Intelligent robot cloud remote control linkage system and method
By constructing a three-tiered collaborative architecture encompassing cloud-based embodied intelligent decision-making and scheduling, edge perception processing, multi-robot state synchronization, and safety monitoring, the real-time performance, accuracy, and safety issues of remote robot control in existing technologies are resolved, enabling efficient collaborative operation of multi-robot clusters.
Patent Information
- Application Number
- CN202611266035.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-20
- Publication Date
- 2026-11-10
AI Technical Summary
In existing technologies, remote robot control systems suffer from insufficient matching between cloud computing power and edge real-time performance, lack of personalized calibration mechanisms, lack of unified state synchronization and collaborative scheduling for multi-robot linkage, and inadequate system security protection. They cannot simultaneously meet the requirements of real-time remote control, accurate action execution, robustness of multi-robot collaboration, and system operational security.
The system employs a cloud-based embodied intelligent decision-making and scheduling module, an edge perception and embodied data preprocessing module, a remote control command parsing and posture calibration module, a multi-robot linkage state synchronization module, an embodied motion generation and execution control module, and a safety monitoring and anomaly self-healing protection module to construct a three-level collaborative architecture, thereby achieving data noise reduction, command calibration, state synchronization, motion generation, and safety monitoring.
It improves the real-time performance, accuracy, and stability of remote control, ensures the consistency and safety of multi-robot cluster linkage, reduces the risk of failure, and enhances the system's adaptability and engineering application value.
Smart Images

Figure CN123069201A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of embodied intelligent robot control technology, specifically to an intelligent robot cloud-based remote control linkage system and method. Background Technology
[0002] With the rapid evolution of embodied intelligence technology, robots are transforming from single autonomous execution to complex operation modes involving remote human-robot collaboration and multi-robot cluster linkage. This places higher demands on the real-time performance, motion accuracy, multi-robot collaborative robustness, and embodied perception closed-loop capabilities of remote control in scenarios such as industrial operation and maintenance, special operations, service interaction, and intelligent inspection. Currently, mainstream robot remote control systems generally suffer from the following technical bottlenecks:
[0003] The control architecture is too simple, mostly using local direct connection or simple cloud forwarding mode. It has not established a three-level collaborative architecture of cloud-based embodied decision-making, edge preprocessing, and on-the-ground execution. The cloud computing power and edge real-time performance cannot be effectively matched, resulting in large fluctuations in remote control latency and delayed command response.
[0004] The remote control commands lack a embodied calibration mechanism. The control input and the robot's pose, joint constraints and environmental perception information do not form a closed loop coupling. Direct execution of commands is prone to motion deviation and posture inaccuracy, and cannot adapt to the physical constraints and environmental interaction requirements of embodied robots.
[0005] The lack of a unified state synchronization and collaborative scheduling mechanism for multi-robot linkage leads to each robot executing tasks independently, which can easily result in motion conflicts, disordered operation sequence, and distorted formation, resulting in insufficient consistency and stability of cluster linkage.
[0006] The embodied motion generation does not combine sensory feedback and dynamic constraints, resulting in stiff motion planning, poor continuity, and a lack of error adaptive compensation capability when interacting with the environment, making it difficult to guarantee execution accuracy and interaction safety.
[0007] The system's security protection mechanism is imperfect, lacking real-time status monitoring, anomaly prediction, and fault self-healing capabilities. When communication interruptions, obstacle intrusions, or joint overloads occur during remote control, it can easily lead to operational failures or equipment damage.
[0008] The information link between cloud-based decision-making and terminal execution is broken, and embodied perception data is not effectively transmitted back to the cloud for model optimization and decision iteration, making it difficult to improve the system's long-term adaptive capabilities and intelligence level.
[0009] Existing technical solutions do not have a dedicated control architecture designed for the physical embodiment, perceptual interactivity and swarm linkage of embodied intelligent robots, and cannot simultaneously meet the requirements of real-time remote control, accurate action execution, robustness of multi-machine collaboration and security of system operation. Summary of the Invention
[0010] The purpose of this invention is to provide a cloud-based remote control and linkage system and method for intelligent robots to solve the problems existing in the prior art.
