A collaborative control method for master-slave hydraulic robotic arms in deep-sea ROVs
By constructing a collaborative control strategy for the hydraulic robotic arm of a deep-sea ROV, the problem of compensating for the attitude stability of the ROV carrier due to the movement of the robotic arm was solved, improving operational safety and positioning accuracy, and realizing active compensation and synchronous movement of the carrier attitude.
Patent Information
- Application Number
- CN202611622542.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
The existing master-slave control method for deep-sea ROV hydraulic manipulators fails to effectively compensate for the attitude stability of the ROV carrier caused by the manipulator's motion, resulting in a decrease in operational safety and positioning accuracy.
By adopting a collaborative control method, a coupled dynamic model of the carrier-robotic arm and a master-slave following error model are constructed by collecting multi-source sensor data. A collaborative control strategy is generated, and the carrier compensation control quantity and joint drive control quantity are output to realize the active compensation and synchronous motion of the ROV carrier by the robotic arm's movements.
It significantly improves the safety and reliability of deep-sea operations, reduces the impact load of the robotic arm's dynamic movement on the carrier, and ensures master-slave following performance and collaborative operation capabilities through real-time monitoring and online adjustment.
Smart Images

Figure CN123103713A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer processing technology, and in particular to a collaborative control method for a master-slave hydraulic robotic arm in a deep-sea ROV. Background Technology
[0002] Remotely operated vehicles (ROVs) are core equipment for deep-sea resource exploration, oil and gas facility maintenance, scientific sampling, and engineering operations. The hydraulic robotic arm mounted on the ROV is the key actuator for completing underwater tasks. Currently, deep-sea hydraulic robotic arms generally employ a master-slave control system, where the operator controls the master arm at the surface, transmitting motion commands to the slave arm underwater, which then follows the master arm's movements to complete the task.
[0003] In underwater operations, the hydraulic manipulator and the ROV (Remotely Operated Vehicle) form a strongly coupled dynamic system. The manipulator's wide-range, high-load movements generate significant reaction forces and torques, which are transmitted to the ROV body through mechanical interfaces, causing attitude instability, position drift, and decreased positioning accuracy. However, existing master-slave control methods only focus on the position tracking of the manipulator joints, failing to incorporate the ROV's attitude stability into the collaborative control framework. This lack of a mechanism for actively compensating for the dynamic coupling between the manipulator's movement and the ROV's attitude seriously threatens the safety and reliability of underwater operations. Summary of the Invention
[0004] To address the aforementioned technical problems, the present invention provides a collaborative control method for a master-slave hydraulic robotic arm in a deep-sea ROV, the method comprising the following steps:
[0005] S01. Collect multi-source sensor data and extract ROV carrier state features and robotic arm joint state features;
[0006] S02. Construct a collaborative control decision model. The collaborative control decision model is based on the joint optimization mechanism of the carrier-manipulator coupling dynamics model and the master-slave following error model. The collaborative control strategy is dynamically generated according to the state characteristics of the ROV carrier and the joint state characteristics of the manipulator.
[0007] S03. Based on the aforementioned collaborative control strategy, output the carrier compensation control quantity and the joint drive control quantity respectively. The carrier compensation control quantity is used to offset the impact load effect of the robotic arm's movement on the ROV carrier, and the joint drive control quantity is used to drive the slave joint to follow the master hand's command.
[0008] S04. Generate a coordinated execution instruction sequence based on the carrier compensation control quantity and the drive control quantity of each joint. The coordinated execution instruction sequence includes the time synchronization execution logic of the carrier thruster instruction and the hydraulic servo valve instruction of each joint.
[0009] S05. Collect execution feedback data, update the ROV carrier status characteristics and robotic arm joint status characteristics, and judge the current collaborative control effect through the collaborative status evaluation mechanism. When it is determined that the carrier posture deviation or joint tracking deviation exceeds the preset threshold, online adjustment of the collaborative control strategy is triggered. If the operation termination condition is not met, return to S02 for iterative execution. If it is met, archive the operation results.
[0010] Preferably, the multi-source sensing data includes carrier posture data, carrier position data, joint angle data, joint angular velocity data, and joint hydraulic cylinder pressure data.
[0011] The ROV carrier state characteristics include the carrier attitude quaternion, the three-axis position vector and the three-axis velocity vector in the inertial coordinate system;
[0012] The joint state characteristics of the robotic arm include the angular position, angular velocity, and pressure difference between the two chambers of the hydraulic cylinder for each joint.
[0013] The construction of the ROV carrier state characteristics in step S01 includes the following steps:
[0014] The multi-source sensor data is filtered and standardized to define the carrier attitude feature dimension and the carrier position feature dimension. The carrier attitude feature dimension includes roll angle, pitch angle, yaw angle and their rate of change. The carrier position feature dimension includes the three-axis position coordinates of the ROV in the inertial coordinate system and their rate of change.
[0015] Based on the influence weights of each feature dimension on collaborative control, a unified ROV carrier state feature vector is generated through weighted integration.
[0016] Preferably, the construction of the robotic arm joint state features in step S01 includes the following steps:
[0017] Define joint motion feature dimension and joint drive feature dimension. The joint motion feature dimension includes the angular position, angular velocity and angular acceleration of each joint. The joint drive feature dimension includes the pressure value and pressure difference of the two chambers of the hydraulic cylinder of each joint.
[0018] Based on the forward kinematics model of the robotic arm, the angular positions of each joint are mapped to the pose and velocity state of the end effector.
[0019] By integrating the motion state, drive state, and end effector state of each joint, a joint state feature vector of the robotic arm is generated.
[0020] Preferably, the carrier-manipulator coupled dynamics model includes a multi-rigid-body dynamics sub-model of the manipulator, a six-degree-of-freedom dynamics sub-model of the ROV carrier, and a force / torque mapping sub-model of the manipulator on the carrier, which is used to describe the reaction force and reaction torque generated by the dynamic motion of the manipulator on the ROV carrier.
[0021] The construction of the carrier-robotic arm coupled dynamics model includes the following steps:
[0022] A multi-rigid-body dynamics model of the robotic arm is established using the Lagrange method to describe the dynamic relationship between the driving torque of each joint and the joint angle, angular velocity and angular acceleration.
[0023] Based on the pre-set ROV carrier six-degree-of-freedom dynamic model, the dynamic relationship between the carrier thruster output and the carrier's six-degree-of-freedom motion is described, and the hydrodynamic added mass coefficient and damping coefficient matrix are introduced.
[0024] Based on the angular velocity, angular acceleration and configuration parameters of each joint of the robotic arm, a force / torque mapping model of the robotic arm on the carrier is established;
[0025] The three sub-models are solved together to form a coupled dynamic model of the carrier-robotic arm.
[0026] Preferably, the master-slave follow error model includes a joint space follow error mapping sub-model and a task space pose error mapping sub-model, which are used to describe the deviation characteristics between the master hand command and the slave hand actual response.
[0027] The construction of the master-slave follower error model includes the following steps:
[0028] Acquire the angle command signals of each joint of the master hand and record the timestamp of the command signal;
[0029] Based on the actual angle feedback from each joint of the hand, the joint space following error between the master hand command and the slave hand actual response is calculated.
[0030] Based on the differential kinematics model of the robotic arm, the joint space following error is mapped to the position and posture errors in the task space;
[0031] By introducing a hydraulic long pipeline transmission delay characteristic model and a hydraulic valve control cylinder nonlinear characteristic model, the joint space following error is compensated for by delay and nonlinear correction, and the compensated master-slave following error is obtained.
