Multi-underwater robot cooperative control system and method based on predictive control

Abstract

The invention discloses a multi-underwater-robot cooperative control system and method based on predictive control, and relates to the technical field of cooperative control. During operation of the system, underwater environment data are collected through sensor equipment, an underwater three-dimensional dynamic environment model is generated by using a data fusion technology, and the underwater three-dimensional dynamic environment model is established based on task requirements; tasks are decomposed and distributed to all the robots, through an optimization algorithm, optimal paths of the robots and the mechanical arms are planned, the motion trails of the underwater robot mechanical arms are dynamically optimized based on model predictive control MPC, a real-time control strategy is provided for task division among the multiple mechanical arms, and the real-time control strategy of the underwater robot mechanical arms is provided based on environment modeling and the real-time state. Future motion states of the robot and the mechanical arm of the robot are predicted, an optimization objective function including task completion time, energy consumption and mechanical arm operation precision is constructed, a rolling optimization method is utilized, a control instruction is updated in real time, and real-time data sharing among multiple underwater robots is supported based on an acoustic communication technology.

Description

Technical Field

[0001] The present invention relates to the technical field of collaborative control, and in particular to a multi-underwater robot collaborative control system and method based on predictive control. Background Art

[0002] Traditional single robot systems often have performance limitations when facing complex underwater environments, such as inaccurate path planning, uneven task distribution, and excessive energy consumption. Therefore, operational robots with collaborative optimization of robotic arms came into being. By equipping each robot with a robotic arm, these operational robots can perform more complex tasks, such as equipment repair and sampling. At the same time, collaborative control between multiple robotic arms can improve operational efficiency and avoid conflicts and interference. However, although the collaborative optimization of robotic arms has improved the ability to operate, during the execution of tasks, changes in environmental factors, the efficiency of collaboration between robots, and the complexity of tasks are still key factors affecting the overall performance of the system.

[0003] However, in the existing technology, although the multi-underwater robot collaborative system has been able to perform some basic tasks, there are still many shortcomings. First, the traditional multi-robot system cannot quickly optimize the path and task according to environmental changes, resulting in low efficiency. Secondly, there is still an efficiency bottleneck in the collaboration between the robotic arms, especially when multiple robots are performing complex tasks at the same time, improper coordination will cause resource conflicts or energy waste. Finally, the existing system has limited adaptability to environmental disturbances (such as water flow, turbulence, etc.), and often requires manual intervention for correction. Therefore, a more intelligent system is needed to achieve more efficient and precise collaborative control to maximize task execution efficiency and system stability.

[0004] The shortcomings of existing multi-underwater robot systems and robotic arm collaborative control systems stem from the limitations of traditional path planning and task allocation methods. Without intelligent control, tasks usually need to be executed according to preset routes and steps, which is particularly rigid in dynamic environments. For example, in complex water flow environments, robots may face problems such as inaccurate paths or excessive energy consumption, which not only affects the execution speed of tasks, but may also cause equipment failure or task interruption. In addition, due to the lack of intelligent task scheduling and load distribution, there is often an uneven load between robotic arms, and some robotic arms may suffer performance degradation or even failure due to excessive task volume. The system lacks prediction and dynamic adjustment of real-time status, which greatly reduces the stability and efficiency of collaborative operations. Once these problems occur, they may cause task delays, waste of energy resources, and even affect the final completion of the task. The multi-underwater robot collaborative control system based on predictive control can effectively solve these problems through real-time prediction and dynamic optimization, provide more flexible and efficient solutions, and ensure the stability of multi-robot systems in complex underwater environments and the efficient completion of tasks. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a multi-underwater robot collaborative control system and method based on predictive control, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-underwater robot collaborative control system based on predictive control, including an environment perception and modeling module, a path planning module, a robotic arm collaborative control module, a predictive control module, a collaborative communication module and a task execution module; The environmental perception and modeling module is used to collect underwater environmental data through sensor equipment and generate an underwater three-dimensional dynamic environmental model using data fusion technology, including obstacle distribution, water flow speed, direction change and operation point location; The path planning module is used to decompose and assign tasks to individual robots based on task requirements, including target area coverage, specific point sampling, and equipment maintenance complex tasks, and plan the optimal path of the robot and manipulator through an optimization algorithm, taking into account the priority of the task, the robot status, and the distribution of resources; The robotic arm collaborative control module dynamically optimizes the motion trajectory of the underwater robot robotic arm based on model predictive control MPC, provides real-time control strategies for task division between multiple robotic arms, and prevents collisions between arms; The predictive control module is used to predict the future motion state of the robot and its mechanical arm based on environmental modeling and real-time status, construct an optimization objective function, including task completion time, energy consumption, and mechanical arm operation accuracy, and use a rolling optimization method to update control instructions in real time; The collaborative communication module is used to support real-time data sharing among multiple underwater robots based on acoustic communication technology, and reduce communication delay and data loss through data compression and efficient protocol design; The task execution module is used to receive the control instructions generated by the upper module, complete the robot movement and manipulator operation tasks, monitor the task completion status and equipment status in real time, and pass the feedback data to the prediction control module.

[0007] Preferably, the environment perception and modeling module includes a data acquisition unit, a data preprocessing unit and a dynamic environment modeling unit; The data acquisition unit is used to collect the original data of the underwater environment in real time through multiple types of sensors, including multi-beam sonar, underwater laser scanner and multi-frequency water flow meter, to obtain environmental information related to obstacle distribution, water flow velocity, turbulence intensity and target position, and obtain: real-time water flow velocity V f , the optimal working speed V of the robot thruster t and environmental sensitivity factor α e , forming an environmental adaptation data set; According to the number of task targets and the difficulty score of the area, the complexity of the current task is obtained, and the distribution rationality and distribution balance of the task targets are described, including the number of targets, the discreteness of the target position distribution and the priority, and the task weight W assigned to the i-th robot is obtained. i , the average distance between the target point and the robot D avg and the number of robots n, forming a task distribution dataset; The current status of each robot, including position, energy margin, and thruster load, is obtained in real time through internal sensors and communication devices to obtain the robot's current task completion rate η r , Energy consumption of a single robot E r , total energy budget of the task E t and resource sensitivity factor β r .

