair of complex curved surfaces such as ship propeller trimming. When the work object is a wide propeller blade with three-dimensional curvature, the clamping structure unfolds its multi-finger configuration and lays flat against the surface of the propeller blade. The composite adsorption foot adaptively adjusts the electromagnetic excitation current and negative pressure distribution according to the blade thickness and material permeability. After fixing, the underwater laser cutting unit performs high-precision trimming operations along the edge of the propeller blade under the anti-disturbance compensation of the nonlinear interference observer. The entire operation logic is consistent with Example 2, fully verifying the reliability of the actuator's all-attitude precise positioning and ultimate penetration processing.

Claims

1. An underwater UUV laser cutting actuator, characterized in that, It includes a UUV platform, a robotic arm module, an execution module, a compensation control module, and a positioning unit; The UUV platform is used to provide underwater navigation power, energy supply, and pressure-resistant mounting space for multiple modules to work together. The robotic arm module is mounted on the support frame of the UUV platform and extends outward through the waterproof hatch when it receives a work command. The execution module is installed at the end flange of the robotic arm module, and the execution module includes a positioning unit and an underwater laser cutting unit; The compensation control module is connected to the UUV platform, the robotic arm module and the execution module via a dual-redundant industrial fieldbus. The UUV platform uses the positioning unit to position itself on the surface of the metal structure.

2. The underwater UUV laser cutting actuator as described in claim 1, characterized in that, The execution module includes multiple finger-like structures, and at least three independently operating composite adsorption feet are arranged in an array on the inner side of the finger-like structures. Each composite adsorption foot integrates a flexible negative pressure suction cup and a magnetic array.

3. The underwater UUV laser cutting actuator as described in claim 2, characterized in that, The total normal holding force of the positioning unit on the target metal surface The mathematical expression is: ； in, This represents the number of working units in the magnetic array. For the first The effective magnetic induction intensity generated by each magnetic attraction unit at the contact surface For the first The effective adsorption area of ​​each magnetic unit. The permeability of free space, For the first The angle between the axis of each magnetic unit and the normal to the surface of the metal being attracted. The number of activated flexible negative pressure suction cups. The absolute hydrostatic pressure of the external environment at the current water depth. For the first The residual absolute pressure inside the cavity of a flexible negative pressure suction cup For the first The effective projected area after the flexible negative pressure suction cup is attached.

4. The underwater UUV laser cutting actuator as described in claim 3, characterized in that, The underwater laser cutting unit includes a semiconductor-pumped solid-state laser generator, an underwater laser collimation and focusing system, and a coaxial auxiliary high-pressure air blowing slag removal head; The semiconductor-pumped solid-state laser generator is used to output a green light band laser with a wavelength of 532nm to reduce absorption and attenuation in seawater. The coaxial auxiliary high-pressure air blowing slag discharge head is connected to a high-pressure gas cylinder, and the coaxial auxiliary high-pressure air blowing slag discharge head is used to continuously spray inert gas at the cutting focal point.

5. The underwater UUV laser cutting actuator as described in claim 4, characterized in that, Each rotary joint of the robotic arm module is coaxially integrated with a bidirectional torque sensor, a high-resolution photoelectric or magnetic absolute encoder, and a harmonic reducer; the rotary joint is covered with an oil-filled, pressure-resistant, compensated outer shell.

6. The underwater UUV laser cutting actuator as described in claim 1, characterized in that, The compensation control module includes a high-frequency inertial measurement unit, a six-dimensional torque sensor, and an acousto-optic composite sensing network. The high-frequency inertial measurement unit includes a three-axis microelectromechanical gyroscope and a three-axis accelerometer, and the six-dimensional torque sensor is installed between the end flange of the robotic arm module and the end effector. The acoustic-optical composite sensing network includes a multibeam forward-looking sonar, a three-dimensional structured light camera, and an underwater low-light camera. The acoustic-optical composite sensing network is used to acquire multi-source heterogeneous morphology data of the target surface in waters with extremely low visibility or high turbidity. The compensation control module adopts a heterogeneous computing architecture based on field-programmable gate arrays and digital signal processors; The field-programmable gate array is used to perform parallel acquisition of IMU high-frequency signals and multi-axis encoder signals and hardware-level Kalman filtering noise reduction at microsecond-level clock cycles. The digital signal processor is used to perform inverse kinematics calculation, trajectory planning, and multi-source data fusion algorithms.