[0032] Preferably, the joint optimization mechanism in step S02 takes the carrier attitude stability and master-slave following accuracy as joint optimization objectives, and uses an iterative optimization algorithm to jointly solve the ROV carrier state characteristics and the robotic arm joint state characteristics to dynamically generate a collaborative control strategy.
[0033] The joint optimization mechanism includes the following steps:
[0034] With the carrier attitude stability index and master-slave follow accuracy index as optimization objectives, a multi-objective optimization function is constructed using a linear weighting method.
[0035] Based on the carrier-manipulator coupled dynamics model, the attitude response of the ROV carrier and the joint response of the manipulator are predicted under the current candidate control input.
[0036] Based on the master-slave error model, the degree of deviation of the predicted response from the expected response is evaluated;
[0037] A cooperative control strategy that minimizes the multi-objective optimization function is obtained by using an iterative optimization algorithm.
[0038] Preferably, the carrier compensation control amount in step S03 includes vertical thrust compensation amount, horizontal longitudinal thrust compensation amount, horizontal lateral thrust compensation amount, and yaw moment compensation amount.
[0039] The generation of the carrier compensation control quantity includes the following steps:
[0040] Based on the carrier attitude compensation component in the aforementioned cooperative control strategy, calculate the required compensation thrust vector in each degree of freedom direction in the inertial coordinate system.
[0041] Based on the thruster spatial arrangement configuration of the ROV, a thruster configuration matrix is constructed, and the compensation thrust vector is allocated to each thruster using the pseudo-inverse method to generate thrust commands for each thruster.
[0042] Preferably, the joint drive control quantity in step S03 includes the opening command and opening change rate command of the hydraulic servo valve of each joint.
[0043] The generation of the drive control quantities for each joint includes the following steps:
[0044] Based on the joint drive components in the aforementioned collaborative control strategy, the desired angular position and angular velocity of each joint are calculated.
[0045] Based on the flow continuity equation and force balance equation of each joint valve-controlled symmetrical hydraulic cylinder, the desired angular position and angular velocity are converted into control current signals of each joint hydraulic servo valve.
[0046] The control current signal is feedforward corrected by introducing hydraulic system pressure fluctuation compensation coefficient and long pipeline transmission delay compensation coefficient.
[0047] Preferably, the time synchronization in step S04 includes the following steps:
[0048] Based on the difference in response characteristics between the carrier thruster and the hydraulic servo valve, a timestamp alignment mechanism for instructions and an execution timing offset configuration mechanism are established to ensure that the carrier thruster and the hydraulic servo valves of each joint operate in coordination within the same control cycle.
[0049] The generation of the cooperative execution instruction sequence includes the following steps:
[0050] Based on the difference in step response time and bandwidth between the carrier thruster and the hydraulic servo valve, the execution sequence and time offset of each instruction are determined.
[0051] A timestamp alignment mechanism for instructions is constructed based on a clock reference of the global control cycle.
[0052] An embedded actuator abnormal state detection logic is used to activate a safety protection strategy when the response feedback of any thruster or hydraulic servo valve exceeds the preset safety range.
[0053] Preferably, the collaborative state evaluation mechanism in step S05 includes:
[0054] The carrier attitude deviation threshold and the tracking deviation threshold of each joint are preset. The effect of cooperative control is judged by quantifying the difference between the updated state characteristics and the expected state characteristics.
[0055] If the carrier attitude deviation is lower than the carrier attitude deviation threshold and the tracking deviation of each joint is lower than the tracking deviation threshold of each joint, then the current cooperative control strategy is maintained.
[0056] If the carrier attitude deviation is not lower than the carrier attitude deviation threshold or the tracking deviation of any joint is not lower than the corresponding tracking deviation threshold of each joint, it is determined to be a coordination mismatch, triggering online adjustment of the coordination control strategy;
[0057] The online adjustment of the collaborative control strategy includes at least one of the following methods:
[0058] Adjust the coupling compensation parameters in the carrier-robotic arm coupled dynamics model;
[0059] Adjust the time delay compensation parameters and nonlinear correction parameters in the master-slave follower error model;
[0060] The multi-objective optimization function is re-solved to generate an updated cooperative control strategy;
[0061] Switch to standby control mode;
[0062] The operation termination conditions include at least one of the following:
[0063] The progress of the task completion target has reached the preset target threshold;
[0064] The degree of difference in state characteristics across multiple consecutive iterations is lower than a preset stability threshold;
[0065] An unrecoverable exception is triggered, and the job execution conditions cannot be met even after exception handling.
[0066] The present invention has at least the following beneficial effects:
[0067] 1. Compared with the existing technology where the carrier control and the robotic arm control are independent, this invention realizes active compensation and collaborative optimization of the ROV carrier attitude during the robotic arm operation, which can minimize the impact of the robotic arm's dynamic movement on the carrier's impact load, and greatly improve the safety and mission reliability of deep-sea operations.
[0068] 2. This invention introduces a hydraulic long pipeline transmission delay characteristic model and a hydraulic valve control cylinder nonlinear characteristic model into the master-slave following error model. By identifying and compensating for the delay and nonlinear parameters online, and by feedforward correction of the control current signal, the joint space following error and task space pose error between the master hand command and the slave hand actual action are effectively suppressed.
[0069] 3. This invention monitors the carrier attitude deviation and joint tracking deviation in real time through a collaborative state assessment mechanism and sets a clear deviation threshold. When a collaborative mismatch is detected, the control strategy is adjusted online to ensure that the system maintains good master-slave following performance and collaborative operation capability in the complex environment of the deep sea.
[0070] 4. This invention introduces a hydraulic long-pipeline transmission delay characteristic model and a hydraulic valve-controlled cylinder nonlinear characteristic model into the master-slave following error model. By identifying and compensating for the delay and nonlinear parameters online, and by feedforward correction of the control current signal, the joint space following error and task space pose error between the master's command and the slave's actual action are effectively suppressed. Simultaneously, this invention monitors the carrier's attitude deviation and joint tracking deviation in real time through a cooperative state evaluation mechanism, and sets a clear deviation threshold. When cooperative mismatch is detected, the control strategy is adjusted online, ensuring that the system maintains good master-slave following performance and cooperative operation capability in the complex environment of the deep sea. Attached Figure Description
[0071] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0072] Figure 1 The flowchart illustrates a collaborative control method for a master-slave hydraulic robotic arm in a deep-sea ROV, as provided in Embodiment 1 of the present invention. Detailed Implementation
[0073] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0074] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0075] Example 1
[0076] This embodiment provides a collaborative control method for a master-slave hydraulic robotic arm in a deep-sea ROV, the method including the following steps, such as... Figure 1 As shown:
[0077] S01. Collect multi-source sensor data and extract ROV carrier state features and robotic arm joint state features;
[0078] The aforementioned multi-source sensor data includes carrier attitude data, carrier position data, joint angle data, joint angular velocity data, and joint hydraulic cylinder pressure data.
[0079] The ROV's state characteristics include the vehicle's attitude quaternion, the three-axis position vector and the three-axis velocity vector in the inertial coordinate system;
[0080] The joint status characteristics of the robotic arm include the angular position, angular velocity, and pressure difference between the two chambers of the hydraulic cylinder;
[0081] Furthermore, the construction of ROV vector state characteristics in step S01 includes the following steps:
[0082] Multi-source sensor data is filtered and standardized to define the carrier attitude feature dimension and the carrier position feature dimension. The carrier attitude feature dimension includes roll angle, pitch angle, yaw angle and their rate of change. The carrier position feature dimension includes the three-axis position coordinates of the ROV in the inertial coordinate system and their rate of change.
[0083] Based on the influence weights of each feature dimension on collaborative control, a unified ROV carrier state feature vector is generated through weighted integration.