[0008] Preferably, the data preprocessing unit is used to denoise, normalize and remove redundant information from the collected raw data, improve data quality through a distributed processing algorithm, and provide data input for a subsequent dynamic environment modeling unit; The dynamic environment modeling unit is used to generate a real-time updated underwater three-dimensional dynamic environment model using Kalman filtering and particle filtering algorithms, including the spatiotemporal distribution information of environmental characteristics, calculate the environmental adaptation coefficient Ce, task distribution balance coefficient Ct and resource utilization efficiency coefficient Cr, and combine them to calculate the system collaborative efficiency index Is.

[0009] Preferably, the environmental adaptability coefficient Ce is calculated by the following formula: ; Where V f Indicates the real-time water flow velocity, V t represents the optimal working speed of the robot thruster, α e represents the environmental sensitivity factor, γ e represents the environmental disturbance correction coefficient; The task distribution balance coefficient Ct is calculated by the following formula: ; Where W i represents the task weight assigned to the i-th robot, D avg represents the average distance between the target point and the robot, and n represents the number of robots; The resource utilization efficiency coefficient Cr is calculated by the following formula: ; Where η r represents the current task completion rate of the robot, Er represents the energy consumption of a single robot, Et represents the total energy consumption budget of the task, β r represents the resource sensitivity factor; The system synergy efficiency index Is is calculated by the following formula: ; Where ω1, ω2 and ω3 represent the weights of the coefficients, respectively, satisfying ω1+ω2+ω3=1, Ce, Ct and Cr represent the environmental adaptation coefficient Ce, the task distribution balance coefficient Ct and the resource utilization efficiency coefficient Cr.

[0010] Preferably, the path planning module includes a task decomposition unit and a multi-robot path generation unit; The task decomposition unit is used to decompose a complex task into multiple subtasks based on the global task goal using a priority evaluation algorithm, and reasonably allocate them in combination with robot performance and task difficulty; The multi-robot path generation unit is used to combine obstacle distribution and water flow characteristics, calculate the optimal path for each robot through the Dijkstra algorithm, avoid path intersections and resource conflicts, and dynamically optimize the path planning scheme according to the remaining energy, task path length and thruster efficiency of each robot.

[0011] Preferably, the robotic arm collaborative control module includes a single-arm dynamic optimization unit and a multi-arm collaborative scheduling unit; The single-arm dynamic optimization unit constructs an initial trajectory planning model of the robotic arm based on the target point position, task requirements, and obstacle distribution, uses the MPC algorithm to predict the motion state of the robotic arm within multiple time steps, optimizes the trajectory in real time to avoid motion deviation, updates the dynamic trajectory planning parameters using the real-time data collected by the sensor, and in the grasping or operation task, senses whether the target grasping force is appropriate through the force feedback of the end effector and adjusts the motion speed and direction; when the task is completed or fails, the state is fed back to the predictive control module; The multi-arm collaborative scheduling unit is used to coordinate the motion of each robotic arm through dynamic task division and path planning, prevent path conflicts, and at the same time optimize the distribution of the workload to improve the collaborative operation efficiency.

[0012] Preferably, the predictive control module includes a state prediction unit and an instruction generation unit; The state prediction unit is used to construct a dynamic prediction model of multiple underwater robots and their robotic arms, predict the change trend of the system collaborative efficiency index Is in the future multiple time steps, evaluate the current system state, and determine the current system state level through comparison with the high-efficiency qualified threshold A and the lowest qualified threshold B, and define three levels of comparison and evaluation strategies: high-efficiency level, critical qualified level, and unqualified level: High-efficiency state: Is≥A, obtain the high-efficiency level evaluation result, continue to execute the existing task allocation scheme and path planning, without changing the overall structure of the task, maintain the allocation ratio and task objectives of all robots, regularly check the operation accuracy of each robot and robotic arm according to the current system state, in the high-efficiency state, pay special attention to the key nodes in the task execution, including the target sampling position and the key operation points, during the task completion process, make small adjustments according to the sensor feedback, appropriately reduce the motion speed of the robotic arm by 20%, avoid the tiny position error caused by the water flow change in the dynamic environment, regularly check the remaining power of each robot, and estimate the energy consumption according to the task progress. Once the monitored power is lower than the set threshold of 20%, issue a low-power warning and plan the nearest charging point or supplementary task for the robot with the即将耗尽电量的机器人规划最近的充电点或补充任务; Qualified state: B≤Is<A, obtain the critical qualified level evaluation result, locally optimize the interaction area or obstacle-dense area between the robots and robotic arms in the current task, optimize the path distance, generate a shorter return path for each robot's robotic arm, assign high-priority tasks to the robots with more remaining power and lower energy consumption, and in the area with strong water flow, set a path for the robot to avoid the water flow area in advance to reduce the impact of the water flow interference on the task on the path; Inefficient state: Is < B, obtain the unqualified level evaluation result. The system needs to quickly adjust the division of labor of the robotic arm, re-optimize the task path, re-plan the path in areas involving more complex or stronger water flow in the task, prioritize avoiding areas with large water flow disturbances, ensure that the task path is shortened as much as possible, reduce unnecessary path interference, re-decompose the unfinished tasks. If some robots are overloaded, immediately reassign some tasks with lighter loads to robots with more sufficient resources to avoid task incompletion or delay; The instruction generation unit is used to dynamically optimize the system operation strategy according to the level evaluation result, and regenerate the control instructions for multiple robots and robotic arms through a rolling optimization method.

