Industrial robot multi-machine collaborative operation trajectory accurate planning method

Abstract

This invention relates to the field of industrial robot trajectory planning and discloses a method for accurate trajectory planning of multi-robot collaborative operations. The method includes: acquiring parameters, static environment information, and overall collaborative task requirements of multiple industrial robots to be collaborated on; constructing a unified world coordinate system; and building a digital twin scenario model for multi-robot collaborative operations. Subsequently, it identifies strongly coupled collaborative operation nodes and calibrates the strongly coupled constraint system. Then, it generates an initial continuous trajectory in Cartesian space, identifies abnormal trajectory segments, and completes trajectory optimization. After digital twin simulation verification and local replanning, the trajectory is synchronously distributed and closed-loop operation control is executed. This invention solves the problems of inconsistent multi-robot collaborative benchmarks, poor synchronization and high collision risk due to decoupling of strongly coupled constraints, poor trajectory continuity and incomplete conflict verification, inability to achieve global multi-objective optimization of multiple robots, lack of full-process closed-loop control, and poor versatility of the solution in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of industrial robot trajectory planning, and in particular to a method for accurate trajectory planning of multi-robot collaborative operations. Background Technology

[0002] With the rapid development of intelligent manufacturing and flexible production, a single industrial robot can no longer meet the operational needs of complex processes, large-sized workpieces, and high-efficiency production. Multi-robot collaborative operation mode has become the core development direction of industrial robot application due to its advantages such as wide operating range and strong task adaptability. As the core underlying technology of multi-robot collaborative operation, trajectory planning directly determines the accuracy, efficiency, safety and stability of collaborative operation.

[0003] Currently, conventional multi-machine collaborative trajectory planning in this field generally adopts the technical approach of "task decoupling - single-machine planning - splicing and integration," which has several inherent defects. Firstly, the multi-machine collaborative benchmark is not unified. Existing solutions mostly rely on the individual robot's own base coordinate system for single-machine trajectory planning, failing to construct a unified world coordinate system to complete global registration between multiple machines and the work scene. This results in misalignment of the spatial pose benchmarks of multiple machines, failing to meet the requirements of high-precision collaborative assembly, welding, and other high-end operations. Secondly, the handling of coupling constraints is insufficient. Conventional solutions decompose the strong coupling constraints of multi-machine collaboration into multiple independent constraints for individual machines, only setting fixed safe obstacle avoidance distances and single-machine motion limit constraints. They do not fully consider the strong correlation constraints of multi-machine temporal and pose synchronization. Under dynamic conditions, either conservative settings reduce work efficiency, or excessively close distances pose collision safety hazards.

[0004] In terms of trajectory generation and optimization, existing technologies have significant shortcomings. Conventional solutions can only generate trajectories that meet the G1 continuous standard. Sudden curvature changes can easily cause robot joint vibration and excessive impact, which reduces operational accuracy and exacerbates equipment wear. Conflict verification only performs static collision detection at discrete time points of the trajectory and does not conduct time-space dual-dimensional matching verification throughout the entire operation cycle, making it easy to miss the risk of cross-collision during the dynamic movement of multiple machines. At the same time, it cannot achieve multi-machine global multi-objective collaborative optimization, and can only complete local optimization of single machine single object. It is difficult to balance operation efficiency, operational stability and equipment energy consumption. The optimization results are highly random and have poor engineering feasibility.

[0005] Furthermore, existing solutions generally lack a closed-loop management mechanism for the entire process, often employing an open-loop planning model. Once trajectory planning is complete, execution is immediately initiated without full-process virtual simulation pre-verification, making it impossible to identify trajectory failures under dynamic operating conditions in advance. Even when trajectory correction mechanisms are implemented, they only allow for independent replanning of a single robot, failing to consider the impact of trajectory adjustments on other collaborative robots, easily leading to new collision risks and extremely low operational fault tolerance. Simultaneously, existing solutions are mostly customized developments, lacking versatility and adaptability. They cannot quickly adapt to different robot models and quantities, nor to the changing demands of flexible production, making it difficult to meet the core performance requirements of high-end intelligent manufacturing for multi-robot collaborative operations. Summary of the Invention

[0006] The present invention aims to provide a method for accurate trajectory planning of multi-machine collaborative operation of industrial robots, in order to solve the problems of inconsistent multi-machine collaborative benchmarks, poor synchronization and high collision risk caused by decoupling of strong coupling constraints, poor trajectory continuity and incomplete conflict verification, inability to achieve global multi-objective optimization of multi-machines, lack of closed-loop management of the whole process, and poor versatility of the solution in the existing technology.