[0011] To achieve the above objectives, the present invention provides the following technical solution: a cloud-based remote control linkage system for intelligent robots, comprising a cloud-based embodied intelligent decision-making and scheduling module, an edge perception and embodied data preprocessing module, a remote control command parsing and posture calibration module, a multi-robot linkage state synchronization module, an embodied action generation and execution control module, and a safety monitoring and abnormal self-healing protection module.
[0012] The cloud-based intelligent decision-making and scheduling module is used to realize remote control task analysis, global collaborative scheduling and decision iterative optimization, and output control constraints, linkage formation rules and task execution sequences to the lower level.
[0013] The edge perception and embodied data preprocessing module is used to fuse multi-source embodied perception data, complete data noise reduction, feature extraction and spatiotemporal alignment, and transmit standardized perception features and ontology state information back upwards.
[0014] The remote control command parsing and attitude calibration module is used to receive cloud control commands and local perception data, complete command mapping, deviation correction and constraint adaptation, and output control commands that conform to the physical characteristics of the robot.
[0015] The multi-robot linkage state synchronization module is used to build a distributed synchronization network to achieve real-time alignment of the poses, actions, and task states of multiple robots, and to complete conflict resolution and formation optimization.
[0016] The embodied motion generation and execution control module is used to generate embodied motion sequences based on calibrated instructions and synchronization status, combined with robot dynamics constraints, to achieve precise motion execution.
[0017] The security monitoring and anomaly self-healing protection module is used to monitor the system's operating status in real time, identify abnormal risks, execute hierarchical protection strategies, and complete fault self-healing and reset.
[0018] Furthermore, the cloud-based embodied intelligent decision-making and scheduling module includes a task parsing unit, an embodied model reasoning unit, a multi-machine collaborative scheduling unit, and a decision iterative optimization unit;
[0019] The task parsing unit is used to decompose remote control tasks into sub-task sequences, execution constraints, and linkage interaction rules; the embodied model reasoning unit is used to combine perceptual features and ontological state to reason about the robot's adaptive action paradigm and instruction mapping relationship; the multi-machine collaborative scheduling unit is used to allocate robot operation roles, plan formation patterns, and formulate execution sequences; the decision iterative optimization unit is used to store historical operation data and optimize embodied model parameters and scheduling rules.
[0020] Furthermore, the edge perception and embodied data preprocessing module includes a multi-source perception fusion unit, a data denoising and feature extraction unit, a spatiotemporal alignment unit, and a state encapsulation and uploading unit;
[0021] The multi-source perception fusion unit is used to synchronously access multimodal perception data from vision, force, odometer, and joint status; the data denoising and feature extraction unit is used to remove perception noise and extract environmental geometric features, target interaction features, and body joint features; the spatiotemporal alignment unit is used to complete the timestamp calibration and spatial coordinate unification of multi-source perception data; and the state encapsulation and upload unit is used to encapsulate and upload standardized data and cache key information locally.
[0022] Furthermore, the remote control command parsing and attitude calibration module includes a command receiving and decoding unit, an embodied attitude mapping unit, a command calibration calculation unit, and a constraint adaptation output unit;
[0023] The instruction calibration calculation unit is equipped with an embodied attitude calibration formula:
[0024]
[0025] in, To calibrate the target pose vector of the end effector, Let be the rotation transformation matrix. This is the original control command pose vector. For translation transformation vectors, This is the real-time pose compensation vector; the constraint adaptation output unit is used to match the calibration command with the robot's kinematic constraints and output compliant control commands.
[0026] Furthermore, the multi-robot linkage state synchronization module includes a distributed synchronization communication unit, a state acquisition and broadcasting unit, a cooperative cost calculation unit, and a conflict resolution and formation optimization unit;
[0027] The collaborative cost calculation unit is configured with a collaborative cost function:
[0028]
[0029] in, The value of the collaborative cost function. The total number of robots, For neighborhood robot ensemble, For collaborative weighting coefficients, , For pose vector, , The velocity vector is used to adjust the motion parameters according to the cooperative cost function value, thereby eliminating cluster linkage conflicts.