[0013] Preferably, the collaborative communication module includes a data exchange unit and a clock synchronization unit; The data exchange unit is used for real-time exchange of key data among multiple underwater robots, including robot position, task progress, environmental dynamic information, and energy status. The data content includes real-time position information, velocity vector, task status identifier, energy consumption data, obstacle position, and water flow intensity. The data is packaged in layers, divided into high priority and normal priority; Acoustic communication combined with Turbo code or LDPC code technology is used to reduce the bit error rate and adapt to the high-noise underwater environment. In high-traffic or weak-signal situations, the data exchange unit uses dynamic compression technology to reduce bandwidth requirements, dynamically adjusts the communication frequency based on task requirements, transmits high-priority data in real-time, and delays the transmission of low-priority data; The clock synchronization unit is used to adopt an adaptive time synchronization protocol to achieve clock calibration among robots through timestamp messages broadcast over the network. In multi-robot communication, by collecting data packets of multiple timestamp paths, the minimum offset method is used to select the best path for time calibration. Considering the characteristics of multi-path propagation underwater, a robust calibration algorithm is designed to eliminate time differences caused by different paths, and the time offset among multiple robots is monitored in real-time, and the cumulative error is calculated and gradually corrected.

[0014] Preferably, the task execution module includes an execution instruction parsing unit and a feedback adjustment unit; The execution instruction parsing unit is used to parse the control instructions generated by the predictive control module, convert them into specific execution actions of the robot thrusters and robotic arms, and monitor the task execution status in real-time, including the robot motion trajectory, robotic arm operation accuracy, and task completion progress, and record the execution deviation; The feedback adjustment unit is used to upload key feedback data generated during task execution, including path deviation and energy consumption overrun, to the predictive control module.

[0015] A multi-underwater-robot collaborative control method based on predictive control includes the following steps: Step 1: Collect underwater environmental data through sensor equipment and use data fusion technology to generate an underwater three-dimensional dynamic environment model, including obstacle distribution, water flow speed, direction change, and operation point location; Step 2: Based on the task requirements, including target area coverage, specific point sampling, and complex equipment maintenance tasks, the tasks are decomposed and assigned to each robot. Through the optimization algorithm, the optimal path of the robot and the manipulator is planned, taking into account the priority of the task, the robot status, and the distribution of resources; Step 3: Based on model predictive control (MPC), the motion trajectory of the underwater robot's manipulator is dynamically optimized to provide a real-time control strategy for the division of tasks between multiple manipulators and prevent collisions between arms. Step 4: Based on the environment modeling and real-time status, predict the future motion state of the robot and its manipulator, build the optimization objective function, including task completion time, energy consumption and manipulator operation accuracy, and use the rolling optimization method to update the control instructions in real time; Step 5: Based on acoustic communication technology, support real-time data sharing among multiple underwater robots, and reduce communication delays and data loss through data compression and efficient protocol design; Step 6: Receive control instructions generated by the upper-level module, complete robot movement and manipulator operation tasks, monitor task completion and equipment status in real time, and pass feedback data to the predictive control module.

[0016] The present invention provides a multi-underwater robot collaborative control system and method based on predictive control, which has the following beneficial effects: (1) When the system is running, it collects underwater environmental data through sensor equipment and generates an underwater three-dimensional dynamic environment model using data fusion technology. Based on task requirements, the system decomposes and assigns tasks to each robot. Through optimization algorithms, it plans the optimal path of the robot and the manipulator. Based on model predictive control (MPC), it dynamically optimizes the motion trajectory of the underwater robot's manipulator arm and provides real-time control strategies for the division of tasks among multiple manipulator arms. Based on environmental modeling and real-time status, it predicts the future motion state of the robot and its manipulator arm and constructs an optimization objective function, including task completion time, energy consumption, and manipulator arm operation accuracy. Using the rolling optimization method, it updates control instructions in real time. Based on acoustic communication technology, it supports real-time data sharing among multiple underwater robots.

[0017] (2) The multi-underwater robot collaborative control system based on predictive control successfully realizes the efficient scheduling and execution of multi-robot collaborative tasks in complex underwater environments by integrating six core modules: environment perception and modeling, path planning, robot collaborative control, predictive control, collaborative communication, and task execution. The environment perception and modeling module provides accurate underwater environmental data, so that path planning and task allocation can be adjusted in real time according to factors such as water flow and obstacles, ensuring that the collaborative operation of multiple robots is more accurate. The robot collaborative control module improves the operating efficiency of the robot through dynamic optimization and anti-collision mechanisms, avoiding interference between robot arms. At the same time, the predictive control module realizes dynamic adjustment based on real-time data, ensuring energy optimization and path planning during task execution, greatly improving the overall efficiency of the system.

[0018] (3) Compared with traditional single-robot or simple collaborative control systems, this system has made significant technical improvements in many aspects. In traditional systems, robots often lack sufficient coordination and intelligent scheduling, and path planning and task allocation cannot adapt to real-time changes in complex environments, resulting in low efficiency and energy waste. The multi-underwater robot collaborative control system based on predictive control adopts real-time data prediction and rolling optimization methods, which enables the system to adjust the path and task allocation in real time according to the environment and task requirements, avoiding resource conflicts and energy waste. In addition, by introducing robotic arm collaborative control and environmental adaptability optimization algorithms, this system can ensure the accuracy of the robotic arm in a dynamic environment and achieve an efficient balance in task allocation, greatly improving the collaborative ability and stability of the multi-robot system.

