An edge-computing-based industrial robot remote debugging system and method

Abstract

The application belongs to the technical field of industrial robots, and discloses an industrial robot remote debugging system and method based on edge computing, which comprises an edge state acquisition module, is arranged on an edge node of an industrial robot, acquires running state information of the industrial robot, converts the running state information into a standardized state segment, analyzes a debugging instruction issued by a remote debugging end, and disassembles the debugging instruction into a structured instruction sequence; an edge state reconstruction module, which establishes a space-time reconstruction result of a field running state based on the standardized state segment and the structured instruction sequence, and generates a dynamic conflict domain indicating a conflict risk; and an action calibration module, which maps the structured instruction sequence into a debugging action according to time, performs segmented calibration on the debugging action according to the dynamic conflict domain, backfills parameters of the calibrated debugging action according to a preset priority, and forms a calibrated action sequence; thus, the debugging efficiency and operation stability of the industrial robot in a complex scene are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of industrial robot technology, and more specifically, to a remote debugging system and method for industrial robots based on edge computing. Background Technology

[0002] Existing remote debugging systems and methods for industrial robots mainly suffer from the following problems:

[0003] With the continuous improvement of industrial automation and flexible manufacturing, industrial robots are being deployed in large numbers in assembly, handling, inspection, welding and other work scenarios, and remote debugging technology is gradually becoming an important requirement in industrial sites.

[0004] Existing remote commissioning systems for industrial robots typically rely on the command sequence itself or motion simulation based on simplified models, and can only check the corresponding discrete postures or trajectory segments of the robot. These methods cannot accurately determine whether the robot's movements may interfere with obstacles under real environmental parameters, nor can they identify the risk of spatial conflicts during continuous movements in advance during the commissioning phase.

[0005] Traditional collision detection is mostly based on discrete sampling points, which can only identify single-point collisions and cannot characterize the spatial occupancy of the action unit throughout the entire execution time, let alone quantify the degree of interaction between the spatial volume occupied and surrounding obstacles. Due to the lack of a continuous spatial representation of the action trajectory, existing technologies struggle to calculate the geometric volume of potential conflict areas and further convert it into risk indicators that can be used for debugging purposes.

[0006] On the other hand, industrial environments are often limited by edge computing resources. Existing debugging systems are often unable to construct the spatial envelope of actions in real time at the field equipment end, nor can they complete highly complex Boolean geometric operations. This makes risk filtering often rely on remote computing or human experience, which is difficult to meet the requirements of real-time performance and security in a variable production environment.

[0007] For multi-robot collaborative scenarios, existing technologies still rely on empirical parameters or fixed safety distances for spatial conflict relationship analysis, making it difficult to provide fine-grained, computable interference metrics. Due to the lack of realistic interference models based on robot reachability domains and trajectory spaces, scheduling strategies are often coarse, easily leading to problems such as accidental avoidance, insufficient avoidance, or operational blockage. Furthermore, scheduling is often prioritized based on task sequence or action type, failing to consider task urgency, risk level, execution coupling relationships, and the potential costs of task failure, making it difficult to achieve reasonable resource allocation when multiple robots are operating in parallel.

[0008] Furthermore, existing technologies typically lack a secure closed-loop mechanism from motion generation to command issuance, often issuing commands directly without trusted verification or secure authorization. This exposes remote debugging processes to risks such as accidental triggering, unauthorized execution, and cross-robot interference. Existing technologies cannot automatically divide multi-robot actions into parallel or sequential execution batches based on spatial conflicts and task attributes, significantly limiting the efficiency of multi-robot collaboration.

[0009] In view of this, the present invention proposes an edge computing-based remote debugging system and method for industrial robots to solve the above problems. Summary of the Invention

[0010] To overcome the aforementioned shortcomings of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a remote debugging system for industrial robots based on edge computing, comprising:

[0011] The edge status acquisition module is deployed at the edge node of the industrial robot to acquire the operating status information of the industrial robot, convert the operating status information into standardized status fragments, and parse the debugging instructions issued by the remote debugging terminal, breaking them down into structured instruction sequences.

[0012] The edge state reconstruction module establishes the spatiotemporal reconstruction results of the field operation state based on standardized state fragments and structured instruction sequences, and generates a dynamic conflict domain indicating conflict risk.

[0013] The action calibration module maps the structured instruction sequence into debugging actions according to time, performs segmented calibration of the debugging actions according to the dynamic conflict domain, and backfills the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence.

[0014] The edge-side verification module performs reversible verification on the calibration action sequence, generates forward execution simulation results and reverse playback simulation results, and calculates the difference between the simulation results; if the difference exceeds the preset difference threshold, the calibration action sequence is rolled back to the previous level calibration point for recalibration, and a reliable action sequence is output.

[0015] The remote collaborative scheduling module performs remote authorization, security locking, and multi-robot collaborative scheduling of trusted action sequences, outputs batch execution instructions, and sends them to the corresponding industrial robots.

[0016] Preferably, the method for converting operational status information into standardized state fragments includes:

[0017] The edge status acquisition module is deployed at the edge node of the industrial robot. The edge node includes various sensors used to collect the operating status information of the industrial robot during operation. The sensors include position sensors, torque sensors, current sensors, and environmental sensors. The operating status information includes the position and posture information, kinematic information, electrical drive status information, health status information, and environmental information of the industrial robot.

[0018] The running status information is synchronized by network time protocol, and the time-synchronized running status information is normalized to map data with different dimensions and amplitude ranges to a unified numerical range, so as to obtain normalized running status information.