[0007] To achieve the above objectives, the present invention provides the following method:

[0008] This invention provides a method for accurate trajectory planning in multi-robot collaborative operations of industrial robots:

[0009] S1: Obtain the kinematic and dynamic parameters of multiple industrial robots to be coordinated, the global static environment information of the work scene, and the total task requirements of multi-robot collaborative operation; construct a unified world coordinate system; complete the global registration of the multiple industrial robots to be coordinated and the work scene in the unified world coordinate system; and construct a digital twin scene model of multi-robot collaborative operation based on the registration results.

[0010] S2: Based on the unified world coordinate system and the overall task requirements of the multi-machine collaborative operation, the overall task requirements of the multi-machine collaborative operation are decomposed into sub-tasks corresponding to the multiple industrial robots to be collaborated. The strong coupling constraint system of multi-machine collaboration is simultaneously calibrated. The strong coupling constraint system of multi-machine collaboration includes the safety obstacle avoidance distance threshold between multiple machines, the timing matching window of synchronous operation, the end pose synchronization deviation threshold of collaborative operation, and the body motion limit constraints of each industrial robot.

[0011] S3: In the multi-machine collaborative operation digital twin scenario model, for each sub-task of the multiple industrial robots to be collaborated, the Cartesian space initial continuous trajectory of each industrial robot corresponding to the sub-task is generated by combining the kinematic and dynamic parameters of the corresponding industrial robot and the multi-machine collaborative strong coupling constraint system; based on the unified world coordinate system, the Cartesian space initial continuous trajectory is subjected to full-time spatiotemporal matching verification to identify abnormal trajectory segments that do not conform to the multi-machine collaborative strong coupling constraint system;

[0012] S4: With the optimization objectives of minimizing total operation time, optimizing trajectory smoothness, and minimizing joint drive energy consumption, a multi-machine global collaborative multi-objective optimization function is constructed. The multi-machine collaborative strong coupling constraint system calibrated in step S2 is used as the boundary condition, and the abnormal trajectory segment identified in step S3 is used as the core optimization interval. The initial continuous trajectory in the Cartesian space is then solved through global collaborative optimization to obtain the optimized pre-execution trajectory of each industrial robot.

[0013] S5: Import the optimized pre-execution trajectory into the multi-machine collaborative operation digital twin scenario model, carry out full-process closed-loop virtual simulation verification, simulate abnormal working conditions during the operation, identify problem trajectory segments that do not conform to the multi-machine collaborative strong coupling constraint system during the simulation, perform local dynamic replanning for the problem trajectory segments, and generate the final multi-machine collaborative operation executable trajectory.

[0014] S6: The final multi-machine collaborative operation executable trajectory is synchronously sent to the controller of the corresponding industrial robot according to the timing matching window of the synchronous operation calibrated in step S2, driving each industrial robot to perform collaborative operation; during the operation, the actual operation data of each industrial robot is collected in real time and the deviation is compared with the final multi-machine collaborative operation executable trajectory. When the deviation exceeds the preset threshold, local trajectory replanning is triggered to ensure the synchronization of multi-machine collaborative operation.

[0015] Further, in step S1, the base coordinate system pose data of the multiple industrial robots to be coordinated is acquired, and the global registration of the multiple industrial robots to be coordinated with the work scene is completed in the unified world coordinate system based on the base coordinate system pose data. The constructed multi-robot collaborative operation digital twin scene model is a dynamic mapping model that can map the operating pose of the industrial robots, changes in the work scene environment, and the progress of the work task in real time. The multi-robot collaborative operation digital twin scene model includes an industrial robot body twin unit, a work scene environment twin unit, a collaborative task mapping unit, and a simulation verification unit. The industrial robot body twin unit is used to map the kinematics and dynamic characteristics of the multiple industrial robots to be coordinated. The work scene environment twin unit is used to map the global static environment and dynamic changes of the work scene. The collaborative task mapping unit is used to map the total task requirements and the decomposed sub-tasks of the multi-robot collaborative operation. The simulation verification unit is used to realize the full-process virtual simulation of the trajectory and the simulation of abnormal working conditions.

[0016] Further, step S2 specifically involves: based on the unified world coordinate system and the overall task requirements of the multi-machine collaborative operation, identifying strongly coupled operation nodes of the multi-machine collaborative operation, wherein the strongly coupled operation nodes are operation process nodes that require multiple industrial robots to cooperate synchronously to complete the operation and have strong spatial and temporal correlations; and decomposing the overall task requirements of the multi-machine collaborative operation into operation sub-tasks that correspond one-to-one with the multiple industrial robots to be collaborated on and are bound to the strongly coupled operation nodes.