[0030] Furthermore, the embodied action generation and execution control module includes an action trajectory planning unit, a dynamic constraint adaptation unit, an error compensation execution unit, and an execution status feedback unit;
[0031] The error compensation execution unit is equipped with a formula for compensating for errors in the execution of specific actions:
[0032]
[0033] in, This is the joint torque compensation vector. , , The coefficient matrix, This is the pose deviation vector. The speed deviation vector is used; the execution status feedback unit is used to transmit execution parameters back in real time to form closed-loop control.
[0034] Furthermore, the security monitoring and anomaly self-healing protection module includes a full-dimensional status monitoring unit, an anomaly identification and prediction unit, a hierarchical protection execution unit, and a fault self-healing reset unit;
[0035] The all-dimensional status monitoring unit monitors the quality of the communication link, the status of cloud command transmission and reception, the validity of edge perception data, the robot's operating parameters, environmental obstacle information, and the multi-robot collaborative status in real time. The anomaly identification and prediction unit builds an anomaly feature library and identifies anomaly types such as communication interruption, command anomaly, joint overload, obstacle intrusion, and multi-robot conflict through status threshold judgment, trend prediction, and pattern matching. The graded protection execution unit executes graded protection strategies according to the anomaly level: minor anomalies execute local corrections, moderate anomalies execute deceleration and pause, and severe anomalies execute emergency stop and physical locking. The fault self-healing and reset unit automatically resumes operation execution after the anomaly is eliminated, re-establishing the communication link, synchronizing status data, and calibrating execution parameters according to the preset reset logic, without manual intervention.
[0036] A linkage method for a cloud-based remote control and linkage system for intelligent robots includes the following steps:
[0037] Step 1: System initialization and cloud-edge-device link establishment, complete model loading, coordinate system calibration, communication configuration and cluster networking, and the system enters standby state;
[0038] Step 2: The cloud receives remote control tasks, completes task analysis, embodied reasoning and global scheduling, and issues control and scheduling commands;
[0039] Step 3: Remote control command parsing and attitude calibration, complete command decoding, attitude mapping, deviation correction and constraint adaptation, and output control commands;
[0040] Step 4: Synchronize and optimize the state of multiple robots in linkage, realize state acquisition, broadcasting, cost calculation and conflict resolution, and ensure the consistency of cluster linkage;
[0041] Step 5: Generate embodied motion and execute it in a closed loop, completing trajectory planning, dynamics adaptation, error compensation and status feedback to achieve precise operation;
[0042] Step 6: Safety monitoring and anomaly self-healing protection, real-time monitoring of status, identification of anomalies, execution of graded protection and completion of fault reset;
[0043] Step 7: Task completion and system iteration optimization, record job data, optimize model rules, and reset all modules to standby.
[0044] Furthermore, in step 4, the multi-robot linkage adopts a distributed synchronization network, supporting both single-master-multiple-slave and fully distributed synchronization modes; in step 5, the embodied motion execution combines joint space and Cartesian space dual trajectory planning to ensure motion continuity and dynamic compliance.
[0045] Compared with the prior art, the beneficial effects of the present invention are:
[0046] This invention employs a three-tiered cloud-edge-device collaborative architecture: cloud-based embodied decision-making, edge real-time preprocessing, and precise execution by the robot itself. This architecture fully leverages the computing power advantages of the cloud and the real-time requirements of the edge, reducing remote control latency fluctuations and improving command response efficiency and system adaptability. Through a remote control command posture calibration mechanism, closed-loop coupling is achieved between control input, robot physical constraints, and environmental perception information, eliminating command deviations and coordinate system mismatches, thus improving the accuracy and stability of remote control. A scheduling mechanism based on a distributed synchronization network and a collaborative cost function enables real-time alignment of multiple robot states, automatic conflict resolution, and dynamic optimization of formation, improving the consistency, timing rationality, and operational efficiency of cluster linkage. Combining dynamic constraints and error compensation algorithms, a continuous and smooth embodied action sequence is generated, correcting execution deviations in real time and ensuring the compliance, accuracy, and safety of robot-environment interaction. A comprehensive state monitoring, anomaly prediction, hierarchical protection, and fault self-healing mechanism reduces the risk of remote control failures, minimizes the need for manual intervention, and enhances the system's continuous operation capability and environmental adaptability. The cloud-based decision-making iterative optimization mechanism enables continuous upgrades to models and rules, and the modular design supports the access of multiple types of embodied robots and the adaptation to multiple scenarios and tasks, thus possessing broad engineering application value. Attached Figure Description
[0047] Figure 1 This is a system module diagram of the present invention;
[0048] Figure 2 This is a schematic diagram of the cloud-based embodied intelligent decision-making and scheduling module of the present invention;
[0049] Figure 3 This is a schematic diagram of the edge sensing and embodied data preprocessing module of the present invention;
[0050] Figure 4 This is a schematic diagram of the remote control command parsing and attitude calibration module of the present invention;
[0051] Figure 5 This is a flowchart of the method of the present invention. Detailed Implementation
[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0053] Please see Figure 1-5 The present invention provides an intelligent robot cloud remote control linkage system, including a cloud-based embodied intelligent decision-making and scheduling module, an edge perception and embodied data preprocessing module, a remote control command parsing and posture calibration module, a multi-robot linkage state synchronization module, an embodied action generation and execution control module, and a safety monitoring and abnormal self-healing protection module.