[0019] The normalized operating status information is divided into discrete state segments according to the preset sampling period; each discrete state segment is processed for format standardization and attribute identification, and a unique identifier is assigned to each discrete state segment, thereby forming a standardized state segment.

[0020] Preferably, the method for decomposing into a structured instruction sequence includes:

[0021] The edge situation acquisition module is also used to receive debugging instructions issued by the remote debugging terminal, and to parse and decompose the debugging instructions, decompose the debugging instructions into basic action units, and map each basic action unit to a standardized state segment in time or logic.

[0022] The mapped basic action units are sorted by execution order to form a structured instruction sequence. The structured instruction sequence includes different instruction units, each of which includes action type, target parameters, execution time, and other control attributes.

[0023] Preferably, the method for establishing the spatiotemporal reconstruction result of the on-site operating status includes:

[0024] The system receives standardized state segments and structured action instruction sequences. It maps each basic action unit in the structured action instruction sequence to the corresponding standardized state data segment according to a preset execution time window, establishes the temporal and spatial correspondence between the basic action unit and the standardized state segment, generates the predicted motion trajectory of the basic action unit, and thus establishes the spatiotemporal reconstruction result of the on-site operation state.

[0025] Preferably, the method for generating a dynamic conflict domain indicating conflict risk includes:

[0026] Based on the predicted motion trajectory of the basic motion unit, a motion trajectory envelope is constructed within the preset execution time window of the basic motion unit. The motion trajectory envelope is intersected with the preset set of environmental obstacles to generate a dynamic conflict domain.

[0027] A conflict indication function is defined using a dynamic conflict domain to indicate the potential conflict risks that may occur during the execution of basic motion units; conflict risks include potential interference, joint limit triggering, or the risk of accumulated motion errors.

[0028] Preferably, the method for mapping structured instruction sequences to debugging actions over time includes:

[0029] Based on the semantic tags, action parameters, and instruction dependencies in the structured instruction sequence, the start and end times of the execution of each basic action unit are determined, and each basic action unit is mapped to a unified global time axis.

[0030] The basic motion units are sorted according to the global timeline to generate a motion scheduling sequence. Based on the preset execution time window of each basic motion unit, the motion requirements in the motion scheduling sequence are transformed into debugging actions to be performed by the industrial robot.

[0031] Preferably, the method for forming the calibration action sequence includes:

[0032] The continuous motion trajectory of the debugging action within the preset execution time window is discretized to obtain different time slice trajectory segments; based on the dynamic conflict domain, it is detected whether each time slice trajectory segment intersects with the dynamic conflict domain;

[0033] When a time slice trajectory segment enters the dynamic conflict domain, the control parameters of the time slice trajectory segment are adjusted. The adjustment includes modifying the path offset, the joint angle redundancy allocation method, or the end effector attitude, so that the corrected time slice trajectory segment avoids the dynamic conflict domain, thereby forming a conflict-free calibration time slice trajectory segment.

[0034] Adjustable parameters generated during calibration in the calibration time slice trajectory segment are backfilled according to a preset priority. Each calibration time slice trajectory segment with completed parameter backfilling is then recombined according to the time sequence of the basic motion units to generate a calibration motion sequence with definite motion parameters, no interference, and executable by the industrial robot.

[0035] Preferably, the method for outputting a reliable action sequence includes:

[0036] A forward execution simulation is constructed based on the calibration action sequence to simulate and deduce the execution process of the industrial robot from the initial state to the preset target state, and the forward execution simulation results are obtained. Based on the forward execution simulation results, a corresponding reverse playback sequence is constructed, and the reverse execution process is simulated and deduced to obtain the reverse playback simulation results.

[0037] Differential calculations are performed on the forward execution simulation results and the reverse playback simulation results to obtain the differences in key poses, joint angles, and control parameters. When the difference does not exceed the preset difference threshold, the calibration action sequence is determined as a reliable action sequence. When the difference exceeds the preset difference threshold, the calibration action sequence is rolled back to the previous level calibration point and re-calibrated until a reliable action sequence that meets the preset difference threshold requirement is obtained.

[0038] Preferably, the method for outputting batch instructions and sending them to the corresponding industrial robots includes:

[0039] Based on the action sets of each industrial robot in the trusted action sequence, the reachable domain of the corresponding industrial robot is constructed. The reachable domain is obtained by kinematic mapping from the joint feasible domain of the industrial robot. Spatial intersection operation is performed on the reachable domains of any two industrial robots to obtain the interaction interference region of the two industrial robots in the work space, and interference weights are defined for the interaction interference region.

[0040] An interference graph is constructed based on the interference weights among all industrial robots. Nodes in the interference graph correspond to industrial robots, and the weights of the edges correspond to the interference weights. Using the spatial conflict factor and task attribute factor in the interference graph, the scheduling priority of each industrial robot is calculated according to the preset trade-off coefficients.

[0041] The spatial conflict factor is the sum of the interference weights of each node, and the task attribute factor is the task attribute parameter of each industrial robot. The task attribute factor is the comprehensive metric value after weighted fusion of the task attribute parameters of each industrial robot. The task attribute parameters include task urgency, safety sensitivity, task coupling, quality criticality, and resource occupancy.

[0042] Based on scheduling priority, time windows are allocated and batches are divided for each industrial robot action in the trusted action sequence to generate execution batch instructions. After generating the execution batch instructions, remote authorization information and security lock parameters are attached to each batch instruction, and the authorized execution batch instructions are sent to the corresponding industrial robots respectively.