[0017] Furthermore, the multi-machine collaborative strong coupling constraint system calibrated in step S2 also includes: joint torque limit constraints of each industrial robot and curvature continuity constraints of the Cartesian space trajectory; the safe obstacle avoidance distance threshold between the multiple machines is a dynamic safe obstacle avoidance distance threshold that is adjusted in real time with the running speed of the industrial robot, and the dynamic safe obstacle avoidance distance threshold is positively correlated with the real-time running speed of the industrial robot. The faster the running speed, the larger the corresponding safe obstacle avoidance distance threshold.

[0018] ;

[0019] In the formula, This is the initial value for the dynamic safe obstacle avoidance distance. Based on the safe distance, For the robot's rated speed, For the density of obstacles in the scene, This is the accuracy coefficient for the operation. , , These are the weighting coefficients corresponding to speed, obstacle density, and operation accuracy, respectively, which are obtained by training a constrained self-learning optimization model.

[0020] Furthermore, in step S3, the generated Cartesian space initial continuous trajectory is a Cartesian space G2 continuous trajectory, which is a trajectory with continuous curvature throughout without abrupt changes.

[0021] Further, the full-time spatiotemporal matching verification in step S3 involves performing spatiotemporal dual-dimensional matching verification at any time node on the continuous time dimension of the entire operation cycle for the continuous trajectory of the Cartesian space G2 of all industrial robots under the unified world coordinate system. The content of the spatiotemporal dual-dimensional matching verification is: verifying the operation timing matching degree of each industrial robot in the time dimension, and verifying the relative pose and safe distance between multiple robots in the spatial dimension. The abnormal trajectory segments that do not conform to the strong coupling constraint system of multi-robot collaboration are trajectory segments where the spatial distance between multiple robots is less than the safe obstacle avoidance distance threshold, the timing deviation exceeds the timing matching window of synchronous operation, and the pose deviation exceeds the end pose synchronization deviation threshold of collaborative operation at any time node.

[0022] Further, in step S4, the constrained non-dominated sorting genetic algorithm NSGA-III is used to complete the global collaborative optimization solution. The multi-machine collaborative strong coupling constraint system calibrated in step S2 is used as the rigid constraint condition of the algorithm, and the spatiotemporal parameters of the initial continuous trajectory in the Cartesian space are used as the optimization decision variables to obtain the Pareto optimal solution set. The shortest total operation time is the first priority, the optimal trajectory smoothness is the second priority, and the lowest joint drive energy consumption is the third priority. The optimal solution is selected from the Pareto optimal solution set according to the progressive rule. The progressive rule is as follows: first, the solution set subset with the best first priority index is selected, then the secondary solution set subset with the best second priority index is selected from the solution set subset, and finally the global optimal solution with the best third priority index is selected from the secondary solution set subset. Based on the global optimal solution, the optimized pre-execution trajectory of each industrial robot is generated.

[0023] Furthermore, the abnormal working conditions simulated in step S5 include dynamic obstacle intrusion, robot load fluctuation, and joint torque exceeding limits. During the simulation, the operating posture, timing, and joint torque data of each industrial robot are collected at all times. Based on the multi-machine collaborative strong coupling constraint system, the local dynamic replanning of the problematic trajectory segment is completed. The local dynamic replanning is as follows: with the multi-machine collaborative strong coupling constraint system as the rigid boundary and smooth trajectory connection as the core requirement, trajectory replanning is only performed on the time interval corresponding to the problematic trajectory segment and the preset transition intervals before and after it. The non-associated trajectory segments of the unconstrained exceeding limit problem are not changed. At the same time, dynamic adjustments are made to adapt to the real-time abnormal working conditions in the simulation process. After the local dynamic replanning is completed, the full-process closed-loop virtual simulation verification is carried out again until there are no problems in the entire simulation process that do not conform to the multi-machine collaborative strong coupling constraint system. Then, the final multi-machine collaborative operation executable trajectory is generated.

[0024] Furthermore, the actual operating data of each industrial robot collected in real time in step S6 includes the actual operating posture data, joint torque data, and actual operation progress data of each industrial robot; when comparing the deviation with the final multi-robot collaborative operation executable trajectory, a multi-dimensional comparison of posture deviation, timing deviation, and torque deviation is carried out simultaneously; the multi-dimensional comparison is judged using a standardized deviation rate, which is calculated according to the following formula:

[0025] ;

[0026] In the formula, The deviation rate, This is the actual deviation value. The constraint threshold is the corresponding dimension; the time deviation is the deviation between the actual operation progress of the industrial robot and the preset operation progress of the final multi-robot collaborative operation executable trajectory.