[0054] The cloud-based embodied intelligent decision-making and scheduling module is the core of the system, including a task parsing unit, an embodied model reasoning unit, a multi-machine collaborative scheduling unit, and a decision iterative optimization unit.
[0055] Task parsing unit: Receives task input from remote control terminal, breaks it down into sub-task sequence, execution constraints, workspace range and linkage interaction rules, and completes the conversion of task semantics into machine-executable logic;
[0056] Embodied Model Reasoning Unit: Loads a pre-trained embodied intelligent model, combines edge-retrieved sensory features and ontological state, and reasones about the robot's adaptive action paradigm, joint movement range, environmental interaction taboos, and the mapping relationship between control commands.
[0057] Multi-machine collaborative scheduling unit: Based on task requirements and robot cluster status, it allocates job roles, plans formation patterns, determines execution sequence, and generates global collaborative scheduling instructions to avoid multi-machine movement conflicts and resource competition.
[0058] Decision-making iterative optimization unit: Stores historical control data, execution results and perception feedback, and continuously learns to optimize embodied model parameters, scheduling rules and calibration parameters to improve the system's long-term adaptive capability.
[0059] The edge perception and embodied data preprocessing module is deployed on the edge computing unit of a single robot, including a multi-source perception fusion unit, a data denoising and feature extraction unit, a spatiotemporal alignment unit, and a state encapsulation and uploading unit;
[0060] Multi-source perception fusion unit: Synchronously accesses robot visual perception, force and tactile perception, odometry perception, and joint state perception data to build a unified perception input interface and realize the synchronous acquisition of multi-modal data;
[0061] Data denoising and feature extraction unit: Adaptive filtering algorithm is used to remove random noise and outliers from the perceived data, and to extract environmental geometric features, target interaction features, body joint features and contact force features;
[0062] Spatiotemporal alignment unit: Based on the robot system clock, it completes the timestamp calibration and spatial coordinate unification of multi-source sensing data, eliminating sampling delay and coordinate system deviation of different sensors;
[0063] State encapsulation and upload unit: encapsulates standardized perception features and ontology state data into a fixed-format data packet, uploads it to the cloud decision module through a communication link, and caches key data locally for real-time edge computing.
[0064] The remote control command parsing and attitude calibration module includes a command receiving and decoding unit, a embodied attitude mapping unit, a command calibration calculation unit, and a constraint adaptation output unit. The command receiving and decoding unit receives the original remote control commands from the cloud, performs command verification, format decoding, and validity judgment, and eliminates abnormal commands and redundant data. The embodied attitude mapping unit establishes the mapping relationship between the control input space and the robot's joint space and end effector pose space, converting abstract control commands into target robot pose parameters. The command calibration calculation unit corrects command deviations based on the embodied attitude calibration formula, eliminating control input errors, coordinate system deviations, and mechanical installation deviations. The formula is as follows:
[0065]
[0066] in, To calibrate the target pose vector of the end effector, Let be the rotation transformation matrix. This is the original control command pose vector. For translation transformation vectors, This is the real-time pose compensation vector, calculated from edge-aware data.