[0043] A remote debugging method for industrial robots based on edge computing includes:

[0044] S1. Deployed at the edge node of the industrial robot, it acquires the operating status information of the industrial robot, converts the operating status information into standardized status fragments, and parses the debugging instructions issued by the remote debugging terminal, breaking them down into structured instruction sequences.

[0045] S2. Based on standardized state fragments and structured instruction sequences, establish the spatiotemporal reconstruction results of the field operation status and generate a dynamic conflict domain indicating conflict risk.

[0046] S3. Map the structured instruction sequence to debugging actions according to time, perform segmented calibration of the debugging actions according to the dynamic conflict domain, and backfill the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence.

[0047] S4. Perform reversible verification on the calibration action sequence, generate forward execution simulation results and reverse playback simulation results, and calculate the difference between the simulation results; if the difference exceeds the preset difference threshold, roll back the calibration action sequence to the previous level calibration point for recalibration, and output a reliable action sequence.

[0048] S5. Perform remote authorization, security locking, and multi-robot collaborative scheduling on trusted action sequences, output batch execution instructions, and send them to the corresponding industrial robots.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] By explicitly modeling the spatial occupancy of basic motion units during continuous execution and generating motion envelopes through Minkowski expansion, remote robot debugging achieves capabilities similar to real-time spatial prediction. Intersection operations with a pre-set set of obstacles can identify potential interference caused by trajectory deviations, posture changes, or error propagation in advance, significantly reducing risks during offline programming and remote debugging. A conflict indication function enables instantaneous quantification of conflict risks, allowing the system to provide clear motion safety feedback early in debugging, improving the interpretability and robustness of robot motion planning. This method makes the remote debugging process more automated, efficient, and controllable, helping to reduce repeated trial-and-error time and improve the safety and accuracy of remote debugging.

[0051] By mapping reachable domains based on trusted action sequences, calculating interaction interference zones, and constructing interference relationship graphs, a quantifiable and computable expression of spatial conflict relationships among multiple robots is achieved. By fusing spatial conflict factors with task attribute factors and using trade-off coefficients to calculate scheduling priorities, the scheduling process simultaneously considers workspace safety and task importance. Through time window allocation and batching of subtasks based on scheduling priorities, the system can automatically generate efficient execution batch instructions, enabling automated orchestration of parallel and sequential actions during multi-robot operations, thus improving collaborative efficiency. By attaching safety locking parameters based on behavioral trustworthiness to execution batches, closed-loop control is achieved between simulation trustworthiness verification and actual execution. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the structure of a remote debugging system for industrial robots based on edge computing according to the present invention;

[0053] Figure 2This is a schematic diagram of a remote debugging method for industrial robots based on edge computing according to the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0055] Please see Figure 1 As shown, this embodiment provides a remote debugging system for industrial robots based on edge computing, specifically including the following steps:

[0056] The edge status acquisition module is deployed at the edge node of the industrial robot to acquire the operating status information of the industrial robot, convert the operating status information into standardized status fragments, and parse the debugging instructions issued by the remote debugging terminal, breaking them down into structured instruction sequences.

[0057] The edge state reconstruction module establishes the spatiotemporal reconstruction results of the field operation state based on standardized state fragments and structured instruction sequences, and generates a dynamic conflict domain indicating conflict risk.

[0058] The action calibration module maps the structured instruction sequence into debugging actions according to time, performs segmented calibration of the debugging actions according to the dynamic conflict domain, and backfills the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence.

[0059] The edge-side verification module performs reversible verification on the calibration action sequence, generates forward execution simulation results and reverse playback simulation results, and calculates the difference between the simulation results; if the difference exceeds the preset difference threshold, the calibration action sequence is rolled back to the previous level calibration point for recalibration, and a reliable action sequence is output.

[0060] The remote collaborative scheduling module performs remote authorization, security locking, and multi-robot collaborative scheduling of trusted action sequences, outputs batch execution instructions, and sends them to the corresponding industrial robots.

[0061] Methods for converting runtime status information into standardized state fragments include:

[0062] The edge status acquisition module is deployed at the edge node of the industrial robot. The edge node includes different sensors to collect the operating status information of the industrial robot during operation. The sensors include position sensors, torque sensors, current sensors and environmental sensors.

[0063] Operational status information includes the industrial robot's position and orientation information, kinematic information, electrical drive status information, health status information, and environmental information;

[0064] Position and orientation information includes the angles of each joint, the position and orientation of the end effector; kinematic information includes the velocities of each joint, joint accelerations, end effector velocities and accelerations, joint torques and end effector torques; electrical drive status information includes the drive current, voltage, motor temperature and driver temperature of each joint; health status information includes sensor abnormal status information and power supply status information; environmental information includes the temperature and humidity of the environment in which the industrial robot is located.

[0065] The running status information is synchronized by network time protocol, and the time-synchronized running status information is normalized to map data with different dimensions and amplitude ranges to a unified numerical range, so as to obtain normalized running status information.

[0066] The normalized operating status information is divided into discrete state segments according to the preset sampling period; each discrete state segment is processed for format standardization and attribute identification, and a unique identifier is assigned to each discrete state segment, thereby forming a standardized state segment.

[0067] Methods for breaking down instructions into structured instruction sequences include:

[0068] The edge situation acquisition module is also used to receive debugging instructions issued by the remote debugging terminal, and to parse and decompose the debugging instructions. The debugging instructions are decomposed into basic action units, and each basic action unit is mapped to a standardized state segment in time or logic to ensure that each basic action unit can be executed accurately within the corresponding standardized state segment.

[0069] The mapped basic action units are sorted by execution order to form a structured instruction sequence. The structured instruction sequence includes different instruction units, each of which includes action type, target parameters, execution time, and other control attributes.