[0027] Furthermore, in step S6, when the deviation exceeds a preset threshold, a local trajectory replanning for multi-machine global collaboration is triggered based on the multi-machine collaborative strong coupling constraint system. The local trajectory replanning refers to replanning the trajectory only for the trajectory interval corresponding to the deviation exceeding the limit and the preset transition intervals before and after, with the multi-machine collaborative strong coupling constraint system as the rigid boundary and multi-machine collaborative synchronization as the core requirement. It does not change the non-associated trajectory segments without deviation problems, and at the same time, it adapts to the real-time operation data of the industrial robots to complete dynamic adjustments. The operation trajectories of all associated industrial robots are updated synchronously. The associated industrial robots are all industrial robots that have strong coupling operation nodes with the industrial robot with the deviation exceeding the limit, ensuring that the entire process of multi-machine collaborative operation complies with the requirements of the multi-machine collaborative strong coupling constraint system.

[0028] The beneficial effects of this invention are as follows: By constructing a unified world coordinate system and building a supporting multi-machine collaborative digital twin scene model, this invention solves the problem of misalignment of multi-machine collaborative benchmarks, providing a precise virtual-real mapping carrier for the entire trajectory planning process; it breaks through the conventional approach of "decoupling and dimensionality reduction" in existing technologies, taking multi-machine global coupling collaboration as the core, and through strong coupling of task binding of operation nodes, construction of a full-dimensional strong coupling constraint system, and constraint adaptive calibration iteration mechanism, it solves the pain points of poor synchronization and high collision risk caused by decoupling processing in existing technologies, balancing operation efficiency and operational safety; and it generates a G2 continuous trajectory avoidance mechanism. The robot's joint vibration impact, combined with all-time spatiotemporal dual-dimensional matching and verification, solves the collision omission problem of existing discrete point static detection technology; by constructing a multi-machine global multi-objective optimization function, the optimization solution is completed by using the constrained NSGA-III algorithm, which breaks through the limitation of single-machine local optima in existing technology; a new enhanced spatiotemporal linkage and collaborative compensation mechanism for coupled operation nodes fills the technical gap of existing technology that can only replan after the fact, reducing the frequency of replanning triggering; finally, a closed-loop control system for the entire process is constructed, which greatly improves the fault tolerance and stability of the operation. The solution has strong adaptability and has extremely high value for large-scale promotion. Attached Figure Description

[0029] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0030] Figure 1 A flowchart illustrating a method for precise trajectory planning in multi-robot collaborative operation of industrial robots, provided by an embodiment of the present invention;

[0031] Figure 2 This is a flowchart of a multi-machine collaborative strongly coupled constraint adaptive calibration method provided in an embodiment of the present invention.

[0032] Figure 3 This is a flowchart of a multi-machine global multi-objective optimization solution provided in an embodiment of the present invention;

[0033] Figure 4 This is a core architecture diagram of a digital twin scenario provided in an embodiment of the present invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, 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.

[0035] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0036] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0037] Currently, conventional multi-machine collaborative trajectory planning in this field generally adopts the technical approach of "task decoupling - single-machine planning - splicing and integration," which has several inherent defects. Firstly, the multi-machine collaborative benchmark is not unified. Existing solutions mostly rely on the individual robot's own base coordinate system for single-machine trajectory planning, failing to construct a unified world coordinate system to complete global registration between multiple machines and the work scene. This results in misalignment of the spatial pose benchmarks of multiple machines, failing to meet the requirements of high-precision collaborative assembly, welding, and other high-end operations. Secondly, the handling of coupling constraints is insufficient. Conventional solutions decompose the strong coupling constraints of multi-machine collaboration into multiple independent constraints for individual machines, only setting fixed safe obstacle avoidance distances and single-machine motion limit constraints. They do not fully consider the strong correlation constraints of multi-machine temporal and pose synchronization. Under dynamic conditions, either conservative settings reduce work efficiency, or excessively close distances pose collision safety hazards.