[0067] Constraint Adaptation Output Unit: Matches the calibrated pose command with the robot joint limits, kinematic constraints, and environmental space constraints, eliminates commands that exceed the limits, and outputs the final control command that conforms to the physical constraints.
[0068] The multi-robot linkage state synchronization module includes a distributed synchronization communication unit, a state acquisition and broadcasting unit, a cooperative cost calculation unit, and a conflict resolution and formation optimization unit. The distributed synchronization communication unit constructs a point-to-point distributed communication network between robots to achieve low-latency state data interaction, supporting both single-master-multiple-slave and fully distributed synchronization modes. The state acquisition and broadcasting unit collects real-time information such as the robot's pose, joint angles, execution progress, and fault status, broadcasts synchronization data to other robots in the cluster, and simultaneously receives state information from other robots. The cooperative cost calculation unit evaluates the cluster's linkage consistency and conflict risk based on a cooperative cost function, as shown in the following formula:
[0069]
[0070] in, The value of the collaborative cost function. The total number of robots, For neighborhood robot ensemble, For collaborative weighting coefficients, , For pose vector, , This is the velocity vector.
[0071] Conflict resolution and formation optimization unit: Adjusts robot motion parameters based on the cooperative cost function value. When the cost function exceeds the threshold, conflict resolution is triggered to correct the motion trajectory and formation, ensuring stable cluster linkage.
[0072] The embodied motion generation and execution control module includes a motion trajectory planning unit, a dynamic constraint adaptation unit, an error compensation execution unit, and an execution state feedback unit. The motion trajectory planning unit generates continuous and smooth joint space and Cartesian space trajectories based on calibrated control commands and multi-machine synchronization status, ensuring motion continuity and curvature consistency. The dynamic constraint adaptation unit performs dynamic smoothing of the trajectory by considering the robot's mass distribution, inertial parameters, load state, and joint torque limitations, avoiding joint overload and motion impact. The error compensation execution unit corrects trajectory deviations in real time based on the embodied motion execution error compensation formula, improving execution accuracy. The formula is as follows:
[0073]
[0074] in, This is the joint torque compensation vector. , , The coefficient matrix, This is the deviation vector between the actual pose and the target pose; This is the deviation vector between the actual speed and the target speed.
[0075] Execution status feedback unit: collects joint execution parameters, end-effector pose error, and interaction force information in real time, and sends them back to the command calibration module and status synchronization module to form closed-loop control.
[0076] The safety monitoring and anomaly self-healing protection module includes a full-dimensional status monitoring unit, an anomaly identification and prediction unit, a graded protection execution unit, and a fault self-healing reset unit. The full-dimensional status monitoring unit monitors real-time communication link quality, cloud command transmission and reception status, edge perception data validity, robot operating parameters, environmental obstacle information, and multi-robot collaborative status. The anomaly identification and prediction unit constructs an anomaly feature library and identifies anomaly types such as communication interruption, command anomalies, joint overload, obstacle intrusion, and multi-robot conflict through status threshold judgment, trend prediction, and pattern matching. The graded protection execution unit executes graded protection strategies based on the anomaly level: minor anomalies trigger local corrections, moderate anomalies trigger deceleration and pause, and severe anomalies trigger emergency shutdown and physical locking. The fault self-healing reset unit automatically resumes operation after the anomaly is eliminated, re-establishing the communication link, synchronizing status data, and calibrating execution parameters according to preset reset logic, without manual intervention.
[0077] A linkage method for a cloud-based remote control and linkage system for intelligent robots includes the following steps:
[0078] Step 1: System initialization and cloud-edge-device link establishment
[0079] The cloud-based embodied intelligent decision-making and scheduling module and the edge modules of each robot are activated to complete the loading of the embodied model, coordinate system calibration, communication link configuration and cluster networking; the cloud sends out the initial parameters, the edge devices complete the self-check of the perception sensors and the status reporting, and the system enters the standby state.
[0080] Step 2: Remote control task reception and global scheduling
[0081] The cloud receives task input from the remote control terminal, which is then broken down into sub-tasks and execution constraints by the task parsing unit; the embodied model reasoning unit combines edge perception data to generate control rules and action paradigms; and the multi-machine collaborative scheduling unit assigns roles, plans formations and execution sequences for robots in the cluster, and issues global scheduling instructions.