[0070] It should be noted that the action type indicates the action category, such as joint space motion, Cartesian space motion, grasping or releasing operation, tool operation, waiting or triggering action; the target parameter is used to define the execution target of the action, such as joint angle, end effector position and attitude, grasping state or task parameters; the execution time is used to indicate the start, end or time window of the action; other control attributes include speed and acceleration limits of action execution, action mode or execution method, triggering conditions or logical constraints, safety or protection parameters, and feedback or monitoring requirements for action execution.

[0071] Methods for establishing spatiotemporal reconstruction results of on-site operational status include:

[0072] The system receives standardized state fragments and structured action instruction sequences. It maps each basic action unit in the structured action instruction sequence to the corresponding standardized state data fragment according to a preset execution time window, establishes the correspondence between the basic action unit and the standardized state fragment in time and space, generates the predicted motion trajectory of the basic action unit, and ensures that the action can be accurately executed in the field, thereby establishing the spatiotemporal reconstruction result of the field operation state.

[0073] Methods for generating dynamic conflict domains that indicate conflict risk include:

[0074] Based on the predicted motion trajectory of the basic action unit, a motion trajectory envelope is constructed within the preset execution time window of the basic action unit. The motion trajectory envelope is as follows: ;in, Represents the basic action unit; Represents basic action unit The execution time window; Represents basic action unit At the point of time The pose includes joint position, end effector position, and attitude information; Indicates radius as The closed ball is used to represent the basic action unit at a given time point. The space occupied; Represents the Minkowski sum operation, used to convert action pose. Expanding into the spatial area that a basic motion unit may occupy (that is, "expanding" the point pose into spatial volume); Indicates time window The union of the expansion spaces at all points in time within the window forms the spatial extent that the action unit may occupy within the entire time window, which is the motion trajectory envelope. ;

[0075] The motion trajectory envelope intersects with a pre-defined set of environmental obstacles to generate a dynamic conflict domain; the dynamic conflict domain is: ;in, Represents the dynamic conflict domain, i.e., the basic action unit. The spatial region that may interfere with the pre-set set of environmental obstacles during execution; This represents the set of pre-defined environmental obstacles, which is all obstacles known before the industrial robot performs its actions, including static equipment, workbenches, walls, or other modeled fixed objects. This represents the set intersection operation, used to find the intersection of the motion trajectory envelope and the preset set of environmental obstacles, forming a dynamic conflict domain;

[0076] A conflict indication function is defined using a dynamic conflict domain to indicate the potential conflict risks that may occur during the execution of basic motion units; conflict risks include potential interference, joint limit triggering, or the risk of accumulated motion errors.

[0077] The conflict indication function is: ;in, Represents basic action unit A Boolean indication of whether a conflict exists; This indicates a risk of conflict (potential interference, joint limitation triggering, or risk of cumulative motion error). Indicates safety; Represents a dynamic conflict domain The volume, used to quantify the size of the conflict region, can be obtained through voxel discretization calculation;

[0078] This solution addresses the following technical problems of existing technologies: Existing technologies typically rely solely on the instruction sequence itself or simple motion simulation, failing to accurately determine whether robot actions will interfere with obstacles during execution under real-world environmental parameters, and also failing to identify action conflicts in advance during remote debugging. Traditional methods primarily detect single-point collisions (such as collision detection based on discrete sampling points), failing to characterize the spatial volume range that an action unit may occupy within the entire time window, and also failing to quantify the interaction between this spatial occupancy range and obstacles.

[0079] Current technologies lack a method to align the continuous spatial representation of motion trajectories with the geometric model of obstacles. Therefore, it is impossible to obtain the volume of the conflict area, let alone convert it into a clear risk indicator for remote debugging. Industrial sites typically lack high-performance computing capabilities, and existing debugging systems cannot complete the construction of motion space envelopes and Boolean intersection calculations in real time at the edge, thus failing to achieve fast and safe risk filtering at the equipment level.

[0080] Compared to existing technologies, the advantages are as follows: Explicitly modeling the spatial occupancy of basic motion units during continuous execution, and generating motion envelopes through Minkowski expansion, enables remote robot debugging to possess capabilities similar to real-time spatial prediction. Intersection operations with a pre-set set of obstacles can identify potential interference caused by trajectory deviations, posture changes, or error propagation in advance, thus significantly reducing the risks during offline programming and remote debugging. The conflict indication function enables instantaneous quantification of conflict risks, allowing the system to provide clear motion safety feedback early in the debugging process, improving the interpretability and robustness of robot motion planning. This method makes the remote debugging process more automated, efficient, and controllable, helping to reduce repeated trial-and-error time and improve the safety and accuracy of remote debugging.

[0081] Methods for mapping structured instruction sequences to debugging actions over time include:

[0082] Based on the semantic tags, action parameters, and instruction dependencies in the structured instruction sequence, the start and end times of the execution of each basic action unit are determined, and each basic action unit is mapped to a unified global time axis.

[0083] It should be noted that when parsing structured instruction sequences, semantic tags are first used to identify the action type corresponding to the instruction. Different action types imply different time determinants. For example, semantic tags such as heating, cooling, holding, moving, positioning, and gripping specify the time generation method of the action. The system combines the time model corresponding to the semantic tags with the action parameters given in the instruction, such as target temperature, moving distance, speed, holding duration, acceleration and deceleration parameters, to calculate the theoretical execution time of each basic action unit.