[0038] In terms of trajectory generation and optimization, existing technologies have significant shortcomings. Conventional solutions can only generate trajectories that meet the G1 continuous standard. Sudden curvature changes can easily cause robot joint vibration and excessive impact, which reduces operational accuracy and exacerbates equipment wear. Conflict verification only performs static collision detection at discrete time points of the trajectory and does not conduct time-space dual-dimensional matching verification throughout the entire operation cycle, making it easy to miss the risk of cross-collision during the dynamic movement of multiple machines. At the same time, it cannot achieve multi-machine global multi-objective collaborative optimization, and can only complete local optimization of single machine single object. It is difficult to balance operation efficiency, operational stability and equipment energy consumption. The optimization results are highly random and have poor engineering feasibility.

[0039] Furthermore, existing solutions generally lack a closed-loop management mechanism for the entire process, often employing an open-loop planning model. Once trajectory planning is complete, execution is immediately initiated without full-process virtual simulation pre-verification, making it impossible to identify trajectory failures under dynamic operating conditions in advance. Even when trajectory correction mechanisms are implemented, they only allow for independent replanning of a single robot, failing to consider the impact of trajectory adjustments on other collaborative robots, easily leading to new collision risks and extremely low operational fault tolerance. Simultaneously, existing solutions are mostly customized developments, lacking versatility and adaptability. They cannot quickly adapt to different robot models and quantities, nor to the changing demands of flexible production, making it difficult to meet the core performance requirements of high-end intelligent manufacturing for multi-robot collaborative operations.

[0040] The present invention aims to provide a method for accurate trajectory planning of multi-machine collaborative operation of industrial robots, in order to solve the problems of inconsistent multi-machine collaborative benchmarks, poor synchronization and high collision risk caused by decoupling of strong coupling constraints, poor trajectory continuity and incomplete conflict verification, inability to achieve global multi-objective optimization of multi-machines, lack of closed-loop management of the whole process, and poor versatility of the solution in the existing technology.

[0041] like Figure 1 and Figure 4 As shown in the figure, a specific embodiment of the present invention provides a method for accurate trajectory planning of multi-robot collaborative operations of industrial robots, including the following steps:

[0042] S1: Obtain the kinematic and dynamic parameters of multiple industrial robots to be coordinated, the global static environment information of the work scene, and the overall task requirements of multi-robot collaborative operation. Construct a unified world coordinate system and complete the global registration of multiple industrial robots to be coordinated and the work scene in the unified world coordinate system. Based on the registration results, construct a digital twin scene model of multi-robot collaborative operation.

[0043] In this embodiment of the invention, the base coordinate system pose data of multiple industrial robots to be coordinated are also acquired, and global registration of the multiple industrial robots to be coordinated with the work scene is completed in a unified world coordinate system based on the base coordinate system pose data. The constructed multi-robot collaborative operation digital twin scenario model is a dynamic mapping model that can map the running pose of industrial robots, changes in the work scene environment, and the execution progress of work tasks in real time. The multi-robot collaborative operation digital twin scenario model includes an industrial robot body twin unit, a work scene environment twin unit, a collaborative task mapping unit, and a simulation verification unit. The industrial robot body twin unit is used to map the kinematics and dynamic characteristics of the multiple industrial robots to be coordinated, the work scene environment twin unit is used to map the global static environment and dynamic changes of the work scene, the collaborative task mapping unit is used to map the total task requirements of multi-robot collaborative operation and the decomposed sub-tasks, and the simulation verification unit is used to realize the full-process virtual simulation of the trajectory and the simulation of abnormal working conditions.

[0044] S2: As Figure 2 As shown, based on the unified world coordinate system and the overall task requirements of multi-machine collaborative operation, the overall task requirements of multi-machine collaborative operation are decomposed into sub-tasks corresponding to multiple industrial robots to be collaborated. The strong coupling constraint system of multi-machine collaboration is simultaneously calibrated. The strong coupling constraint system of multi-machine collaboration includes the safety obstacle avoidance distance threshold between multiple robots, the timing matching window of synchronous operation, the end pose synchronization deviation threshold of collaborative operation, and the body motion limit constraints of each industrial robot.