[0019] (4) The implementation of this system not only optimizes existing technologies and improves the performance of multiple underwater robots in complex tasks, but also greatly improves the efficiency and accuracy of task completion. Through highly integrated collaborative control and prediction mechanisms, the system can quickly respond to changes in complex underwater environments and adjust strategies in real time, thereby improving the timeliness and success rate of tasks. At the same time, energy consumption has been optimized, and the working time and efficiency of the robot system have been significantly improved. Compared with current technical means, this system can significantly reduce manual intervention during task execution, has a higher level of automation, and has enhanced real-time performance and reliability, ultimately achieving an improvement in task success rate and execution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a block diagram of a multi-underwater robot collaborative control system based on predictive control in the present invention; Figure 2 The present invention is a schematic diagram of the steps of a method for cooperative control of multiple underwater robots based on predictive control. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Example 1 The present invention provides a multi-underwater robot cooperative control system based on predictive control, please refer to Figure 1 , including environment perception and modeling module, path planning module, robotic arm collaborative control module, predictive control module, collaborative communication module and task execution module; The environmental perception and modeling module is used to collect underwater environmental data through sensor equipment and generate an underwater three-dimensional dynamic environmental model using data fusion technology, including obstacle distribution, water flow speed, direction change and operation point location; The path planning module is used to decompose and assign tasks to individual robots based on task requirements, including target area coverage, specific point sampling, and equipment maintenance complex tasks, and plan the optimal path of the robot and manipulator through an optimization algorithm, taking into account the priority of the task, the robot status, and the distribution of resources; The robotic arm collaborative control module dynamically optimizes the motion trajectory of the underwater robot robotic arm based on model predictive control MPC, provides real-time control strategies for task division between multiple robotic arms, and prevents collisions between arms; The predictive control module is used to predict the future motion state of the robot and its mechanical arm based on environmental modeling and real-time status, construct an optimization objective function, including task completion time, energy consumption, and mechanical arm operation accuracy, and use a rolling optimization method to update control instructions in real time; The collaborative communication module is used to support real-time data sharing among multiple underwater robots based on acoustic communication technology, and reduce communication delay and data loss through data compression and efficient protocol design; The task execution module is used to receive the control instructions generated by the upper module, complete the robot movement and manipulator operation tasks, monitor the task completion status and equipment status in real time, and pass the feedback data to the prediction control module.

[0023] In this embodiment, underwater environmental data is collected by sensor equipment, and a three-dimensional underwater dynamic environment model is generated by using data fusion technology, including obstacle distribution, water flow speed, direction change and operation point location. Based on task requirements, including target area coverage, specific point sampling and equipment maintenance complex tasks, the tasks are decomposed and assigned to each robot. The optimal path of the robot and the manipulator is planned by an optimization algorithm, considering the priority of the task, the robot state and the distribution of resources. Based on the model predictive control MPC, the motion trajectory of the underwater robot manipulator is dynamically optimized, and a real-time control strategy is provided for the task division between multiple manipulators to prevent collisions between arms. Based on environmental modeling and real-time status, the future motion state of the robot and its manipulator is predicted, and an optimization objective function is constructed, including task completion time, energy consumption and manipulator operation accuracy. The rolling optimization method is used to update the control instructions in real time. Based on acoustic communication technology, real-time data sharing between multiple underwater robots is supported. Through data compression and efficient protocol design, communication delay and data loss are reduced. The control instructions generated by the upper module are received, the robot movement and manipulator operation tasks are completed, the task completion status and equipment status are monitored in real time, and the feedback data is passed to the predictive control module.

[0024] Example 2 This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the environment perception and modeling module includes a data acquisition unit, a data preprocessing unit and a dynamic environment modeling unit; The data acquisition unit is used to collect the original data of the underwater environment in real time through multiple types of sensors, including multi-beam sonar, underwater laser scanner and multi-frequency water flow meter, to obtain environmental information related to obstacle distribution, water flow velocity, turbulence intensity and target position, and obtain: real-time water flow velocity V f , the optimal working speed V of the robot thruster t and environmental sensitivity factor α e , forming an environmental adaptation data set; According to the number of task targets and the difficulty score of the area, the complexity of the current task is obtained, and the distribution rationality and distribution balance of the task targets are described, including the number of targets, the discreteness of the target position distribution and the priority, and the task weight W assigned to the i-th robot is obtained. i , the average distance between the target point and the robot D avg and the number of robots n, forming a task distribution dataset; The current status of each robot, including position, energy margin, and thruster load, is obtained in real time through internal sensors and communication devices to obtain the robot's current task completion rate η r , Energy consumption of a single robot E r , total energy budget of the task E tand resource sensitivity factor β r .

[0025] The data preprocessing unit is used to denoise, normalize and remove redundant information from the collected raw data, improve data quality through a distributed processing algorithm, and provide data input for the subsequent dynamic environment modeling unit; The dynamic environment modeling unit is used to generate a real-time updated underwater three-dimensional dynamic environment model using Kalman filtering and particle filtering algorithms, including the spatiotemporal distribution information of environmental characteristics, calculate the environmental adaptation coefficient Ce, task distribution balance coefficient Ct and resource utilization efficiency coefficient Cr, and combine them to calculate the system collaborative efficiency index Is.

[0026] In this embodiment, the environmental perception and modeling module realizes accurate perception and real-time monitoring of complex underwater environments through efficient data acquisition, preprocessing and dynamic modeling, significantly improving the adaptability and coordination efficiency of the system. The data acquisition unit can comprehensively obtain the key characteristic data of the underwater environment, including water flow velocity, obstacle distribution and task target position, through real-time monitoring of multiple types of sensors, and provides the system with rich environmental adaptation data. These data provide a solid data foundation for task allocation, path planning and subsequent control decisions, ensuring the efficient collaboration of robots and manipulators in complex environments. The accuracy and quality of the original collected data are guaranteed by denoising, normalization and redundant information elimination of the data preprocessing unit, providing clear and reliable input data for the dynamic environment modeling unit. The dynamic environment modeling unit uses advanced algorithms such as Kalman filtering and particle filtering, which can not only update the underwater three-dimensional environment model in real time, but also accurately calculate the environmental adaptation coefficient, task distribution balance coefficient and resource utilization efficiency coefficient. The calculation of these coefficients provides a basis for subsequent collaborative control, thereby effectively improving the efficiency of task execution and the rational allocation of resources.

[0027] Example 3 This embodiment is explained in Example 1, please refer to Figure 1 Specifically: the environmental adaptability coefficient Ce is calculated by the following formula: ; Where V f Indicates the real-time water flow velocity, V t represents the optimal working speed of the robot thruster, α e represents the environmental sensitivity factor, γ e represents the environmental disturbance correction coefficient; The task distribution balance coefficient Ct is calculated by the following formula: ; Where W irepresents the task weight assigned to the i-th robot, D avg represents the average distance between the target point and the robot, and n represents the number of robots; The resource utilization efficiency coefficient Cr is calculated by the following formula: ; Where η r represents the current task completion rate of the robot, Er represents the energy consumption of a single robot, Et represents the total energy consumption budget of the task, β r represents the resource sensitivity factor; The system synergy efficiency index Is is calculated by the following formula: ; Where ω1, ω2 and ω3 represent the weights of the coefficients, respectively, satisfying ω1+ω2+ω3=1, Ce, Ct and Cr represent the environmental adaptation coefficient Ce, the task distribution balance coefficient Ct and the resource utilization efficiency coefficient Cr.