[0084] Simultaneously, the system further analyzes the dependencies in the instruction sequence. These dependencies indicate whether a certain action must be completed before it can begin, or whether it can be performed in parallel with other actions. Based on these dependencies, the system constructs sequential constraints between actions, ensuring that the earliest start time of a given action equals the end time or condition trigger time of all its preceding actions. Ultimately, the start and end times of each basic action unit are determined by these two dimensions: the start time is determined by the dependencies, and the end time is the sum of the start time and the execution duration calculated from semantic tags and action parameters, thus forming a deterministic time positioning.

[0085] The basic motion units are sorted according to the global timeline to generate a motion scheduling sequence. Based on the preset execution time window of each basic motion unit, the motion requirements in the motion scheduling sequence are transformed into debugging actions to be performed by the industrial robot. The debugging actions include target pose, joint angle changes, velocity and acceleration constraints.

[0086] It should be noted that the instruction unit is parsed into a joint angle sequence, and a continuous executable trajectory is generated using interpolation and trajectory planning algorithms. Safety verification is then performed by combining dynamic conflict domain, joint limit, and velocity acceleration constraints, thereby forming debugging actions that can be directly recognized and executed by the industrial robot remote debugging system.

[0087] Methods for generating calibration action sequences include:

[0088] The continuous motion trajectory of the debugging action within the preset execution time window is discretized to obtain different time slice trajectory segments; based on the dynamic conflict domain, it is detected whether each time slice trajectory segment intersects with the dynamic conflict domain;

[0089] It should be noted that the method for obtaining different time-slice trajectory segments includes: acquiring the continuous trajectory of the industrial robot within the time window, including the joint angle sequence that changes continuously over time, the end effector pose sequence, and the corresponding velocity and acceleration curves. Then, based on a preset time step, the entire time window is divided into multiple equally spaced discrete time nodes. The trajectory segment between every two adjacent discrete time nodes is defined as a time-slice trajectory segment.

[0090] When a time slice trajectory segment enters the dynamic conflict domain, the control parameters of the time slice trajectory segment are adjusted. The adjustment includes modifying the path offset, the joint angle redundancy allocation method, or the end effector attitude, so that the corrected time slice trajectory segment avoids the dynamic conflict domain, thereby forming a conflict-free calibration time slice trajectory segment.

[0091] The control parameters of the time slice trajectory segment include joint angle commands describing the motion of each joint, joint angular velocity, joint angular acceleration, pose parameters describing the motion of the end effector (including position and attitude angle), end effector velocity, end effector acceleration, redundancy allocation parameters for handling redundancy (e.g., the redundancy joint angle selection strategy for a multi-joint robotic arm under a given end effector pose), path offset for collision avoidance, and time scaling factor for smoothing the execution action.

[0092] Adjustable parameters generated during calibration in the calibration time slice trajectory segments are backfilled according to a preset priority. The preset priority is used to determine the parameter recovery order under the condition of balancing action safety, control stability and equipment dynamic performance requirements. The calibration time slice trajectory segments with completed parameter backfilling are recombined according to the time order of the basic action units to generate a calibration action sequence with definite motion parameters, no interference and can be executed by the industrial robot.

[0093] Methods for outputting reliable action sequences include:

[0094] A forward execution simulation is constructed based on the calibration action sequence to simulate and deduce the execution process of the industrial robot from the initial state to the preset target state, and the forward execution simulation results are obtained. Based on the forward execution simulation results, a corresponding reverse playback sequence is constructed, and the reverse execution process is simulated and deduced to obtain the reverse playback simulation results.

[0095] It should be noted that the system takes the motion commands, target pose parameters, and corresponding time series in the calibration action sequence as input, and calls upon the known forward kinematics model and dynamic constraint model of the industrial robot to continuously simulate the entire process of the robot gradually transitioning from the initial state to the preset target state. Specifically, the system maps the calibration commands in the joint space to the end effector pose trajectory in Cartesian space through forward kinematic equations, and calculates the state evolution of the robot throughout the calibration process by combining velocity constraints, acceleration constraints, and joint torque boundary conditions, obtaining forward execution simulation results including joint angles, joint velocities, end effector pose, and dynamic components at each time step.

[0096] After obtaining the forward execution simulation results, the system further constructs a one-to-one corresponding reverse playback sequence based on the forward trajectory. The reverse playback sequence is achieved by reversing the state sequence of the forward execution simulation results in time and performing inverse kinematics solutions on the state variables at each moment. This ensures that the joint commands and end-effector pose trajectories during reverse playback are theoretically precisely reversible with the forward process. Then, using the reverse playback sequence as input, the system calls the robot's kinematics and dynamics model to reverse-engineer the process of the robot gradually regressing from a preset target state to the initial state, obtaining the reverse playback simulation results.

[0097] Differential calculations are performed on the forward execution simulation results and the reverse playback simulation results to obtain the differences in key poses, joint angles, and control parameters. When the difference does not exceed the preset differential threshold, the calibration action sequence is determined as a reliable action sequence. When the difference exceeds the preset differential threshold, the calibration action sequence is rolled back to the previous level calibration point and re-calibrated until a reliable action sequence that meets the preset differential threshold requirement (i.e., the difference does not exceed the preset differential threshold) is obtained.

[0098] Methods for outputting batch execution instructions and sending them to the corresponding industrial robots include:

[0099] Based on the action set of each industrial robot in the reliable action sequence, the reachability domain of the corresponding industrial robot is constructed. The reachability domain is obtained by kinematic mapping from the joint feasible domain of the industrial robot.