[0045] In this embodiment of the invention, based on the overall task requirements of a unified world coordinate system and multi-machine collaborative operation, strongly coupled operation nodes are identified. These nodes are operation procedures where multiple industrial robots need to coordinate synchronously to complete tasks, exhibiting strong spatial and temporal correlations. The overall task requirements of multi-machine collaborative operation are decomposed into sub-tasks that correspond one-to-one with multiple industrial robots to be collaborated on and are bound to the strongly coupled operation nodes. The calibrated multi-machine collaborative strong coupling constraint system also includes: joint torque limit constraints for each industrial robot and curvature continuity constraints for Cartesian space trajectories. The safe obstacle avoidance distance threshold between multiple robots is a dynamic safe obstacle avoidance distance threshold that adjusts in real-time with the operating speed of the industrial robots. The dynamic safe obstacle avoidance distance threshold is related to the real-time operation of the industrial robots. Speed ​​is positively correlated; the faster the operating speed, the larger the corresponding safety obstacle avoidance distance threshold. It also includes a multi-machine collaborative strongly coupled constraint adaptive dynamic calibration and iterative optimization mechanism based on a digital twin scenario. The execution steps are as follows: Dataset construction: Collect historical operation data from the same scenario, digital twin simulation data, and real-time operating condition feature data of the current operation. After denoising and normalization preprocessing, a standardized training dataset is constructed to eliminate the influence of units. Self-learning model construction: Using the standardized dataset as input and the various thresholds of the constraint system as output, a BP neural network constraint self-learning optimization model is constructed. The comprehensive loss function is 100% achievement of operation accuracy, zero collisions, and the shortest operation time. Gradient descent is used to train until convergence. The convergence criterion is that the change in loss over 100 consecutive iterations is less than... Adaptive differential calibration: Input the working condition features of this operation into the trained model and output the initial value of the constraint threshold; tighten the timing and pose constraints of strongly coupled operation nodes to ensure accuracy, and relax the constraints of uncoupled operation nodes within a safe range to improve efficiency;

[0046] ;

[0047] In the formula, This is the initial value for the dynamic safe obstacle avoidance distance. Based on the safe distance, For the robot's rated speed, For the density of obstacles in the scene, This is the accuracy coefficient for the operation. , , These are the weighting coefficients corresponding to speed, obstacle density, and operational accuracy, respectively. These weighting coefficients are obtained by training the constraint self-learning optimization model. Dynamic adjustment throughout the entire process: During trajectory planning, simulation, and operation, real-time data on robot speed, load, scene changes, and operational progress are collected to dynamically adjust constraint thresholds. When speed increases, load increases, or obstacle density increases, the safety distance is increased; when accuracy fluctuations increase, timing and pose constraints are tightened, with the adjustment range not exceeding ±30% of the initial value. Full-cycle iterative optimization: After a single batch of operations is completed, the entire process data is added to the training dataset for incremental model training. A full retraining is performed every 10 batches of operations to continuously adapt to changes in working conditions.

[0048] S3: In the digital twin scenario model of multi-machine collaborative operation, for each sub-task of multiple industrial robots to be coordinated, the initial continuous trajectory of each industrial robot in Cartesian space is generated by combining the kinematic and dynamic parameters of the corresponding industrial robot and the strong coupling constraint system of multi-machine collaboration. Based on the unified world coordinate system, the initial continuous trajectory in Cartesian space is checked for spatiotemporal matching at all times to identify abnormal trajectory segments that do not conform to the strong coupling constraint system of multi-machine collaboration.

[0049] In this embodiment of the invention, the generated initial continuous trajectory in Cartesian space is a Cartesian space G2 continuous trajectory, which is a trajectory with continuous curvature throughout without abrupt changes. The full-time spatiotemporal matching verification is carried out in a unified world coordinate system, performing spatiotemporal dual-dimensional matching verification at any time node on the continuous time dimension of the entire operation cycle for the Cartesian space G2 continuous trajectory of all industrial robots. The content of the spatiotemporal dual-dimensional matching verification is: in the time dimension, verifying the operation timing matching degree of each industrial robot, and in the spatial dimension, verifying the relative pose and safe distance between multiple robots. Abnormal trajectory segments that do not conform to the strong coupling constraint system of multi-robot collaboration are trajectory segments where the spatial distance between multiple robots is less than the safe obstacle avoidance distance threshold, the timing deviation exceeds the timing matching window of synchronous operation, and the pose deviation exceeds the end pose synchronization deviation threshold of collaborative operation at any time node.

[0050] S4: As Figure 3 As shown, with the optimization objectives of shortest total operation time, optimal trajectory smoothness, and lowest joint drive energy consumption, a multi-machine global collaborative multi-objective optimization function is constructed. The multi-machine collaborative strong coupling constraint system calibrated in step S2 is used as the boundary condition, and the abnormal trajectory segments identified in step S3 are used as the core optimization interval. The initial continuous trajectory in Cartesian space is solved by global collaborative optimization to obtain the optimized pre-execution trajectory of each industrial robot.