[0028] The path planning module includes a task decomposition unit and a multi-robot path generation unit; The task decomposition unit is used to decompose a complex task into multiple subtasks based on the global task goal and utilize a priority evaluation algorithm, and reasonably allocate the tasks in combination with the robot performance and task difficulty; The multi-robot path generation unit is used to combine obstacle distribution and water flow characteristics, calculate the optimal path for each robot through the Dijkstra algorithm, avoid path intersections and resource conflicts, and dynamically optimize the path planning scheme according to the remaining energy, task path length and thruster efficiency of each robot.

[0029] The robotic arm collaborative control module includes a single-arm dynamic optimization unit and a multi-arm collaborative scheduling unit; The single-arm dynamic optimization unit constructs an initial trajectory planning model for the robot arm based on the target point position, task requirements, and obstacle distribution, uses the MPC algorithm to predict the motion state of the robot arm in multiple time steps, optimizes the trajectory in real time, avoids motion deviation, and uses real-time data collected by sensors to update dynamic trajectory planning parameters. In grasping or operating tasks, the force feedback of the end effector is used to sense whether the target grasping force is appropriate, and adjusts the motion speed and direction; when the task is completed or failed, the state is fed back to the predictive control module; The multi-arm collaborative scheduling unit is used to coordinate the movement of each robotic arm through dynamic task division and path planning, prevent path conflicts, optimize workload distribution, and improve collaborative work efficiency.

[0030] In this embodiment, the system can realize accurate multi-underwater robot collaborative control by introducing the environmental adaptability coefficient Ce, the task distribution balance coefficient Ct and the resource utilization efficiency coefficient Cr, and effectively improve the stability of task execution and the rationality of resource allocation. The environmental adaptability coefficient Ce can dynamically evaluate and correct the adaptability of the robot under different environmental conditions by comparing the real-time water flow velocity with the working state of the robot propeller. This enables the robot to automatically adjust the operation mode to cope with the impact of environmental disturbances (such as water flow changes) on task execution, and improves the adaptability of the system in complex underwater environments. In task allocation, the task distribution balance coefficient Ct accurately measures the rationality and fairness of task allocation. By considering the distance between the target point and the robot and the task weight, it ensures the balance of the task load of each robot, thereby avoiding the efficiency reduction caused by excessive load of some robots. This not only effectively improves the efficiency of task allocation, but also ensures that each robot can complete the task in the optimal state. In addition, the resource utilization efficiency coefficient Cr realizes dynamic optimization of energy consumption by analyzing the relationship between the robot energy consumption and the total energy consumption of the task. Under the guidance of the system collaborative efficiency index Is, the robot can adjust the energy allocation and usage strategy according to the real-time feedback of task execution, ensuring that the system maximizes the use of each robot's resources without adding additional burden, thereby effectively improving the system's operating efficiency and the reliability of task completion.

[0031] Example 4 This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the prediction control module includes a state prediction unit and an instruction generation unit; The state prediction unit is used to construct a dynamic prediction model of multiple underwater robots and their manipulators, predict the change trend of the system collaborative efficiency index Is in multiple future time steps, and evaluate the current system state. By comparing with the efficient qualified threshold A and the minimum qualified threshold B, the current system state level is determined, and a comparison evaluation strategy of three levels is defined: efficient level, critical qualified level and unqualified level: Efficient state: Is≥A. Obtain the evaluation result of the efficient level, continue to execute the existing task allocation scheme and path planning, without changing the overall structure of the task, maintain the allocation ratio and task objectives of all robots, regularly check the operation accuracy of each robot and manipulator according to the current system state. In the efficient state, pay special attention to the key nodes in task execution, including the target sampling position and key operation points. During the task completion process, make small adjustments according to the sensor feedback, appropriately reduce the movement speed of the manipulator by 20% to avoid tiny position errors caused by the water flow change in the dynamic environment, regularly check the remaining power of each robot, and estimate the energy consumption according to the task progress. Once the monitored power is lower than the set threshold of 20%, issue a low power warning and plan the nearest charging point or supplementary task for the robot with the upcoming exhausted power; Qualified state: B≤Is<A. Obtain the critical qualified level evaluation result, locally optimize the interaction area between the robots and manipulators or the obstacle-dense area in the current task, optimize the path distance, generate a shorter return path for each robot manipulator, assign high-priority tasks to the robots with more remaining power and lower energy consumption. In the area with strong water flow, set a path for the robot to avoid the water flow area in advance to reduce the impact of water flow interference on the task on the path; Inefficient state: Is<B. Obtain the unqualified level evaluation result. The system needs to quickly adjust the manipulator division of labor, re-optimize the task path, re-plan the path in the area involving complex or strong water flow in the task, give priority to avoiding the area with large water flow disturbance, ensure that the task path is shortened as much as possible, reduce unnecessary path interference, re-decompose the unfinished task. If some robots are overloaded, immediately re-assign some tasks with lighter loads to the robots with more sufficient resources to avoid task incompletion or delay; The instruction generation unit is used to dynamically optimize the system operation strategy according to the level evaluation result, and regenerate the control instructions for multiple robots and manipulators through the rolling optimization method.