[0100] The reachable domain is: ;in, Industrial robots The reachable domain; Industrial robots Joint configuration (joint angles or set of joint displacements); Industrial robots The joint feasible domain (the set of all joint configurations allowed by mechanical structure and safety constraints); Industrial robots The positive kinematic mapping function is used to map joint configurations to the spatial pose of the end effector;

[0101] Perform spatial intersection operation on the reachable domains of any two industrial robots to obtain the interaction interference zone of the two industrial robots in the work space, and define the interference weight for the interaction interference zone;

[0102] The interactive interference region is: ;in, Industrial robots With industrial robots The intersection of the interactive interference regions, i.e., the reachable domains; Industrial robots The reachable domain; and Indicates the serial number of the industrial robot;

[0103] An interference graph is constructed based on the interference weights among all industrial robots. Nodes in the interference graph correspond to industrial robots, and the weights of the edges correspond to the interference weights. Using the spatial conflict factor and task attribute factor in the interference graph, the scheduling priority of each industrial robot is calculated according to the preset trade-off coefficients.

[0104] The scheduling priority is: ;in, Industrial robots Scheduling priority; This represents a tradeoff coefficient, ranging from [0,1], used to adjust the weight between the spatial conflict factor and the task attribute factor; This represents the interference weight, used to quantify the degree of conflict between two industrial robots in space; Indicates the spatial conflict factor; Indicates task attribute factors;

[0105] The spatial conflict factor is the sum of the interference weights of each node, and the task attribute factor is the task attribute parameter of each industrial robot. The task attribute factor is the comprehensive metric value after weighted fusion of the task attribute parameters of each industrial robot. The task attribute parameters include task urgency, safety sensitivity, task coupling, quality criticality, and resource occupancy.

[0106] It should be noted that the task attribute parameters are derived from the task level information of the robot's current task. During the task initialization phase, the system presets task attribute tags for each task to describe its urgency, risk level, the impact of task failure, and its dependence on equipment or processes. For example, high-risk actions (such as working close to people, grasping fragile parts, or contacting high-temperature or high-pressure equipment), actions with clear time constraints, or critical actions that would cause the system to stop if failed are marked with higher-level task attribute tags. The system assigns corresponding weights to each attribute based on these task attribute tags; the higher the urgency, the greater the weight; the higher the risk level, the greater the weight; and the more complex the task coupling or the stricter the quality requirements, the greater the corresponding weight.

[0107] Based on scheduling priority, time windows are allocated and batches are divided for each industrial robot action in the trusted action sequence to generate execution batch instructions. After generating the execution batch instructions, remote authorization information and security lock parameters are attached to each batch instruction, and the authorized execution batch instructions are sent to the corresponding industrial robots respectively.

[0108] This solution addresses the following technical problems in existing technologies: First, existing technologies lack a refined and quantifiable way of expressing spatial conflict relationships between multiple robots, failing to identify the degree of dynamic interference at the real reachability level. This results in coarse scheduling strategies, prone to false avoidance or insufficient avoidance. Second, existing technologies often prioritize tasks solely based on task sequence or operation type, leading to inefficient resource allocation during multi-robot collaborative execution. This can result in high-priority tasks being delayed or high-risk tasks lacking safety boundaries. Third, existing technologies lack a safe closed-loop mechanism between trusted action sequences and actual execution instructions. They typically issue instructions directly according to task sequence without remote authorization or dynamic safety locking strategies, increasing the risk of false triggering, unauthorized execution, or cross-robot interference during remote debugging. Fourth, existing technologies lack a batch-level scheduling mechanism, failing to intelligently divide multi-robot actions into parallel or sequential execution batches based on spatial conflicts and task attributes, resulting in low efficiency in multi-robot collaboration.

[0109] The preset difference threshold is set by the staff. By collecting the differences of different simulation results, the average of multiple differences is taken as the preset difference threshold.

[0110] In this embodiment, by explicitly modeling the spatial occupancy of basic motion units during continuous execution and generating motion envelopes through Minkowski expansion, remote robot debugging acquires capabilities similar to real-time spatial prediction. Intersection operations with a pre-set set of obstacles can identify potential interference caused by trajectory deviations, posture changes, or error propagation in advance, significantly reducing risks during offline programming and remote debugging. A conflict indication function enables instantaneous quantification of conflict risks, allowing the system to provide clear motion safety feedback early in debugging, improving the interpretability and robustness of robot motion planning. This method makes the remote debugging process more automated, efficient, and controllable, helping to reduce repeated trial-and-error time and improve the safety and accuracy of remote debugging.

[0111] By mapping reachable domains based on trusted action sequences, calculating interaction interference zones, and constructing interference relationship graphs, a quantifiable and computable expression of spatial conflict relationships among multiple robots is achieved. By fusing spatial conflict factors with task attribute factors and using trade-off coefficients to calculate scheduling priorities, the scheduling process simultaneously considers workspace safety and task importance. Through time window allocation and batching of subtasks based on scheduling priorities, the system can automatically generate efficient execution batch instructions, enabling automated orchestration of parallel and sequential actions during multi-robot operations, thus improving collaborative efficiency. By attaching safety locking parameters based on behavioral trustworthiness to execution batches, closed-loop control is achieved between simulation trustworthiness verification and actual execution. Example

[0112] Please see Figure 2 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. This embodiment provides a remote debugging method for industrial robots based on edge computing, including:

[0113] S1. Deployed at the edge node of the industrial robot, it acquires the operating status information of the industrial robot, converts the operating status information into standardized status fragments, and parses the debugging instructions issued by the remote debugging terminal, breaking them down into structured instruction sequences.