[0051] In this embodiment of the invention, the constrained non-dominated sorting genetic algorithm NSGA-III is used to solve the global collaborative optimization problem. The multi-machine collaborative strong coupling constraint system calibrated in step S2 is used as the rigid constraint condition of the algorithm, and the spatiotemporal parameters of the initial continuous trajectory in Cartesian space are used as the optimization decision variables to obtain the Pareto optimal solution set. The shortest total operation time is the first priority, the best trajectory smoothness is the second priority, and the lowest joint drive energy consumption is the third priority. The optimal solution is selected from the Pareto optimal solution set according to the progressive rule. The progressive rule is as follows: first, the solution set subset with the best first priority index is selected, then the secondary solution set subset with the best second priority index is selected from the solution set subset, and finally the global optimal solution with the best third priority index is selected from the secondary solution set subset. Based on the global optimal solution, the optimized pre-execution trajectory of each industrial robot is generated.

[0052] S5: Import the optimized pre-execution trajectory into the digital twin scenario model of multi-machine collaborative operation, carry out full-process closed-loop virtual simulation verification, simulate abnormal working conditions in the operation process, identify problem trajectory segments that do not conform to the strong coupling constraint system of multi-machine collaboration in the simulation process, perform local dynamic replanning for problem trajectory segments, and generate the final executable trajectory of multi-machine collaborative operation.

[0053] In this embodiment of the invention, the abnormal working conditions in the simulated operation process include dynamic obstacle intrusion, robot load fluctuation, and joint torque exceeding the limit. During the simulation, the operating posture, timing, and joint torque data of each industrial robot are collected at all times. Based on the multi-machine collaborative strong coupling constraint system, the local dynamic replanning of the problem trajectory segment is completed. The local dynamic replanning is as follows: with the multi-machine collaborative strong coupling constraint system as the rigid boundary and the smooth connection of the trajectory as the core requirement, the trajectory is replanned only for the time interval corresponding to the problem trajectory segment and the preset transition intervals before and after, without changing the non-associated trajectory segments of the unconstrained exceeding the limit problem, and at the same time, it adapts to the real-time abnormal working conditions in the simulation process to complete the dynamic adjustment. After the local dynamic replanning is completed, the full-process closed-loop virtual simulation verification is carried out again until there are no problems in the entire simulation process that do not conform to the multi-machine collaborative strong coupling constraint system, and then the final multi-machine collaborative operation executable trajectory is generated.

[0054] S6: The final multi-machine collaborative operation executable trajectory is synchronously sent to the controller of the corresponding industrial robot according to the timing matching window of the synchronous operation calibrated in step S2, driving each industrial robot to perform collaborative operation; during the operation, the actual operation data of each industrial robot is collected in real time and the deviation is compared with the final multi-machine collaborative operation executable trajectory. When the deviation exceeds the preset threshold, local trajectory replanning is triggered to ensure the synchronization of multi-machine collaborative operation.

[0055] In this embodiment of the invention, the real-time collected actual operating data of each industrial robot includes the actual operating posture data, joint torque data, and actual operation progress data of each industrial robot; when comparing the deviation with the final multi-robot collaborative operation executable trajectory, a multi-dimensional comparison of posture deviation, timing deviation, and torque deviation is carried out simultaneously; the multi-dimensional comparison is judged using a standardized deviation rate, which is calculated according to the following formula:

[0056] ;