[0084] Furthermore, the construction of the robotic arm joint state features in S01 includes the following steps:
[0085] Define the joint motion feature dimension and the joint drive feature dimension. The joint motion feature dimension includes the angular position, angular velocity and angular acceleration of each joint. The joint drive feature dimension includes the pressure value and pressure difference of the two chambers of the hydraulic cylinder of each joint.
[0086] Based on the forward kinematics model of the robotic arm, the angular positions of each joint are mapped to the pose and velocity state of the end effector.
[0087] By integrating the motion state, drive state, and end effector state of each joint, a joint state feature vector of the robotic arm is generated.
[0088] Specifically, in the raw signal preprocessing stage, the system filters and standardizes the collected sensor data to eliminate measurement noise and dimensional differences between different physical quantities. Specifically, for the carrier attitude data, the system uses the three-axis angular velocities output by the IMU to solve the attitude quaternions through iterative updates of quaternion differential equations. Simultaneously, it combines data from the accelerometer and magnetometer using extended Kalman filtering or complementary filtering for fusion correction to suppress gyroscope integral drift. Then, based on actual needs, the quaternions are converted into roll angle, pitch angle, yaw angle, and their rate of change as attitude feature dimensions. For the carrier position data, the system uses the three-axis velocities in the carrier coordinate system provided by the Doppler velocimeter and combines them with depth gauge data. After coordinate system rotation transformation, these are converted to the inertial coordinate system, and the three-axis position coordinates are obtained through integration. The position coordinates are then differentiated over time to obtain the three-axis velocities as position feature dimensions. For the angle data of each joint of the robotic arm, the system directly reads the angle position of each joint from the angle encoder and obtains the angular velocity and angular acceleration from the first and second derivatives of the angle signal. For the hydraulic cylinder pressure data of each joint, the system directly collects the pressure values of the rod-side and rodless-side chambers and calculates the pressure difference between the two chambers.
[0089] In the feature vector integration and generation stage, the system generates unified carrier state feature vectors and robotic arm joint state feature vectors through weighted integration based on the influence weights of each feature dimension on the quality of collaborative control. For the carrier state feature vector, it integrates attitude quaternions, three-axis position and three-axis velocity in the inertial frame. The weight coefficients are selected based on the degree of coupling influence of each dimension on the pose accuracy of the robotic arm's end effector. Small changes in attitude features can significantly affect the end effector position through the base; therefore, attitude features are given higher weights in precision operation scenarios. For the robotic arm joint state feature vector, its construction process consists of two sub-steps:
[0090] First, at the joint space level, the angular position, angular velocity, angular acceleration of each joint, as well as the pressure value and pressure difference between the two chambers of the hydraulic cylinder are defined as the motion characteristic dimension and the driving characteristic dimension.
[0091] Secondly, at the task space level, the mapping relationship between the joint angles and the end effector pose is established using the DH positive kinematic parameters of the robotic arm. That is, the three-axis position and attitude of the end effector in the base coordinate system are obtained by multiplying the homogeneous transformation matrix, and then the angular velocity of each joint is mapped to the linear velocity and angular velocity of the end effector by the Jacobian matrix.
[0092] Ultimately, the system integrates the motion state, drive state, and end effector state of each joint into a unified robotic arm joint state feature vector.
[0093] S02. Construct a collaborative control decision model. The collaborative control decision model is based on the joint optimization mechanism of the carrier-manipulator coupling dynamics model and the master-slave following error model. It dynamically generates collaborative control strategies according to the ROV carrier state characteristics and the manipulator joint state characteristics.
[0094] Furthermore, the carrier-manipulator coupled dynamics model includes a multi-rigid-body dynamics sub-model for the manipulator, a six-degree-of-freedom dynamics sub-model for the ROV carrier, and a force / torque mapping sub-model for the manipulator's action on the carrier, which is used to describe the reaction force and reaction torque generated by the dynamic motion of the manipulator on the ROV carrier.
[0095] The construction of the carrier-robotic arm coupled dynamics model includes the following steps:
[0096] A multi-rigid-body dynamics model of the robotic arm is established using the Lagrange method to describe the dynamic relationship between the driving torque of each joint and the joint angle, angular velocity and angular acceleration.
[0097] Based on the pre-set ROV carrier six-degree-of-freedom dynamic model, the dynamic relationship between the carrier thruster output and the carrier's six-degree-of-freedom motion is described, and the hydrodynamic added mass coefficient and damping coefficient matrix are introduced.
[0098] Based on the angular velocity, angular acceleration and configuration parameters of each joint of the robotic arm, a force / torque mapping model of the robotic arm on the carrier is established;
[0099] The three sub-models are solved together to form a coupled dynamic model of the carrier-robotic arm.
[0100] Furthermore, the master-slave follower error model includes a joint space follower error mapping sub-model and a task space pose error mapping sub-model, which are used to describe the deviation characteristics between the master hand command and the slave hand's actual response.
[0101] The construction of the master-slave error model includes the following steps:
[0102] Acquire the angle command signals of each joint of the master hand and record the timestamp of the command signal;
[0103] Based on the actual angle feedback from each joint of the hand, the joint space following error between the master hand command and the slave hand actual response is calculated.
[0104] Based on the differential kinematics model of the robotic arm, the joint space following error is mapped to the position and posture errors in the task space;
[0105] By introducing a hydraulic long pipeline transmission delay characteristic model and a hydraulic valve control cylinder nonlinear characteristic model, the joint space following error is compensated for by delay and nonlinear correction, and the compensated master-slave following error is obtained.
[0106] Specifically, the Lagrangian method is used to construct the multi-rigid-body dynamics sub-model of the robotic arm. The principle of this method is to construct a Lagrangian function based on the difference between the system's kinetic and potential energy. By taking partial and total derivatives with respect to generalized coordinates (angles of each joint), the dynamic equations relating the driving torque of each joint to the joint angles, angular velocities, and angular accelerations are derived. In the specific implementation, firstly, the inertial parameters such as the mass, center of mass position, and inertia tensor of each link of the robotic arm, as well as the DH parameters of each joint (link length, torsion angle, joint distance, etc.), are determined. Then, the position, velocity, and acceleration of each link in the base coordinate system are calculated, thereby obtaining the total kinetic energy (including the translational and rotational kinetic energy of each link) and total potential energy (gravitational potential energy; buoyancy potential energy must also be considered during underwater operations). Finally, the Lagrangian equations are substituted into the partial differential equations, and the result is rearranged to obtain the form... The standard dynamic equations are given, where M is the mass matrix (including inertial terms), C is the Coriolis force and centrifugal force terms, G is the gravity and buoyancy terms, τ is the driving torque vector of each joint, and θ is the angle vector of each joint of the robotic arm. For each joint angular velocity vector, Let be the angular acceleration vector of each joint. This equation reveals the quantitative relationship between the joint driving torque and the joint motion state, and can also be used in reverse to calculate the required output driving torque of each joint under a known joint motion trajectory.
[0107] The ROV (Return on Vehicle) six-degree-of-freedom dynamic sub-model is constructed using the classic six-degree-of-freedom dynamic equations for ships and underwater vehicles. These equations describe the motion of the vehicle in six degrees of freedom (puff, sway, heave, roll, pitch, and yaw) in the vehicle coordinate system. Specifically, the position and attitude vectors of the vehicle in the inertial coordinate system are first defined as follows: η=[x,y,z,ϕ,θ] v ,ψ] T Where x, y, z are the three-axis position coordinates in the inertial coordinate system, ϕ is the roll angle, and θ is the angle of rotation. v Let ψ be the pitch angle, ψ be the yaw angle, and η be the attitude vector.