[0032] The collaborative communication module includes a data exchange unit and a clock synchronization unit; The data exchange unit is used for the real-time exchange of key data among multiple underwater robots, including robot position, task progress, environmental dynamic information and energy state. The data content includes real-time position information, velocity vector, task status identifier, energy consumption data, obstacle position and water flow intensity. The data is packaged in layers and divided into high priority and normal priority; Acoustic communication combined with Turbo code or LDPC code technology is used to reduce the error rate and adapt to the high-noise underwater environment. In the case of high traffic or weak signal, the data exchange unit uses dynamic compression technology to reduce the bandwidth requirement, dynamically adjusts the communication frequency based on the task requirements, and transmits high-priority data in real time and delays the transmission of low-priority data; The clock synchronization unit is used to adopt an adaptive time synchronization protocol to realize clock calibration between robots through timestamp messages broadcast on the network. In multi-robot communication, by collecting data packets of multiple timestamp paths, the minimum offset method is used to select the best path for time calibration. According to the characteristics of underwater multi-path propagation, a robust calibration algorithm is designed to eliminate the time difference caused by different paths, monitor the time offset between multiple robots in real time, calculate the accumulated error and make gradual corrections.

[0033] The task execution module includes an execution instruction parsing unit and a feedback adjustment unit; The execution instruction parsing unit is used to parse the control instructions generated by the predictive control module, convert them into specific execution actions of the robot propeller and the manipulator, monitor the task execution status in real time, including the robot motion trajectory, the manipulator operation accuracy and the task completion progress, and record the execution deviation; The feedback adjustment unit is used to upload key feedback data generated during task execution, including path deviation and energy consumption exceeding the limit, to the prediction control module.

[0034] In this embodiment, the system realizes efficient collaborative control of multiple underwater robots and their manipulators by introducing predictive control and collaborative communication modules, significantly improving the accuracy and stability of task execution. In the predictive control module, the state prediction unit can predict the change trend of the system collaborative efficiency index Is in real time by constructing a dynamic prediction model, and quickly determine the state level of the current system based on the comparison with the efficient qualified threshold A and the minimum qualified threshold B. Through this real-time evaluation, the system can make targeted optimization adjustments according to the current state to ensure that the system can achieve optimal control in different states. For example, in an efficient state, the system can avoid small errors caused by water disturbance by slightly adjusting the movement speed of the manipulator; in an inefficient state, it can timely re-plan the task path, optimize the robot load distribution, and avoid task delays or failures. This dynamic and intelligent task adjustment capability greatly improves the stability and task completion rate of the system. The collaborative communication module further enhances the collaborative ability of the system. The data exchange unit ensures the smooth and real-time flow of information by sharing the position, task progress, environmental information and energy status between multiple robots in real time. At the same time, the use of acoustic communication combined with Turbo code or LDPC code technology effectively reduces the bit error rate of underwater communication and enhances the communication reliability of the system in complex underwater environments. In the case of poor communication or weak signal, dynamic compression technology and frequency adjustment strategy ensure the real-time transmission of high-priority data, further optimizing resource utilization. The clock synchronization unit ensures the clock consistency between robots through an adaptive time synchronization protocol, eliminates the time deviation caused by signal delay and different paths, and ensures accurate coordination during task execution. The task execution module accurately converts the control instructions generated by the predictive control module into actual robot movement and manipulator operation through the execution instruction parsing and feedback adjustment unit. The real-time feedback mechanism of these control instructions ensures the accuracy and progress of task execution. At the same time, when deviations occur, the strategy can be adjusted quickly to reduce resource waste and ensure that the task is completed on time. The synergy of this series of functions enables the multi-underwater robot system to achieve accurate and efficient task execution in complex underwater environments, significantly improving the reliability and efficiency of task completion.

[0035] Example 5 A collaborative control method for multiple underwater robots based on predictive control, please refer to Figure 2 , specifically: including the following steps: Step 1: Collect underwater environmental data through sensor equipment and use data fusion technology to generate an underwater three-dimensional dynamic environment model, including obstacle distribution, water flow speed, direction change, and operation point location; Step 2: Based on the task requirements, including target area coverage, specific point sampling, and complex equipment maintenance tasks, the tasks are decomposed and assigned to each robot. Through the optimization algorithm, the optimal path of the robot and the manipulator is planned, taking into account the priority of the task, the robot status, and the distribution of resources; Step 3: Based on model predictive control (MPC), the motion trajectory of the underwater robot's manipulator is dynamically optimized to provide a real-time control strategy for the division of tasks between multiple manipulators and prevent collisions between arms. Step 4: Based on the environment modeling and real-time status, predict the future motion state of the robot and its manipulator, build the optimization objective function, including task completion time, energy consumption and manipulator operation accuracy, and use the rolling optimization method to update the control instructions in real time; Step 5: Based on acoustic communication technology, support real-time data sharing among multiple underwater robots, and reduce communication delays and data loss through data compression and efficient protocol design; Step 6: Receive control instructions generated by the upper-level module, complete robot movement and manipulator operation tasks, monitor task completion and equipment status in real time, and pass feedback data to the predictive control module.

[0036] In this embodiment, the system effectively improves the adaptability and work efficiency of the multi-underwater robot collaborative control system in a complex underwater environment through a series of closely coordinated steps. Step 1 accurately generates an underwater three-dimensional dynamic environment model through real-time data acquisition and data fusion technology of sensor equipment, which can fully understand the distribution of obstacles, water flow speed and operation point location, and provide a reliable environmental basis for subsequent task planning and execution. This data-driven modeling method greatly enhances the real-time perception capability of the system and ensures the efficient collaboration of multiple robots in a dynamic environment. The task decomposition and allocation module in step 2 uses an optimization algorithm to reasonably allocate tasks according to task requirements (such as area coverage, sampling, equipment maintenance, etc.), taking into account the distribution of robot status and resources, and avoiding resource waste and task execution conflicts. By optimizing path planning, robots and manipulators can optimally perform tasks, ensuring that the efficiency of task execution is maximized under limited resources. At the same time, the manipulator collaborative control module in step 3 effectively prevents collisions between manipulators through dynamic optimization based on model predictive control (MPC), improving the safety and accuracy during task execution. In step 4, based on the predictive control mechanism of environmental modeling and real-time status, the future motion state of the robot and the manipulator can be accurately predicted, and the control instructions can be adjusted in real time. This not only effectively improves the time efficiency of task completion, but also optimizes energy consumption control, ensuring the efficient operation of the system in complex tasks. Step 5 significantly improves the efficiency of collaborative communication between multi-robot systems through data sharing and optimization protocols based on acoustic communication technology. In a high-noise environment, communication delays and data loss are reduced through data compression and efficient protocol design, ensuring the stability and real-time performance of data transmission. Finally, step 6 enables the task execution module to accurately complete the robot motion and manipulator operation tasks through a real-time feedback mechanism, and transmits the key data in execution back to the predictive control module in real time, further optimizing the dynamic adjustment and task adjustment strategy of the system. Although the embodiments of the present invention have been shown and described, it is understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the attached claims and their equivalents.