[0082] Step 3: Remote control command parsing and embodied calibration
[0083] The remote control command parsing and attitude calibration module receives the original control commands from the cloud, completes decoding and validity verification; establishes the correspondence between the control space and the body space through attitude mapping, and completes command correction using the embodied attitude calibration formula; and outputs adapted control commands in combination with robot kinematic constraints.
[0084] Step 4: Synchronization and Cooperative Optimization of Multi-Robot Linkage Status
[0085] The multi-robot linkage state synchronization module collects and broadcasts its own status through a distributed network and receives status information from neighboring robots; it calculates linkage consistency index based on the collaborative cost function to determine whether there is motion conflict; if the conflict risk exceeds the standard, it performs trajectory correction and formation optimization to ensure cluster synchronization.
[0086] Step 5: Embodied Motion Generation and Closed-Loop Execution
[0087] The embodied motion generation and execution control module plans a smooth motion trajectory based on calibration instructions and synchronization status; adapts to robot dynamics constraints and corrects execution deviations in real time through error compensation formulas; and continuously feeds back status data during execution, forming an embodied closed loop of perception-decision-execution-feedback.
[0088] Step 6: Security Monitoring and Anomaly Self-Healing Protection
[0089] The safety monitoring module monitors the system's operating status in all dimensions, identifies anomalies in real time and predicts risks; it triggers graded protection strategies according to the anomaly level; after the anomaly is resolved, it executes self-healing reset logic to resynchronize the status, calibrate parameters, and restore remote control linkage operations.
[0090] Step 7: Task Completion and System Iteration Optimization
[0091] After the robot completes the target task, the cloud receives the execution results and perception feedback, and records the task data; the decision iteration optimization unit updates the embodied model and scheduling rules, and each module resets to the standby state, waiting for the next task instruction.
[0092] Example 1: Remote and precise control of a single embodied intelligent robot
[0093] This embodiment demonstrates a remote, human-machine collaborative, precise interactive operation using a single embodied intelligent robot, applicable to scenarios such as remote precision operation, target grasping, and environmental interaction.
[0094] Implementation process:
[0095] System initialization: The embodied intelligent model is loaded in the cloud, the robot's edge device completes the initialization of the perception sensors, coordinate system calibration and communication link establishment, uploads the initial state of the body to the cloud, and the system enters standby mode.
[0096] Task issuance: The remote control terminal inputs precise interactive tasks, and the cloud-based task parsing unit breaks them down into sub-instructions such as end-effector pose target, interactive force constraints, and motion space limitations; the embodied model reasoning unit generates control instruction mapping rules and issues them to the robot edge.
[0097] Command parsing and calibration: The command parsing module receives the original control commands, completes decoding and validity judgment; converts the control input into the end target pose through attitude mapping, uses the embodied attitude calibration formula to eliminate coordinate system deviation and control error, and outputs control commands that conform to joint constraints.
[0098] Embodied motion execution: The motion generation module plans a smooth Cartesian space trajectory to adapt to robot dynamics constraints; the error compensation unit collects end-effector pose deviations in real time and corrects joint torques through error compensation formulas to ensure accurate motion execution; pose and force information are continuously fed back during execution to form a closed loop.
[0099] Safety monitoring and completion: The safety monitoring module monitors the communication, robot body and environment status, and continues to execute if there are no abnormalities; after the robot completes the precise interaction, it uploads the execution results to the cloud, the cloud records the data and optimizes the model, and the system resets and stands by.
[0100] This embodiment achieves high real-time performance, high accuracy, and high safety in remote control of a single embodied robot through cloud-based embodied decision-making, real-time edge preprocessing, embodied command calibration, and closed-loop action execution, effectively avoiding command deviation, inaccurate actions, and uncontrolled interaction.
[0101] Example 2: Multiple embodied intelligent robots working in a cloud-based coordinated formation
[0102] This embodiment demonstrates the remote, coordinated operation of multiple embodied intelligent robots in a cloud-based formation, suitable for scenarios such as cluster inspection, collaborative handling, and formation work.
[0103] Implementation process:
[0104] Cluster initialization: The cloud completes the multi-machine networking, formation rule configuration and collaborative weight setting; each robot's edge terminal completes state self-check, perception data calibration and distributed synchronization network establishment, and reports initial pose and status information.