[0108] Define the velocity vector in the carrier coordinate system as: ν=[u,v,w,p,q,r] T Where u, v, w are the three-axis velocities, p, q, r are the three-axis angular velocities, and ν is the acceleration vector in the carrier coordinate system. Then, based on the Newton-Euler equations, the six-degree-of-freedom dynamic equations of the carrier are established: In this equation, The inertia matrix includes the carrier's own mass and its added mass (hydrodynamic added mass coefficient). The Coriolis centripetal force matrix, The damping coefficient matrix includes both linear and nonlinear damping (reflecting the resistance of water to the motion of the carrier). These are restoring force and restoring torque (i.e., the force and torque generated by gravity and buoyancy). For thrust output, Environmental disturbance forces (such as ocean currents). Let ν represent the reaction force and reaction torque exerted by the robotic arm on the carrier (i.e., the coupling term), ν be the acceleration vector in the carrier coordinate system, and η be the attitude vector. The average velocity is given in the carrier coordinate system.
[0109] The construction of the force / torque mapping sub-model of the robotic arm's action on the carrier establishes a mapping relationship between the joint motion of the robotic arm and the reaction force / torque on the carrier. Its implementation principle is based on the Newton-Euler recursive algorithm in multibody system dynamics: First, based on the current angle, angular velocity, and angular acceleration of each joint, as well as the inertial parameters of each link, the velocity and acceleration of each link are calculated using a forward recursion from the base to the end effector; then, the constraint forces and torques at each joint are calculated using a reverse recursion from the end effector to the base; finally, the total reaction force and reaction torque at the base are transformed to the carrier coordinate system, which is the force / torque exerted by the robotic arm on the carrier. This mapping process can be summarized as follows: ,in: The equivalent force / torque vector exerted by the robotic arm on the carrier has been transformed into the carrier coordinate system; J arm For the Jacobian matrix of the robotic arm, This is the transpose of the Jacobian matrix of the robotic arm; This represents the driving torque vector of each joint of the robotic arm; This represents the force / torque of each joint driving torque mapped onto the carrier or base via the Jacobian transpose, and is not the "joint torque vector" itself; This refers to the inertial coupling term generated by the movement of the robotic arm.
[0110] The joint solution of the three sub-models constitutes a complete coupled dynamic model of the carrier-manipulator. In the joint solution, the system state variables simultaneously include the six-degree-of-freedom motion state of the carrier and the motion state of each joint of the manipulator. The solution process is as follows: given the angular velocity, angular acceleration, and configuration parameters of each joint at the current moment, the reaction force and reaction torque of the manipulator on the carrier are calculated through the force / torque mapping sub-model. These are then substituted as external disturbance forces into the six-degree-of-freedom dynamic equations of the ROV carrier to obtain the acceleration response of the carrier. Simultaneously, the base motion caused by the carrier's attitude change is substituted into the multi-rigid-body dynamic equations of the manipulator as the base excitation of the manipulator. The information interaction in the two directions is advanced step by step through numerical integration methods (such as the fourth-order Runge-Kutta method) to achieve joint dynamic simulation of the coupled system. This joint model comprehensively reflects the impact load transmission mechanism of the manipulator's motion on the ROV carrier and the reverse influence of the ROV carrier's attitude change on the manipulator's end-effector pose, providing an accurate dynamic model foundation for subsequent cooperative compensation control.
[0111] The joint space following error mapping sub-model directly compares the differences between the master hand's commands and the slave hand's responses at the joint level. In implementation, the system first acquires the angle command signals of each joint of the master hand and accurately records the timestamp of each frame of command signals to establish a correspondence between the command issuance time and the expected arrival time. Simultaneously, angle encoders installed on each joint of the slave hand collect the actual angle feedback of each joint in real time. For each control cycle, the system compares the actual angle of each joint of the slave hand at the current moment with the command angle issued by the master hand at the same moment, joint by joint, to calculate the joint space following error vector e. θ (t)=θ m (t)−θ s (t), where θ m (t) represents the joint angle command vector issued by the master hand at time t; θ s (t) represents the angle vectors of each joint actually fed back by the hand at time t; e θ (t) is the joint space following error vector (i.e., joint space error).
[0112] The task space pose error mapping sub-model maps the joint space following error to the task space, transforming the deviation of each joint into a physical quantity reflecting the difference between the actual and desired pose of the end effector, thus more intuitively reflecting the impact of control deviation on the operation. In implementation, based on the Jacobian matrix J(θ) of the robotic arm, the joint space error e is... θ (t) Position error mapped to the task space e p and attitude error e o The relationship is [e p e o ] = J(θ)·e θ (t), where θ(t) is the angle vector of each joint of the robotic arm; J(θ) is the Jacobian matrix of the robotic arm at joint angle θ(t); e p The task space location error vector; [e p e o [ ] represents the attitude error vector in the mission space; J(θ)·e θ (t) represents the stitched task space pose error vector. This mapping reveals the contribution of each joint error to the end effector's position and orientation. For example, the angular error of the upper arm joint near the base significantly amplifies the pose deviation of the end effector, while the error of the forearm joint near the end effector has a relatively smaller impact. Through this mapping, the system can evaluate the overall positioning accuracy of the end effector in the task space and determine whether it meets the requirements for high-precision operation.
[0113] The hydraulic system characteristic correction sub-model incorporates the transmission delay characteristics of long hydraulic pipelines and the nonlinear characteristics of hydraulic valve-controlled cylinders to compensate for delays and correct nonlinearities in joint space following errors. In implementation, to address the transmission delay of long hydraulic pipelines, the system stores the historical command sequence of the master arm and estimates the desired angle values of each joint of the master arm at the current moment based on power curve fitting or Kalman filtering prediction methods, thus eliminating phase lag caused by signal transmission time delay. Regarding the nonlinear characteristics of hydraulic valve-controlled cylinders, the system establishes a flow-pressure-displacement relationship model for the valve-controlled cylinders, with the core relationship being: ,in, The flow rate through the throttle orifice; For flow coefficient; This refers to the flow area of the valve orifice, i.e., the flow area corresponding to the valve orifice opening degree. The pressure difference across the valve port; The hydraulic oil density is used. This relationship reveals the inherent nonlinear characteristics of hydraulic actuators. When the valve opening and differential pressure change, the flow rate exhibits a nonlinear response, leading to a nonlinear deviation between the joint speed response and the commanded speed. Based on this model, the system calculates the deviation between the expected flow rate and the actual flow rate at each joint of the actuator in real time, and adds this deviation to the joint space following error, forming the master-slave following error after time delay compensation and nonlinear correction. The corrected error value more accurately reflects the actual control deviation between the master and slave, providing accurate input for determining the type of compensation action in the subsequent collaborative decision-making system.
[0114] S03. Based on the collaborative control strategy, the carrier compensation control quantity and the joint drive control quantity are output respectively. The carrier compensation control quantity is used to offset the impact load of the robotic arm movement on the ROV carrier, and the joint drive control quantity is used to drive the slave joint to follow the master hand command.
[0115] Furthermore, the carrier compensation control quantities include vertical thrust compensation, horizontal longitudinal thrust compensation, horizontal lateral thrust compensation, and yaw moment compensation.
[0116] The generation of carrier compensation control quantities includes the following steps:
[0117] Based on the carrier attitude compensation component in the cooperative control strategy, the required compensation thrust vector in each degree of freedom direction in the inertial coordinate system is calculated.
[0118] Based on the thruster spatial arrangement configuration of the ROV, a thruster configuration matrix is constructed, and the compensation thrust vector is allocated to each thruster using the pseudo-inverse method to generate thrust commands for each thruster.