[0114] S2. Based on standardized state fragments and structured instruction sequences, establish the spatiotemporal reconstruction results of the field operation status and generate a dynamic conflict domain indicating conflict risk.

[0115] S3. Map the structured instruction sequence to debugging actions according to time, perform segmented calibration of the debugging actions according to the dynamic conflict domain, and backfill the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence.

[0116] S4. Perform reversible verification on the calibration action sequence, generate forward execution simulation results and reverse playback simulation results, and calculate the difference between the simulation results; if the difference exceeds the preset difference threshold, roll back the calibration action sequence to the previous level calibration point for recalibration, and output a reliable action sequence.

[0117] S5. Perform remote authorization, security locking, and multi-robot collaborative scheduling on trusted action sequences, output batch execution instructions, and send them to the corresponding industrial robots.

[0118] Since the electronic device described in this embodiment is the electronic device used to implement the remote debugging system and method for industrial robots based on edge computing in the embodiments of this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the remote debugging system and method for industrial robots based on edge computing described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the remote debugging system and method for industrial robots based on edge computing in the embodiments of this application falls within the scope of protection of this application.

[0119] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0120] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An edge-computing-based remote commissioning system for industrial robots, characterized in that, include: The edge status acquisition module is deployed at the edge node of the industrial robot to acquire the operating status information of the industrial robot and convert the operating status information into standardized status fragments. It also parses the debugging commands issued by the remote debugging terminal and breaks them down into structured command sequences; The edge state reconstruction module establishes the spatiotemporal reconstruction results of the field operation state based on standardized state fragments and structured instruction sequences, and generates a dynamic conflict domain indicating conflict risk. Methods for generating dynamic conflict domains that indicate conflict risk include: Based on the predicted motion trajectory of the basic action unit, a motion trajectory envelope is constructed within the preset execution time window of the basic action unit. The motion trajectory envelope is as follows: ;in, Represents the basic action unit; Represents basic action unit The execution time window; Represents basic action unit At the point of time The pose includes joint position, end effector position, and attitude information; Indicates radius as The closed ball is used to represent the basic action unit at a given time point. The space occupied; Represents the Minkowski sum operation, used to convert action pose. Expanding into the spatial region that a basic motion unit may occupy, that is, expanding the point pose into a spatial volume; Indicates time window The union of the expansion spaces at all points in time within the window forms the spatial extent that the action unit may occupy within the entire time window, which is the motion trajectory envelope. ; The motion trajectory envelope intersects with a pre-defined set of environmental obstacles to generate a dynamic conflict domain; the dynamic conflict domain is: ;in, Represents the dynamic conflict domain, i.e., the basic action unit. The spatial region that may interfere with the pre-set set of environmental obstacles during execution; This represents the set of pre-defined environmental obstacles, which is all obstacles known before the industrial robot performs its actions, including static equipment, workbenches, walls, or other modeled fixed objects. This represents the set intersection operation, used to find the intersection of the motion trajectory envelope and the preset set of environmental obstacles, forming a dynamic conflict domain; The action calibration module maps the structured instruction sequence into debugging actions according to time, performs segmented calibration of the debugging actions according to the dynamic conflict domain, adjusts the control parameters of the time slice trajectory segment when the time slice trajectory segment enters the dynamic conflict domain, and backfills the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence. The edge-side verification module performs reversible verification on the calibration action sequence, generates forward execution simulation results and reverse playback simulation results, and calculates the difference between the simulation results; if the difference exceeds the preset difference threshold, the calibration action sequence is rolled back to the previous level calibration point for recalibration, and a reliable action sequence is output. The remote collaborative scheduling module performs remote authorization, security locking, and multi-robot collaborative scheduling of trusted action sequences, outputs batch execution instructions, and sends them to the corresponding industrial robots.

2. The edge computing-based remote debugging system for industrial robots according to claim 1, characterized in that, The method for converting operational status information into standardized status fragments includes: The edge status acquisition module is deployed at the edge node of the industrial robot. The edge node includes various sensors used to collect the operating status information of the industrial robot during operation. The sensors include position sensors, torque sensors, current sensors, and environmental sensors. The operating status information includes the position and posture information, kinematic information, electrical drive status information, health status information, and environmental information of the industrial robot. The running status information is synchronized by network time protocol, and the time-synchronized running status information is normalized to map data with different dimensions and amplitude ranges to a unified numerical range, so as to obtain normalized running status information. The normalized operating status information is divided into discrete state segments according to the preset sampling period; each discrete state segment is processed for format standardization and attribute identification, and a unique identifier is assigned to each discrete state segment, thereby forming a standardized state segment. 3.The edge computing based industrial robot remote debugging system according to claim 2, characterized in that, The method for decomposing into a structured instruction sequence includes: The edge situation acquisition module is also used to receive debugging instructions issued by the remote debugging terminal, and to parse and decompose the debugging instructions, decompose the debugging instructions into basic action units, and map each basic action unit to a standardized state segment in time or logic. The mapped basic action units are sorted by execution order to form a structured instruction sequence. The structured instruction sequence includes different instruction units, each of which includes action type, target parameters, execution time, and other control attributes.

4. The edge computing-based remote debugging system for industrial robots according to claim 3, characterized in that, The method for establishing the spatiotemporal reconstruction results of the on-site operating status includes: The system receives standardized state segments and structured action instruction sequences. It maps each basic action unit in the structured action instruction sequence to the corresponding standardized state data segment according to a preset execution time window, establishes the temporal and spatial correspondence between the basic action unit and the standardized state segment, generates the predicted motion trajectory of the basic action unit, and thus establishes the spatiotemporal reconstruction result of the on-site operation state.