[0105] Collaborative task scheduling: The cloud receives collaborative tasks and parses them into parameters such as formation, movement path, operation sequence, and collaborative constraints; the multi-machine scheduling unit assigns navigator / follower roles to each robot, plans the global formation trajectory, and issues scheduling instructions.
[0106] State synchronization and collaborative optimization: Each robot broadcasts its own pose, speed and execution status through a distributed network and receives information from neighboring robots; the collaborative cost calculation unit calculates the linkage cost function in real time to determine the formation consistency and conflict risk; if the cost exceeds the limit, conflict resolution is triggered to correct the motion trajectory and speed parameters.
[0107] Command calibration and motion execution: Each robot independently completes the parsing of control commands and attitude calibration, and generates an adapted motion trajectory by combining the synchronization status; the error compensation unit corrects the execution deviation to ensure the stability of the formation and the consistency of the action sequence.
[0108] Safety protection and self-healing: The safety monitoring module monitors the multi-machine collaborative status, communication links and environmental risks. When local deviations occur, it performs local corrections. When communication fluctuations occur, it performs cache synchronization. When serious conflicts occur, it performs graded shutdown. After the anomaly is eliminated, it automatically resets and resumes formation operation.
[0109] Task completed: The cluster completes the coordinated operation, each robot uploads execution data, the cloud optimizes the collaborative scheduling rules, and the system resets and stands ready.
[0110] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A cloud-based remote control and linkage system for intelligent robots, characterized in that: It includes a cloud-based embodied intelligent decision-making and scheduling module, an edge perception and embodied data preprocessing module, a remote control command parsing and posture calibration module, a multi-robot linkage state synchronization module, an embodied motion generation and execution control module, and a safety monitoring and anomaly self-healing protection module. The cloud-based intelligent decision-making and scheduling module is used to realize remote control task analysis, global collaborative scheduling and decision iterative optimization, and output control constraints, linkage formation rules and task execution sequences to the lower level. The edge perception and embodied data preprocessing module is used to fuse multi-source embodied perception data, complete data noise reduction, feature extraction and spatiotemporal alignment, and transmit standardized perception features and ontology state information back upwards. The remote control command parsing and attitude calibration module is used to receive cloud control commands and local perception data, complete command mapping, deviation correction and constraint adaptation, and output control commands that conform to the physical characteristics of the robot. The multi-robot linkage state synchronization module is used to build a distributed synchronization network to achieve real-time alignment of the poses, actions, and task states of multiple robots, and to complete conflict resolution and formation optimization. The embodied motion generation and execution control module is used to generate embodied motion sequences based on calibrated instructions and synchronization status, combined with robot dynamics constraints, to achieve precise motion execution. The security monitoring and anomaly self-healing protection module is used to monitor the system's operating status in real time, identify abnormal risks, execute hierarchical protection strategies, and complete fault self-healing and reset.
2. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The cloud-based embodied intelligent decision-making and scheduling module includes a task parsing unit, an embodied model reasoning unit, a multi-machine collaborative scheduling unit, and a decision iterative optimization unit. The task parsing unit is used to decompose remote control tasks into sub-task sequences, execution constraints, and linkage interaction rules; the embodied model reasoning unit is used to combine perceptual features and ontological state to reason about the robot's adaptive action paradigm and instruction mapping relationship; the multi-machine collaborative scheduling unit is used to allocate robot job roles, plan formation patterns, and formulate execution sequences; the decision iterative optimization unit is used to store historical job data and optimize embodied model parameters and scheduling rules.
3. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The edge perception and embodied data preprocessing module includes a multi-source perception fusion unit, a data noise reduction and feature extraction unit, a spatiotemporal alignment unit, and a state encapsulation and uploading unit. The multi-source perception fusion unit is used to simultaneously access multimodal perception data from vision, force, odometer, and joint status; the data denoising and feature extraction unit is used to remove perception noise and extract environmental geometric features, target interaction features, and body joint features. The spatiotemporal alignment unit is used to complete the timestamp calibration and spatial coordinate unification of multi-source sensing data; the state encapsulation and upload unit is used to encapsulate and upload standardized data and cache key information locally.
4. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The remote control command parsing and attitude calibration module includes a command receiving and decoding unit, an embodied attitude mapping unit, a command calibration calculation unit, and a constraint adaptation output unit. The instruction calibration calculation unit is equipped with an embodied attitude calibration formula: ; in, To calibrate the target pose vector of the end effector, Let be the rotation transformation matrix. This is the original control command pose vector. For translation transformation vectors, This is the real-time pose compensation vector; the constraint adaptation output unit is used to match the calibration command with the robot's kinematic constraints and output compliant control commands.
5. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The multi-robot linkage state synchronization module includes a distributed synchronization communication unit, a state acquisition and broadcasting unit, a cooperative cost calculation unit, and a conflict resolution and formation optimization unit. The collaborative cost calculation unit is configured with a collaborative cost function: ; in, The value of the collaborative cost function. The total number of robots, For neighborhood robot ensemble, For collaborative weighting coefficients, , For pose vector, , The velocity vector is used to adjust the motion parameters according to the cooperative cost function value, thereby eliminating cluster linkage conflicts.
6. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The embodied action generation and execution control module includes an action trajectory planning unit, a dynamic constraint adaptation unit, an error compensation execution unit, and an execution status feedback unit; The error compensation execution unit is equipped with a formula for compensating for errors in the execution of specific actions: ; in, This is the joint torque compensation vector. , , The coefficient matrix, This is the pose deviation vector. The speed deviation vector is used; the execution status feedback unit is used to transmit execution parameters back in real time to form closed-loop control.
7. The intelligent robot cloud-based remote control and linkage system according to claim 1, characterized in that: The security monitoring and anomaly self-healing protection module includes a full-dimensional status monitoring unit, an anomaly identification and prediction unit, a hierarchical protection execution unit, and a fault self-healing reset unit. The all-dimensional status monitoring unit monitors the quality of the communication link, the status of cloud command transmission and reception, the validity of edge perception data, the robot's operating parameters, environmental obstacle information, and the status of multi-robot collaboration in real time; the anomaly identification and prediction unit builds an anomaly feature library and identifies anomaly types such as communication interruption, command anomaly, joint overload, obstacle intrusion, and multi-robot conflict through status threshold judgment, trend prediction, and pattern matching. The graded protection execution unit executes graded protection strategies according to the level of abnormality. For minor abnormalities, it performs local correction; for moderate abnormalities, it performs deceleration and pause; and for severe abnormalities, it performs emergency shutdown and physical locking. After the abnormality is eliminated, the fault self-healing reset unit re-establishes the communication link, synchronizes status data, and calibrates execution parameters according to the preset reset logic, and automatically resumes operation without manual intervention.
8. The linkage method of a cloud-based remote control linkage system for intelligent robots according to any one of claims 1-7, characterized in that: Includes the following steps: Step 1: System initialization and cloud-edge-device link establishment, complete model loading, coordinate system calibration, communication configuration and cluster networking, and the system enters standby state; Step 2: The cloud receives remote control tasks, completes task analysis, embodied reasoning and global scheduling, and issues control and scheduling commands; Step 3: Remote control command parsing and attitude calibration, complete command decoding, attitude mapping, deviation correction and constraint adaptation, and output control commands; Step 4: Synchronize and optimize the state of multiple robots in linkage, realize state acquisition, broadcasting, cost calculation and conflict resolution, and ensure the consistency of cluster linkage; Step 5: Generate embodied motion and execute it in a closed loop, completing trajectory planning, dynamics adaptation, error compensation and status feedback to achieve precise operation; Step 6: Safety monitoring and anomaly self-healing protection, real-time monitoring of status, identification of anomalies, execution of graded protection and completion of fault reset; Step 7: Task completion and system iteration optimization, record job data, optimize model rules, and reset all modules to standby.
9. The linkage method of a cloud-based remote control linkage system for intelligent robots according to claim 8, characterized in that: In step 4, the multi-robot linkage adopts a distributed synchronization network, supporting both single-master-multiple-slave and fully distributed synchronization modes; in step 5, the embodied motion execution combines joint space and Cartesian space dual trajectory planning to ensure motion continuity and dynamic compliance.