[0119] Furthermore, the drive control quantities for each joint include the opening degree command and the opening degree change rate command of the hydraulic servo valve of each joint.
[0120] The generation of drive control quantities for each joint includes the following steps:
[0121] Based on the joint drive components in the cooperative control strategy, the expected angular position and angular velocity of each joint are calculated.
[0122] Based on the flow continuity equation and force balance equation of each joint valve-controlled symmetrical hydraulic cylinder, the desired angular position and angular velocity are converted into control current signals of each joint hydraulic servo valve.
[0123] A hydraulic system pressure fluctuation compensation coefficient and a long pipeline transmission delay compensation coefficient are introduced to perform feedforward correction on the control current signal.
[0124] Specifically, the above embodiment transforms the collaborative control strategy obtained from the joint optimization solution in step S02 into specific thrust commands for the ROV carrier thruster and specific control signals for the hydraulic servo valves of each joint of the robotic arm, thus realizing the conversion from "macro-control decision" to "micro-execution command". Since the collaborative control strategy includes both carrier attitude compensation components and joint drive components, and these two types of components have completely different physical properties, actuator characteristics, and signal generation logic, step S03 uses different conversion paths according to the characteristics of the two types of components to generate carrier compensation control quantities and joint drive control quantities respectively, ensuring that the control commands can be accurately responded to and effectively executed by the underlying actuators.
[0125] The role of the carrier compensation control variable is to actively counteract the reaction impact load transmitted from the robotic arm's motion to the ROV carrier, ensuring the carrier's attitude remains stable during robotic arm operation and preventing the absolute positioning accuracy of the robotic arm's end effector from being affected by carrier attitude drift. The cooperative control strategy includes attitude compensation components for the six degrees of freedom (vertical, longitudinal, lateral, roll, pitch, and yaw) of the carrier. These components are given in the form of the magnitude and direction of the required compensation thrust in each degree of freedom direction in the inertial coordinate system. However, the number of thrusters on the ROV carrier is usually greater than the number of motion degrees of freedom, and the installation position and thrust direction of each thruster are fixed in the carrier coordinate system. Therefore, it is necessary to transform the compensation thrust vector in the inertial coordinate system to the carrier coordinate system and then distribute it to each thruster. In specific implementation, a thruster configuration matrix is first constructed based on the installation position coordinates and thrust direction vector of each thruster on the ROV carrier. This matrix describes the thrust / torque contribution of each thruster's unit thrust to the six degrees of freedom of the carrier. Then, the pseudo-inverse method is used to obtain the generalized inverse matrix of the thruster configuration matrix. The compensation thrust vector in the inertial coordinate system is mapped to the expected thrust value of each thruster through this generalized inverse matrix, achieving optimal thrust allocation. Since the pseudo-inverse method can obtain the minimum norm solution with optimal energy when the number of thrusters is greater than the number of degrees of freedom, the allocation result can achieve the expected compensation thrust effect with minimal power consumption. Finally, the expected thrust value of each thruster is converted into the corresponding current command according to the thrust-current calibration curve of the thruster itself, and sent to the drive controller of each thruster for execution.
[0126] The role of the joint drive control quantities is to drive the robotic arm's joints to perform posture-following movements according to the master hand's operating instructions, enabling the slave hand to reproduce the master hand's desired motion trajectory. The cooperative control strategy includes the desired angular position and desired angular velocity of each joint. These desired values are reference trajectory points obtained through comprehensive optimization after considering carrier posture compensation and the dynamic characteristics of the hydraulic system. Since the joints of the deep-sea hydraulic robotic arm typically use valve-controlled symmetrical hydraulic cylinders as actuators, and the joint movement is driven by the flow rate into the hydraulic cylinder controlled by the servo valve, it is necessary to convert the desired angular position and angular velocity into control current signals for the hydraulic servo valve. In specific implementation, firstly, based on the kinematic relationship of each joint, the desired joint angular position and angular velocity are converted into the desired piston displacement and desired motion speed of the corresponding hydraulic cylinder. Then, based on the flow continuity equation and force balance equation of the valve-controlled symmetrical hydraulic cylinder, an inverse hydraulic drive model of the joint is established. This model describes the valve core displacement or control current required by the servo valve to achieve the desired piston motion state. By substituting the desired piston displacement and velocity into the inverse model and combining the actual pressure feedback values of the two chambers of the hydraulic cylinder at the current moment, the desired opening degree of the servo valve can be calculated and then converted into a control current signal. Based on this, two types of feedforward corrections are introduced: one is a hydraulic system pressure fluctuation compensation coefficient. When the system oil supply pressure fluctuates due to the simultaneous action of multiple joints, this coefficient corrects the control current in real time based on the deviation between the measured oil supply pressure and the nominal pressure, compensating for the impact of pressure fluctuations on flow control accuracy. The other is a long pipeline transmission delay compensation coefficient. Since the hydraulic pipeline from the surface control console to the underwater robotic arm can be hundreds of meters long, there is a non-negligible delay in the transmission of hydraulic signals. This coefficient predicts future pressure loss changes based on the pipeline model, applying advance correction to the control current to eliminate the constraint of pipeline delay on joint response speed. After the above conversion and correction, the opening degree command and opening change rate command of each joint's hydraulic servo valve are finally generated and sent to the servo valve drive amplifier for execution, driving the slave joint to achieve smooth and precise master hand following movement.
[0127] S04. Generate a coordinated execution instruction sequence based on the carrier compensation control quantity and the drive control quantity of each joint. The coordinated execution instruction sequence includes the time synchronization execution logic of the carrier thruster instruction and the hydraulic servo valve instruction of each joint.
[0128] Furthermore, the time synchronization execution in the above embodiments includes the following steps:
[0129] Based on the difference in response characteristics between the carrier thruster and the hydraulic servo valve, a timestamp alignment mechanism for instructions and an execution timing offset configuration mechanism are established to ensure that the carrier thruster and the hydraulic servo valves of each joint operate in coordination within the same control cycle.
[0130] The generation of a cooperative execution instruction sequence includes the following steps:
[0131] Based on the difference in step response time and bandwidth between the carrier thruster and the hydraulic servo valve, the execution sequence and time offset of each instruction are determined.
[0132] A timestamp alignment mechanism for instructions is constructed based on a clock reference of the global control cycle.
[0133] An embedded actuator abnormal state detection logic is used to activate a safety protection strategy when the response feedback of any thruster or hydraulic servo valve exceeds the preset safety range.
[0134] Specifically, the generated carrier compensation control quantity and joint drive control quantity are converted into a time-synchronized command sequence that can be precisely executed by the ROV carrier thruster and the hydraulic servo valves of each joint of the robotic arm within the same control cycle.
[0135] The first step in achieving the aforementioned time synchronization is to establish a dynamic response characteristic database for each actuator based on the difference in step response time and bandwidth between the carrier thruster and the hydraulic servo valve. This database needs to be established through system identification experiments: under different load conditions on the robotic arm, standard step signals are applied to the hydraulic servo valves of each joint and the thrusters of each degree of freedom of the carrier, recording the time required from command issuance to the actuator output reaching 90% of its steady-state value (i.e., rise time), the overshoot during the response process, and key parameters such as the system's natural frequency and damping ratio. By fitting the experimental data, the transfer function model or equivalent time constant of each actuator can be obtained. Based on this, for a length T... s The global control cycle (typically 50ms~100ms) is used as the global clock reference. The advance of command issuance is calculated for each actuator: For hydraulic servo valves with fast response speed, the command issuance time should be set with a positive offset relative to the start point of the control cycle, that is, the issuance should be appropriately delayed so that the actual action time is aligned with the middle or end of the control cycle; For carrier thrusters with slow response speed, the command should be issued immediately at the start of the control cycle so as to take advantage of its longer response time for preheating and pressure building process, so that its thrust output reaches the target value at the actual demand time.