7. A method for performing underwater UUV laser cutting, the method being used to execute an underwater UUV laser cutting actuator as described in any one of claims 1-6, characterized in that, include: Step S1: The UUV platform uses a multi-beam forward-looking sonar and a three-dimensional structured light camera to perceive the three-dimensional boundary of the target metal structure, solves the thrust distribution matrix of the thruster in the airborne DSP computing unit, navigates to the target surface within a set threshold range, and establishes the initial transformation matrix of the geodetic coordinate system, the UUV body coordinate system and the target working surface coordinate system. In step S2, the robotic arm module guides the execution module to approach the target surface, activates the magnetic array to provide normal preload based on the surface roughness, starts the negative pressure pump system, and dynamically adjusts the suction flow rate using a PID control algorithm until the anti-slip safety threshold is reached, thus completing the physical locking of the target area. Step S3: After physical locking is completed, the acousto-optic composite sensing network performs a fine scan of the cutting area, uses the ICP algorithm to perform local point cloud registration, and generates a smooth NURBS laser cutting spatiotemporal trajectory equation. In step S4, the underwater laser cutting unit starts to vibrate and outputs a continuous laser beam according to the set trajectory. At the same time, the compensation control module uses NDO to estimate the transient ocean current disturbance torque in real time and calculates the inverse solution of the Jacobian matrix to perform reverse micro-displacement compensation on each joint of the high-frequency driven robotic arm module. Step S5: During the entire cutting operation, the high-frequency solution of the end focus tracking error is used to feed the tracking error back to the main control circuit. If the error exceeds the preset tolerance or a sudden loss of pressure occurs, the laser power supply is cut off and the attitude holding or emergency escape logic is executed.

8. The underwater UUV laser cutting method as described in claim 7, characterized in that, In step S2, the suction process of the negative pressure pump system uses a closed-loop PID control algorithm to maintain absolute pressure, and the control voltage of the digital micro electromagnetic proportional valve is... The mathematical expression is: ； in, This is the gain coefficient for the controller proportional to the input. The gain coefficient for integration. The gain coefficient is the derivative. For pressure tracking error, The target safe negative pressure value is set based on the current water depth and the calculated tangential friction force requirements. This refers to the residual absolute pressure that is fed back in real time inside the cavity.

9. The underwater UUV laser cutting execution method as described in claim 8, characterized in that, In step S3, the smooth NURBS laser cutting spatiotemporal trajectory equation The mathematical expression is: ； in, This is an array of control points for the cutting shape extracted from the point cloud model. These are the weighting factors for the corresponding control points. for B-spline basis functions These are the normalized time and arc length node parameters.

10. The underwater UUV laser cutting execution method as described in claim 9, characterized in that, In step S4, the joint output torque of the robotic arm module Issued according to the following dynamic model: ； in, These are the actual angles of each joint in the robotic arm module. The angular velocities of each joint in the robotic arm module are given. The desired angular velocity is calculated by inverse solution of the NURBS trajectory equation. The desired angular acceleration is calculated by inverse solution of the NURBS trajectory equation. For position tracking error, The inertia matrix, The Coriolis force matrix, This is the underwater gravity and buoyancy compensation term. It is a positive definite proportional matrix. The differential gain matrix is... The nonlinear disturbance torque of the water flow collected by the observer; The underwater laser cutting unit obtains the local thickness of the metal structure based on the visual model scan. and the real-time displacement speed of the end effector of the robotic arm module Dynamically adjust the output power of the laser generator This results in an effective linear energy density acting on the cutting surface. Maintaining within a constant interval, its mathematical expression is: ； in, This represents the transmission efficiency coefficient of green laser light in a localized low-pressure cavitation channel. The diameter of the laser focal spot. For the target metal density, Specific heat capacity of the metal This refers to the temperature difference required for a metallic material to rise from ambient water temperature to its molten state.

Images