Claims

1. A multi-underwater robot cooperative control system based on predictive control, characterized in that: It includes environment perception and modeling module, path planning module, robotic arm collaborative control module, predictive control module, collaborative communication module and task execution module; The environmental perception and modeling module is used to collect underwater environmental data through sensor equipment and generate an underwater three-dimensional dynamic environmental model using data fusion technology, including obstacle distribution, water flow speed, direction change and operation point location; The path planning module is used to decompose and assign tasks to individual robots based on task requirements, including target area coverage, specific point sampling, and equipment maintenance complex tasks, and plan the optimal path of the robot and manipulator through an optimization algorithm, taking into account the priority of the task, the robot status, and the distribution of resources; The robotic arm collaborative control module dynamically optimizes the motion trajectory of the underwater robot robotic arm based on model predictive control MPC, provides real-time control strategies for task division between multiple robotic arms, and prevents collisions between arms; The predictive control module is used to predict the future motion state of the robot and its mechanical arm based on environmental modeling and real-time status, construct an optimization objective function, including task completion time, energy consumption, and mechanical arm operation accuracy, and use a rolling optimization method to update control instructions in real time; The collaborative communication module is used to support real-time data sharing among multiple underwater robots based on acoustic communication technology, and reduce communication delay and data loss through data compression and efficient protocol design; The task execution module is used to receive the control instructions generated by the upper module, complete the robot movement and manipulator operation tasks, monitor the task completion status and equipment status in real time, and pass the feedback data to the prediction control module.

2. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The environment perception and modeling module includes a data acquisition unit, a data preprocessing unit and a dynamic environment modeling unit; The data acquisition unit is used to collect the original data of the underwater environment in real time through multiple types of sensors, including multi-beam sonar, underwater laser scanner and multi-frequency water flow meter, to obtain environmental information related to obstacle distribution, water flow velocity, turbulence intensity and target position, and obtain: real-time water flow velocity V f , the optimal working speed V of the robot thruster t and environmental sensitivity factor α e , forming an environmental adaptation data set; According to the number of task targets and the difficulty score of the area, the complexity of the current task is obtained, and the distribution rationality and distribution balance of the task targets are described, including the number of targets, the discreteness of the target position distribution and the priority, and the task weight W assigned to the i-th robot is obtained. i , the average distance between the target point and the robot D avg and the number of robots n, forming a task distribution dataset; The current status of each robot, including position, energy margin, and thruster load, is obtained in real time through internal sensors and communication devices to obtain the robot's current task completion rate η r , Energy consumption of a single robot E r , total energy budget of the task E t and resource sensitivity factor β r .

3. The multi-underwater robot cooperative control system based on predictive control according to claim 2 is characterized in that: The data preprocessing unit is used to denoise, normalize and remove redundant information from the collected raw data, improve data quality through a distributed processing algorithm, and provide data input for the subsequent dynamic environment modeling unit; The dynamic environment modeling unit is used to generate a real-time updated underwater three-dimensional dynamic environment model using Kalman filtering and particle filtering algorithms, including the spatiotemporal distribution information of environmental characteristics, calculate the environmental adaptation coefficient Ce, task distribution balance coefficient Ct and resource utilization efficiency coefficient Cr, and combine them to calculate the system collaborative efficiency index Is.

4. The multi-underwater robot cooperative control system based on predictive control according to claim 3 is characterized in that: The environmental adaptability coefficient Ce is calculated by the following formula: ； Where V f Indicates the real-time water flow velocity, V t represents the optimal working speed of the robot thruster, α e represents the environmental sensitivity factor, γ e represents the environmental disturbance correction coefficient; The task distribution balance coefficient Ct is calculated by the following formula: ； Where W i represents the task weight assigned to the i-th robot, D avg represents the average distance between the target point and the robot, and n represents the number of robots; The resource utilization efficiency coefficient Cr is calculated by the following formula: ； Where η r represents the current task completion rate of the robot, Er represents the energy consumption of a single robot, Et represents the total energy consumption budget of the task, β r represents the resource sensitivity factor; The system synergy efficiency index Is is calculated by the following formula: ； Where ω1, ω2 and ω3 represent the weights of the coefficients, respectively, satisfying ω1+ω2+ω3=1, Ce, Ct and Cr represent the environmental adaptation coefficient Ce, the task distribution balance coefficient Ct and the resource utilization efficiency coefficient Cr.

5. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The path planning module includes a task decomposition unit and a multi-robot path generation unit; The task decomposition unit is used to decompose a complex task into multiple subtasks based on the global task goal and utilize a priority evaluation algorithm, and reasonably allocate the tasks in combination with the robot performance and task difficulty; The multi-robot path generation unit is used to combine the obstacle distribution and water flow characteristics, calculate the optimal path for each robot through the Dijkstra algorithm, avoid path crossing and resource conflicts at the same time, and dynamically optimize the path planning scheme according to the remaining energy, task path length and thruster efficiency of each robot.

6. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The robotic arm cooperative control module includes a single-arm dynamic optimization unit and a multi-arm cooperative scheduling unit; The single-arm dynamic optimization unit constructs an initial trajectory planning model of the robotic arm based on the target point position, task requirements and obstacle distribution, uses the MPC algorithm to predict the motion state of the robotic arm in multiple time steps, optimizes the trajectory in real time, avoids motion deviation, and updates the dynamic trajectory planning parameters using the real-time data collected by the sensor. In the grasping or operation task, the force feedback of the end effector is used to sense whether the target grasping force is appropriate, and the motion speed and direction are adjusted; when the task is completed or fails, the state is fed back to the predictive control module; The multi-arm cooperative scheduling unit is used to coordinate the motion of each robotic arm through dynamic task division and path planning, prevent path conflicts, optimize the distribution of the workload at the same time, and improve the cooperative operation efficiency.

7. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The predictive control module includes a state prediction unit and an instruction generation unit; The state prediction unit is used to construct a dynamic prediction model of multiple underwater robots and their robotic arms, predict the change trend of the system cooperative efficiency index Is in multiple future time steps, and evaluate the current system state. By comparing with the high-efficiency qualified threshold A and the lowest qualified threshold B, the current system state level is determined, and three levels of comparison and evaluation strategies are defined: high-efficiency level, critical qualified level and unqualified level: High-efficiency state: Is≥A, obtain the high-efficiency level evaluation result, continue to execute the existing task assignment scheme and path planning, without changing the overall structure of the task, maintain the allocation ratio and task objectives of all robots, regularly check the operation accuracy of each robot and robotic arm according to the current system state. In the high-efficiency state, pay special attention to the key nodes in the task execution, including the target sampling position and the key operation points. During the task completion process, make small adjustments according to the sensor feedback, appropriately reduce the motion speed of the robotic arm by 20%, avoid the tiny position error caused by the water flow change in the dynamic environment, regularly check the remaining power of each robot, and estimate the energy consumption according to the task progress. Once it is monitored that the power is lower than the set threshold of 20%, issue a low-power warning, and plan the nearest charging point or supplementary task for the robot with the upcoming exhausted power; Qualified state: B≤Is<A, obtain the critical qualified level evaluation result, locally optimize the interaction area or obstacle-dense area between the robots and robotic arms in the current task, optimize the path distance, generate a shorter return path for each robot robotic arm, assign high-priority tasks to the robots with more remaining power and lower energy consumption, and set a path for the robot to avoid the water flow area in advance in the area with strong water flow to reduce the impact of water flow interference on the task on the path; Inefficient state: Is <B, obtain the unqualified level evaluation result. The system needs to quickly adjust the division of labor of the robotic arm, re-optimize the task path, re-plan the path in areas involving more complex or stronger water flow in the task, prioritize avoiding areas with large water flow disturbances, ensure that the task path is shortened as much as possible, reduce unnecessary path interference, re-decompose the unfinished tasks. If some robots are overloaded, immediately re-assign some tasks with lighter loads to robots with more sufficient resources to avoid task incompletion or delays; The instruction generation unit is used to dynamically optimize the system operation strategy according to the level evaluation result, and re-generate the control instructions for multiple robots and robotic arms through a rolling optimization method.

8. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The collaborative communication module includes a data exchange unit and a clock synchronization unit; The data exchange unit is used for real-time exchange of key data among multiple underwater robots, including robot position, task progress, environmental dynamic information, and energy status. The data content includes real-time position information, velocity vector, task status flag, energy consumption data, obstacle position, and water flow intensity. The data is hierarchically packaged and divided into high-priority and normal-priority; Acoustic communication combined with Turbo code or LDPC code technology is used to reduce the error rate and adapt to the high-noise underwater environment. In high-traffic or weak-signal situations, the data exchange unit uses dynamic compression technology to reduce bandwidth requirements and dynamically adjusts the communication frequency based on task requirements. High-priority data is transmitted in real-time, and low-priority data is sent with a delay; The clock synchronization unit is used to adopt an adaptive time synchronization protocol to achieve clock calibration among robots through timestamp messages broadcast over the network. In multi-robot communication, by collecting packets of multiple timestamp paths, the minimum offset method is used to select the best path for time calibration. Considering the characteristics of multi-path propagation underwater, a robust calibration algorithm is designed to eliminate time differences caused by different paths, and the time offset among multiple robots is monitored in real-time, and the cumulative error is calculated and gradually corrected.

9. The multi-underwater robot cooperative control system based on predictive control according to claim 1, characterized in that: The task execution module includes an execution instruction parsing unit and a feedback adjustment unit; The execution instruction parsing unit is used to parse the control instructions generated by the predictive control module and convert them into specific execution actions of the robot thrusters and robotic arms, and monitor the task execution status in real-time, including robot movement trajectory, robotic arm operation accuracy, and task completion progress, and record the execution deviation; The feedback adjustment unit is used to upload the key feedback data generated during task execution, including path deviation and energy consumption overrun, to the predictive control module.

10. A method for cooperative control of multiple underwater robots based on predictive control, applied to a cooperative control system of multiple underwater robots based on predictive control according to any one of claims 1 to 9, characterized in that: It includes the following steps: Step 1: Collect underwater environment data through sensor devices and generate an underwater three-dimensional dynamic environment model using data fusion technology, including obstacle distribution, water flow velocity, direction change, and operation point position; Step 2: Based on task requirements, including target area coverage, specific point sampling, and complex tasks such as equipment maintenance, decompose the tasks and allocate them to each robot, and plan the optimal paths of the robots and robotic arms through an optimization algorithm, considering task priorities, robot status, and resource distribution; Step 3: Based on model predictive control (MPC), the motion trajectory of the underwater robot's manipulator is dynamically optimized to provide a real-time control strategy for the division of tasks between multiple manipulators and prevent collisions between arms. Step 4: Based on the environment modeling and real-time status, predict the future motion state of the robot and its manipulator, build the optimization objective function, including task completion time, energy consumption and manipulator operation accuracy, and use the rolling optimization method to update the control instructions in real time; Step 5: Based on acoustic communication technology, support real-time data sharing among multiple underwater robots, and reduce communication delays and data loss through data compression and efficient protocol design; Step 6: Receive control instructions generated by the upper-level module, complete robot movement and manipulator operation tasks, monitor task completion and equipment status in real time, and pass feedback data to the predictive control module.

Images