[0136] The second step is to establish a timestamp alignment mechanism for instructions. At the start of each global control cycle of the central controller, a synchronization pulse signal is generated by the clock synchronization module. All actuator instructions are marked with a uniform timestamp when generated. This timestamp is based on the global clock and indicates the control cycle number to be executed by the instruction and the relative phase between cycles. Since the communication network inside the ROV system usually uses CAN bus or Ethernet, different actuator nodes may have different data packet transmission delays due to different communication distances and network loads. Therefore, the timestamp alignment mechanism also needs to include a communication delay compensation function: when the central controller performs clock synchronization handshakes with each actuator node, it measures and records the bidirectional communication delay between each node and the control center. When sending instructions, the communication delay of each actuator is superimposed with the aforementioned timing offset to ensure that each instruction, after undergoing different transmission time adjustments on different transmission paths, can arrive at the instruction effective time at each actuator simultaneously.
[0137] The third step is to embed actuator abnormal state detection logic and safety protection strategies. During command execution, each thruster and hydraulic servo valve transmits its actual response signals (such as the actual speed or thrust feedback of the thruster, and the valve core displacement feedback or actual flow feedback of the hydraulic servo valve) back to the central controller in real time. Within each control cycle, the controller compares the transmitted actual response value with the predicted response value for the current control cycle. If the deviation exceeds the preset safety range, the actuator is determined to be in an abnormal state, triggering the safety protection strategy. The safety protection strategy adopts a tiered response mechanism: for minor deviations, local compensation is triggered, and the controller recalculates the compensation control quantity for the current cycle, replacing the corresponding component in the original command; for severe deviations, the system is triggered to stop immediately, and a zero-reset command is sent to the carrier thruster and the robotic arm hydraulic servo valve, allowing the carrier and robotic arm to transition to a safe stationary state in the shortest possible time, preventing system cascading instability caused by a single actuator failure and ensuring absolute safety in deep-sea operations.
[0138] S05. Collect execution feedback data, update the ROV carrier status characteristics and robotic arm joint status characteristics, and judge the current collaborative control effect through the collaborative status evaluation mechanism. When it is determined that the carrier posture deviation or joint tracking deviation exceeds the preset threshold, online adjustment of the collaborative control strategy is triggered. If the operation termination condition is not met, return to S02 for iterative execution. If it is met, archive the operation results.
[0139] Furthermore, the collaborative state evaluation mechanism in the above embodiments includes:
[0140] The carrier attitude deviation threshold and the tracking deviation threshold of each joint are preset. The effect of cooperative control is judged by quantifying the difference between the updated state characteristics and the expected state characteristics.
[0141] If the carrier attitude deviation is lower than the carrier attitude deviation threshold and the tracking deviation of each joint is lower than the tracking deviation threshold of each joint, then the current cooperative control strategy is maintained.
[0142] If the carrier attitude deviation is not lower than the carrier attitude deviation threshold or the tracking deviation of any joint is not lower than the corresponding tracking deviation threshold of each joint, it is determined to be a coordination mismatch, triggering online adjustment of the coordination control strategy;
[0143] Online adjustments to the collaborative control strategy include at least one of the following methods:
[0144] Adjust the coupling compensation parameters in the carrier-robotic arm coupled dynamics model;
[0145] Adjust the time delay compensation parameters and nonlinear correction parameters in the master-slave follower error model;
[0146] The multi-objective optimization function is re-solved to generate an updated cooperative control strategy;
[0147] Switch to standby control mode;
[0148] The termination conditions for the operation include at least one of the following:
[0149] The progress of the task completion target has reached the preset target threshold;
[0150] The degree of difference in state characteristics across multiple consecutive iterations is lower than a preset stability threshold;
[0151] An unrecoverable exception is triggered, and the job execution conditions cannot be met even after exception handling.
[0152] Specifically, after the hand joint drive sequence generated in step S04 is sent to the hydraulic servo valves of each joint and the ROV carrier thruster, the system executes one control cycle and then restarts a new round of sensor data acquisition. This process is consistent with the acquisition method in step S01, acquiring the ROV carrier's attitude response (including roll angle, pitch angle, yaw angle, and angular velocity) through an IMU, the carrier's position response through a Doppler velocimeter and depth gauge, the actual rotation angle of each joint through angle encoders, and the actual pressure response through pressure sensors in both chambers of each joint's hydraulic cylinder. This newly acquired raw sensor data is input into the same state feature extraction module as in step S01. After filtering, standardization, and weighted integration, updated ROV carrier state feature vectors and robotic arm joint state feature vectors are obtained.
[0153] Furthermore, the system internally presets carrier attitude deviation thresholds and joint tracking deviation thresholds. The carrier attitude deviation thresholds are determined during the system design phase based on the ROV carrier's attitude stability margin requirements. For example, the deviation thresholds for roll and pitch angles are typically set to ±3°~±5°, and the yaw angle deviation threshold is set to ±5°~±8°. The tracking deviation thresholds for each joint are determined comprehensively based on the robotic arm's end-effector pose accuracy requirements and the transmission accuracy of each joint, typically ±1%~±3% of the corresponding joint's range of motion. After obtaining the updated state feature vector, the system compares the updated carrier attitude features with the desired attitude features to calculate the carrier attitude deviation. This desired attitude feature is the carrier's desired attitude corresponding to the master operator's command (usually the initial attitude when maintaining the current attitude without robotic arm disturbance). Simultaneously, the updated joint angle positions are compared with the master operator's desired command after time delay compensation in step S1 to calculate the tracking deviation of each joint. Then, the calculated deviation is compared with a preset threshold: if the carrier attitude deviation is lower than the carrier attitude deviation threshold, and the tracking deviation of each joint is lower than the corresponding joint tracking deviation threshold, the current collaborative control effect is considered good, the current collaborative control strategy is maintained, and the system directly enters the operation termination condition judgment stage. Conversely, if the carrier attitude deviation is not lower than the carrier attitude deviation threshold, or if the tracking deviation of any joint is not lower than the corresponding threshold, the system is considered to have experienced collaborative mismatch, meaning that the movement of the robotic arm has caused a significant impact disturbance to the carrier attitude, or the robotic arm joints have failed to effectively follow the master hand command. In this case, it is necessary to trigger online adjustment of the collaborative control strategy.
[0154] The purpose of online adjustment is to suppress coordination mismatch and restore the system to a stable following state by adjusting control strategy parameters or switching control modes without interrupting system operation. This invention provides four online adjustment methods; the specific method or combination thereof used depends on the type and severity of the coordination mismatch.
[0155] The first approach involves adjusting the coupling compensation parameters in the carrier-manipulator coupled dynamics model within the framework of the current multi-objective optimization function. Specifically, this involves adjusting the compensation gain coefficient of the manipulator base reaction force on the carrier's attitude disturbance in the dynamics model, thereby increasing or decreasing the strength of the coupling compensation and enabling the carrier thruster to more actively counteract the impact loads caused by the manipulator's movement.
[0156] The second approach involves adjusting the delay compensation and nonlinear correction parameters in the master-slave tracking error model. When it is determined that the joint tracking deviation is mainly due to communication delay or the nonlinear response of the hydraulic actuator, the tracking performance of the joint can be improved by increasing the order of the delay prediction compensation or adjusting the nonlinear friction compensation coefficient.