5. The edge computing based remote commissioning system of industrial robots according to claim 4, characterized in that, The method for generating a dynamic conflict domain that indicates conflict risk includes: Based on the predicted motion trajectory of the basic motion unit, a motion trajectory envelope is constructed within the preset execution time window of the basic motion unit. The motion trajectory envelope is intersected with the preset set of environmental obstacles to generate a dynamic conflict domain. A conflict indication function is defined using a dynamic conflict domain to indicate the potential conflict risks that may occur during the execution of basic motion units; conflict risks include potential interference, joint limit triggering, or the risk of accumulated motion errors.

6. The edge computing based remote commissioning system of industrial robots according to claim 5, characterized in that, The method for mapping structured instruction sequences to debugging actions over time includes: Based on the semantic tags, action parameters, and instruction dependencies in the structured instruction sequence, the start and end times of the execution of each basic action unit are determined, and each basic action unit is mapped to a unified global time axis. The basic motion units are sorted according to the global timeline to generate a motion scheduling sequence. Based on the preset execution time window of each basic motion unit, the motion requirements in the motion scheduling sequence are transformed into debugging actions to be performed by the industrial robot.

7. The remote debugging system for industrial robots based on edge computing according to claim 6, characterized in that, The method for forming the calibration action sequence includes: The continuous motion trajectory of the debugging action within the preset execution time window is discretized to obtain different time slice trajectory segments; based on the dynamic conflict domain, it is detected whether each time slice trajectory segment intersects with the dynamic conflict domain; Adjustments include modifying path offsets, joint angle redundancy allocation methods, or end effector attitudes to ensure that the corrected time slice trajectory segments avoid dynamic conflict domains, thereby forming conflict-free calibration time slice trajectory segments. Adjustable parameters generated during calibration in the calibration time slice trajectory segment are backfilled according to a preset priority. Each calibration time slice trajectory segment with completed parameter backfilling is then recombined according to the time sequence of the basic motion units to generate a calibration motion sequence with definite motion parameters, no interference, and executable by the industrial robot.

8. The edge computing based remote commissioning system of industrial robots according to claim 7, characterized in that, The method for outputting a reliable action sequence includes: A forward execution simulation is constructed based on the calibration action sequence to simulate and deduce the execution process of the industrial robot from the initial state to the preset target state, and the forward execution simulation results are obtained. Based on the forward execution simulation results, a corresponding reverse playback sequence is constructed, and the reverse execution process is simulated and deduced to obtain the reverse playback simulation results. Differential calculations are performed on the forward execution simulation results and the reverse playback simulation results to obtain the differences in key poses, joint angles, and control parameters. When the difference does not exceed the preset difference threshold, the calibration action sequence is determined as a reliable action sequence. When the difference exceeds the preset difference threshold, the calibration action sequence is rolled back to the previous level calibration point and re-calibrated until a reliable action sequence that meets the preset difference threshold requirement is obtained.

9. The edge computing based remote commissioning system of industrial robots according to claim 8, characterized in that, The method for outputting batch instructions and sending them to the corresponding industrial robots includes: Based on the action sets of each industrial robot in the trusted action sequence, the reachable domain of the corresponding industrial robot is constructed. The reachable domain is obtained by kinematic mapping from the joint feasible domain of the industrial robot. Spatial intersection operation is performed on the reachable domains of any two industrial robots to obtain the interaction interference region of the two industrial robots in the work space, and interference weights are defined for the interaction interference region. An interference graph is constructed based on the interference weights among all industrial robots. Nodes in the interference graph correspond to industrial robots, and the weights of the edges correspond to the interference weights. Using the spatial conflict factor and task attribute factor in the interference graph, the scheduling priority of each industrial robot is calculated according to the preset trade-off coefficients. The spatial conflict factor is the sum of the interference weights of each node, and the task attribute factor is the task attribute parameter of each industrial robot. The task attribute factor is the comprehensive metric value after weighted fusion of the task attribute parameters of each industrial robot. The task attribute parameters include task urgency, safety sensitivity, task coupling, quality criticality, and resource occupancy. Based on scheduling priority, time windows are allocated and batches are divided for each industrial robot action in the trusted action sequence to generate execution batch instructions. After generating the execution batch instructions, remote authorization information and security lock parameters are attached to each batch instruction, and the authorized execution batch instructions are sent to the corresponding industrial robots respectively.

10. An edge computing-based remote debugging method for industrial robots, implemented by the edge computing-based remote debugging system for industrial robots according to any one of claims 1 to 9, characterized in that, include: S1. Deployed at the edge nodes of industrial robots to acquire the operating status information of industrial robots and convert the operating status information into standardized state fragments; It also parses the debugging commands issued by the remote debugging terminal and breaks them down into structured command sequences; S2. Based on standardized state fragments and structured instruction sequences, establish the spatiotemporal reconstruction results of the field operation status and generate a dynamic conflict domain indicating conflict risk. S3. Map the structured instruction sequence to debugging actions according to time, perform segmented calibration of the debugging actions according to the dynamic conflict domain, and backfill the parameters of the calibrated debugging actions according to the preset priority to form a calibration action sequence. S4. Perform reversible verification on the calibration action sequence, generate forward execution simulation results and reverse playback simulation results, and calculate the difference between the simulation results; if the difference exceeds the preset difference threshold, roll back the calibration action sequence to the previous level calibration point for recalibration, and output a reliable action sequence. S5. Perform remote authorization, security locking, and multi-robot collaborative scheduling on trusted action sequences, output batch execution instructions, and send them to the corresponding industrial robots.

Images