[0157] The third approach is to re-trigger the joint optimization solution of step S02, and re-solve the multi-objective optimization function under the new state conditions to generate a brand-new cooperative control strategy to replace the current strategy. This approach is suitable for scenarios where cooperative mismatch is severe and the first two parameter adjustments are insufficient to restore stability.
[0158] The fourth method is to switch to a backup control mode, such as switching from a precise master-slave position following mode to a speed following mode, or switching to a preset safe recovery motion mode, which is suitable for emergency handling in extreme working conditions. The system automatically selects one or more of the above adjustment methods in combination and executes them according to the degree of coordination mismatch, until the deviation converges to within the threshold.
[0159] After completing the above assessment and necessary online adjustments, the system determines whether the job termination conditions are currently met. Job termination conditions include three categories:
[0160] The first category is normal completion. When the progress of the task reaches the preset target threshold, for example, when the end effector of the robotic arm has reached the target pose and completed the predetermined sampling or operation action, the task is judged to be successfully completed, and all task execution results (including the status log and control command record of the whole process) are archived.
[0161] The second type is stable convergence. When the degree of difference in state characteristics is lower than the preset stability threshold for multiple consecutive iterations (e.g., 10 consecutive control cycles), it indicates that the system has entered a stable state and no further work is required. It can be regarded as the natural termination of the work.
[0162] The third category is abnormal termination. When the system detects an unrecoverable anomaly (such as failure of a critical sensor, a sudden drop in hydraulic system pressure, or a complete communication interruption), and the basic work execution conditions still cannot be met after anomaly handling, the work is determined to be abnormally terminated. The fault log and all data before the anomaly occurred are archived for subsequent analysis. As long as any of the above termination conditions are not met, the system returns to step S02 to re-execute collaborative decision-making and optimization control until the work is successfully completed or abnormally terminated.
[0163] Example 2
[0164] Based on Example 1, the joint optimization mechanism in step S02 of this example takes the carrier attitude stability and master-slave following accuracy as the joint optimization objectives. It uses an iterative optimization algorithm to jointly solve the ROV carrier state characteristics and the robotic arm joint state characteristics to dynamically generate a cooperative control strategy.
[0165] The implementation of the joint optimization mechanism includes the following steps:
[0166] With the carrier attitude stability index and master-slave follow accuracy index as optimization objectives, a multi-objective optimization function is constructed using a linear weighting method.
[0167] Based on the coupled dynamics model of the carrier and the robotic arm, the attitude response of the ROV carrier and the joint response of the robotic arm are predicted under the current candidate control input.
[0168] Based on the master-slave error model, the degree of deviation of the predicted response from the expected response is evaluated;
[0169] A collaborative control strategy that minimizes the multi-objective optimization function is found through an iterative optimization algorithm.
[0170] Specifically, since vehicle attitude stability and master-slave tracking accuracy belong to two different evaluation dimensions, and they often compete with each other. For example, excessively pursuing high-speed tracking response of the robotic arm may lead to greater reaction impact on the vehicle, while excessively pursuing vehicle attitude stability requires the robotic arm to decelerate or pause its movements, thus sacrificing tracking accuracy. Therefore, both objectives need to be included in the optimization function for comprehensive trade-off. Specifically, a linear weighted method is used to construct a multi-objective optimization function: the vehicle attitude stability index and the master-slave tracking accuracy index are normalized separately to eliminate the differences in their dimensions and numerical scales. After assigning corresponding weight coefficients to each objective, a unified multi-objective optimization function is formed by weighted summation. The selection of weight coefficients is determined based on the task priority of the current operation scenario. In scenarios with extremely high requirements for end-point positioning accuracy, such as fine docking and precision assembly, the weight of master-slave tracking accuracy can be appropriately increased; when the ROV is maneuvering over a wide range or subjected to strong environmental disturbances, the weight of vehicle attitude stability is increased accordingly.
[0171] Before solving the optimization problem, a dynamic model describing the coupling relationship between the ROV and the robotic arm is first established. This model combines the six-degree-of-freedom motion equations of the ROV with the multi-joint dynamic equations of the robotic arm through force and torque transmission terms at the robotic arm base. Given the current system state (i.e., the ROV state characteristics and robotic arm joint state characteristics output from step S01) and a set of candidate control inputs (including the thrust command of the ROV thruster and the control voltage of the hydraulic servo valves of each joint of the robotic arm), forward numerical integration is performed using this coupled dynamic model to predict the attitude response trajectory of the ROV and the motion response trajectory of each joint of the robotic arm within a relatively short time window in the future. During the prediction process, the dynamic response delay of the hydraulic system and the thrust response characteristics of the thruster are considered simultaneously to improve the accuracy of the prediction.
[0172] Master-slave follower error refers to the deviation between the actual response pose of each joint of the robotic arm and the pose expected by the master hand's operating command. This step is based on a preset master-slave follower error model, which includes multiple dimensions such as joint space position error, velocity error, and the comprehensive pose error of the end effector in the workspace. The predicted robotic arm joint response is compared with the master hand's expected command at the current moment, the quantified value of each error dimension is calculated, and a single master-slave follower error index is formed according to a preset weight relationship. This index, together with the carrier posture stability index in the first step, constitutes the two basic terms of the multi-objective optimization function.
[0173] After establishing the objective function and enabling response prediction, the algorithm searches for the optimal solution that minimizes the multi-objective optimization function in a multi-dimensional decision space using the current system state as the initial condition and the control input as the decision variable, through numerical optimization algorithms. Since the objective function has nonlinear and nonconvex characteristics, iterative algorithms suitable for handling complex nonlinear optimization problems, such as sequential quadratic programming or genetic algorithms, can be used. In each optimization iteration, the algorithm generates a set of candidate control inputs, predicts the system response through a second stage, evaluates the error index through a third stage, calculates the objective function value, and adjusts the search direction and step size for the next iteration based on this value. After multiple iterations until the convergence condition is met, the algorithm outputs a cooperative control strategy that comprehensively optimizes the vehicle's attitude stability and master-slave tracking accuracy—a set of optimal control input commands—which are then sent to the ROV thruster control system and the robotic arm joint servo control system for execution. As the system operates continuously, step S02 is retried in each control cycle, achieving dynamic automatic processing of time-varying operating conditions.
[0174] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A collaborative control method for a master-slave hydraulic robotic arm in a deep-sea ROV, characterized in that, The method includes the following steps: S01. Collect multi-source sensor data and extract ROV carrier state features and robotic arm joint state features; S02. Construct a collaborative control decision model. The collaborative control decision model is based on the joint optimization mechanism of the carrier-manipulator coupling dynamics model and the master-slave following error model. The collaborative control strategy is dynamically generated according to the state characteristics of the ROV carrier and the joint state characteristics of the manipulator. S03. Based on the aforementioned collaborative control strategy, output the carrier compensation control quantity and the joint drive control quantity respectively. The carrier compensation control quantity is used to offset the impact load effect of the robotic arm's movement on the ROV carrier, and the joint drive control quantity is used to drive the slave joint to follow the master hand's command. S04. Generate a coordinated execution instruction sequence based on the carrier compensation control quantity and the drive control quantity of each joint. The coordinated execution instruction sequence includes the time synchronization execution logic of the carrier thruster instruction and the hydraulic servo valve instruction of each joint. S05. Collect execution feedback data, update the ROV carrier status characteristics and robotic arm joint status characteristics, and judge the current collaborative control effect through the collaborative status evaluation mechanism. When it is determined that the carrier posture deviation or joint tracking deviation exceeds the preset threshold, online adjustment of the collaborative control strategy is triggered. If the operation termination condition is not met, return to S02 for iterative execution. If it is met, archive the operation results.
2. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The multi-source sensor data includes carrier attitude data, carrier position data, joint angle data, joint angular velocity data, and joint hydraulic cylinder pressure data. The ROV carrier state characteristics include the carrier attitude quaternion, the three-axis position vector and the three-axis velocity vector in the inertial coordinate system; The joint state characteristics of the robotic arm include the angular position, angular velocity, and pressure difference between the two chambers of the hydraulic cylinder for each joint. The construction of the ROV carrier state characteristics in step S01 includes the following steps: The multi-source sensor data is filtered and standardized to define the carrier attitude feature dimension and the carrier position feature dimension. The carrier attitude feature dimension includes roll angle, pitch angle, yaw angle and their rate of change. The carrier position feature dimension includes the three-axis position coordinates of the ROV in the inertial coordinate system and their rate of change. Based on the influence weights of each feature dimension on collaborative control, a unified ROV carrier state feature vector is generated through weighted integration.
3. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 2, characterized in that, The construction of the robotic arm joint state features in step S01 includes the following steps: Define joint motion feature dimension and joint drive feature dimension. The joint motion feature dimension includes the angular position, angular velocity and angular acceleration of each joint. The joint drive feature dimension includes the pressure value and pressure difference of the two chambers of the hydraulic cylinder of each joint. Based on the forward kinematics model of the robotic arm, the angular positions of each joint are mapped to the pose and velocity state of the end effector. By integrating the motion state, drive state, and end effector state of each joint, a joint state feature vector of the robotic arm is generated.
4. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The carrier-manipulator coupled dynamics model includes a multi-rigid-body dynamics sub-model of the manipulator, a six-degree-of-freedom dynamics sub-model of the ROV carrier, and a force / torque mapping sub-model of the manipulator's action on the carrier, which is used to describe the reaction force and reaction torque generated by the dynamic motion of the manipulator on the ROV carrier. The construction of the carrier-robotic arm coupled dynamics model includes the following steps: A multi-rigid-body dynamics model of the robotic arm is established using the Lagrange method to describe the dynamic relationship between the driving torque of each joint and the joint angle, angular velocity and angular acceleration. Based on the pre-set ROV carrier six-degree-of-freedom dynamic model, the dynamic relationship between the carrier thruster output and the carrier's six-degree-of-freedom motion is described, and the hydrodynamic added mass coefficient and damping coefficient matrix are introduced. Based on the angular velocity, angular acceleration and configuration parameters of each joint of the robotic arm, a force / torque mapping model of the robotic arm on the carrier is established; The three sub-models are solved together to form a coupled dynamic model of the carrier-robotic arm.
5. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The master-slave follow error model includes a joint space follow error mapping sub-model and a task space pose error mapping sub-model, which are used to describe the deviation characteristics between the master hand command and the slave hand actual response. The construction of the master-slave follower error model includes the following steps: Acquire the angle command signals of each joint of the master hand and record the timestamp of the command signal; Based on the actual angle feedback from each joint of the hand, the joint space following error between the master hand command and the slave hand actual response is calculated. Based on the differential kinematics model of the robotic arm, the joint space following error is mapped to the position and posture errors in the task space; By introducing a hydraulic long pipeline transmission delay characteristic model and a hydraulic valve control cylinder nonlinear characteristic model, the joint space following error is compensated for by delay and nonlinear correction, and the compensated master-slave following error is obtained.
6. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The joint optimization mechanism in step S02 takes the carrier attitude stability and master-slave following accuracy as the joint optimization objectives. It uses an iterative optimization algorithm to jointly solve the ROV carrier state characteristics and the robotic arm joint state characteristics to dynamically generate a cooperative control strategy. The joint optimization mechanism includes the following steps: With the carrier attitude stability index and master-slave follow accuracy index as optimization objectives, a multi-objective optimization function is constructed using a linear weighting method. Based on the carrier-manipulator coupled dynamics model, the attitude response of the ROV carrier and the joint response of the manipulator are predicted under the current candidate control input. Based on the master-slave error model, the degree of deviation of the predicted response from the expected response is evaluated; A cooperative control strategy that minimizes the multi-objective optimization function is obtained by using an iterative optimization algorithm.
7. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The carrier compensation control quantity in step S03 includes vertical thrust compensation quantity, horizontal longitudinal thrust compensation quantity, horizontal lateral thrust compensation quantity, and yaw moment compensation quantity. The generation of the carrier compensation control quantity includes the following steps: Based on the carrier attitude compensation component in the aforementioned cooperative control strategy, calculate the required compensation thrust vector in each degree of freedom direction in the inertial coordinate system. Based on the thruster spatial arrangement configuration of the ROV, a thruster configuration matrix is constructed, and the compensation thrust vector is allocated to each thruster using the pseudo-inverse method to generate thrust commands for each thruster.
8. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 7, characterized in that, The joint drive control quantities in step S03 include the opening command and opening change rate command of the hydraulic servo valve of each joint. The generation of the drive control quantities for each joint includes the following steps: Based on the joint drive components in the aforementioned collaborative control strategy, the desired angular position and angular velocity of each joint are calculated. Based on the flow continuity equation and force balance equation of each joint valve-controlled symmetrical hydraulic cylinder, the desired angular position and angular velocity are converted into control current signals of each joint hydraulic servo valve. The control current signal is feedforward corrected by introducing hydraulic system pressure fluctuation compensation coefficient and long pipeline transmission delay compensation coefficient.
9. The method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The time synchronization execution in step S04 includes the following steps: Based on the difference in response characteristics between the carrier thruster and the hydraulic servo valve, a timestamp alignment mechanism for instructions and an execution timing offset configuration mechanism are established to ensure that the carrier thruster and the hydraulic servo valves of each joint operate in coordination within the same control cycle. The generation of the cooperative execution instruction sequence includes the following steps: Based on the difference in step response time and bandwidth between the carrier thruster and the hydraulic servo valve, the execution sequence and time offset of each instruction are determined. A timestamp alignment mechanism for instructions is constructed based on a clock reference of the global control cycle. An embedded actuator abnormal state detection logic is used to activate a safety protection strategy when the response feedback of any thruster or hydraulic servo valve exceeds the preset safety range.
10. A method for collaborative control of a master-slave hydraulic robotic arm in a deep-sea ROV according to claim 1, characterized in that, The collaborative status assessment mechanism in step S05 includes: The carrier attitude deviation threshold and the tracking deviation threshold of each joint are preset. The effect of cooperative control is judged by quantifying the difference between the updated state characteristics and the expected state characteristics. If the carrier attitude deviation is lower than the carrier attitude deviation threshold and the tracking deviation of each joint is lower than the tracking deviation threshold of each joint, then the current cooperative control strategy is maintained. If the carrier attitude deviation is not lower than the carrier attitude deviation threshold or the tracking deviation of any joint is not lower than the corresponding tracking deviation threshold of each joint, it is determined to be a coordination mismatch, triggering online adjustment of the coordination control strategy; The online adjustment of the collaborative control strategy includes at least one of the following methods: Adjust the coupling compensation parameters in the carrier-robotic arm coupled dynamics model; Adjust the time delay compensation parameters and nonlinear correction parameters in the master-slave follower error model; The multi-objective optimization function is re-solved to generate an updated cooperative control strategy; Switch to standby control mode; The operation termination conditions include at least one of the following: The progress of the task completion target has reached the preset target threshold; The degree of difference in state characteristics across multiple consecutive iterations is lower than a preset stability threshold; An unrecoverable exception is triggered, and the job execution conditions cannot be met even after exception handling.