[0057] In the formula, The deviation rate, This is the actual deviation value. The constraint thresholds are defined for the corresponding dimensions; the time-series deviation is the difference between the actual operation progress of the industrial robot and the preset operation progress of the final multi-robot collaborative executable trajectory; when the deviation exceeds the preset threshold, a local trajectory replanning is triggered based on the multi-robot collaborative strong coupling constraint system; local trajectory replanning refers to replanning the trajectory only for the trajectory interval corresponding to the deviation exceeding the limit and the preset transition intervals before and after, with the multi-robot collaborative strong coupling constraint system as the rigid boundary and multi-robot collaborative synchronization as the core requirement, without changing the non-associated trajectory segments without deviation problems, and simultaneously adapting to the real-time operation data of the industrial robot to complete dynamic adjustments; synchronously updating all relevant... The system tracks the operational trajectories of industrial robots, identifying all industrial robots with strongly coupled operational nodes that are associated with the robot exhibiting deviations exceeding limits. This ensures that the entire multi-robot collaborative operation process complies with the requirements of the strongly coupled multi-robot collaborative constraint system. It also includes a spatiotemporal linkage and collaborative compensation mechanism for multi-robot trajectories based on strongly coupled operational nodes. The execution steps are as follows: Associated Robot Locking: Before operation, based on the identified strongly coupled operational nodes, all associated industrial robots are locked, establishing a millisecond-level real-time data synchronization channel; Data Synchronization Acquisition and Processing: During the execution of strongly coupled operational nodes, the pose and timing of all associated robots are synchronously acquired at a frequency of 100Hz. Torque data is used to calculate the deviation rate (actual deviation / corresponding constraint threshold) for each dimension, and normalization is performed. Deviation risk prediction: a deviation warning threshold is set at 50%-80% of the constraint threshold. When the deviation rate of any associated robot reaches the warning threshold and continues to increase for three consecutive sampling periods, a risk of deviation accumulation is identified, triggering collaborative compensation. When the deviation exceeds the constraint threshold, global collaborative replanning is directly triggered, clarifying the mechanism boundary. Collaborative compensation quantity calculation: After compensation is triggered, a quadratic planning model is constructed to solve for the spatiotemporal collaborative compensation quantity, using a strongly coupled constraint system as a rigid boundary and minimizing multi-machine synchronization error and trajectory adjustment as optimization objectives. The maximum dimensional offset does not exceed 30% of the temporal matching window, and the maximum spatial dimension correction does not exceed 30% of the pose deviation threshold, ensuring that the constraint boundaries are not exceeded. Compensation trajectories are synchronously distributed: based on the solved compensation amount, G2 continuous compensation trajectories are generated only for the trajectory segments corresponding to strongly coupled operation nodes and the transition intervals before and after them, ensuring smooth connection with the original trajectory, and are synchronously distributed to all associated robot controllers. Compensation effect closed-loop verification: after compensation execution, it is verified in real time whether the multi-machine synchronization deviation has decreased to below the warning threshold; if the deviation continues to increase, compensation is terminated and replanning is triggered; if the target is met, the compensation data is added to the training dataset to form a closed-loop linkage.

[0058] The beneficial effects of this invention are as follows: By constructing a unified world coordinate system and building a supporting multi-machine collaborative digital twin scene model, this invention solves the problem of misalignment of multi-machine collaborative benchmarks, providing a precise virtual-real mapping carrier for the entire trajectory planning process; it breaks through the conventional approach of "decoupling and dimensionality reduction" in existing technologies, taking multi-machine global coupling collaboration as the core, and through strong coupling of task binding of operation nodes, construction of a full-dimensional strong coupling constraint system, and constraint adaptive calibration iteration mechanism, it solves the pain points of poor synchronization and high collision risk caused by decoupling processing in existing technologies, balancing operation efficiency and operational safety; and it generates a G2 continuous trajectory avoidance mechanism. The robot's joint vibration impact, combined with all-time spatiotemporal dual-dimensional matching and verification, solves the collision omission problem of existing discrete point static detection technology; by constructing a multi-machine global multi-objective optimization function, the optimization solution is completed by using the constrained NSGA-III algorithm, which breaks through the limitation of single-machine local optima in existing technology; a new enhanced spatiotemporal linkage and collaborative compensation mechanism for coupled operation nodes fills the technical gap of existing technology that can only replan after the fact, reducing the frequency of replanning triggering; finally, a closed-loop control system for the entire process is constructed, which greatly improves the fault tolerance and stability of the operation. The solution has strong adaptability and has extremely high value for large-scale promotion.

[0059] The above descriptions are merely embodiments of the present invention. Commonly known technical solutions or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. An industrial robot multi-machine collaborative work trajectory accurate planning method, characterized in that, The method includes: S1: Obtain the kinematic and dynamic parameters of multiple industrial robots to be coordinated, the global static environment information of the work scene, and the total task requirements of multi-robot collaborative operation; construct a unified world coordinate system; complete the global registration of the multiple industrial robots to be coordinated and the work scene in the unified world coordinate system; and construct a digital twin scene model of multi-robot collaborative operation based on the registration results. S2: Based on the unified world coordinate system and the overall task requirements of the multi-machine collaborative operation, the overall task requirements of the multi-machine collaborative operation are decomposed into sub-tasks corresponding to the multiple industrial robots to be collaborated. The strong coupling constraint system of multi-machine collaboration is simultaneously calibrated. The strong coupling constraint system of multi-machine collaboration includes the safety obstacle avoidance distance threshold between multiple machines, the timing matching window of synchronous operation, the end pose synchronization deviation threshold of collaborative operation, and the body motion limit constraints of each industrial robot. S3: In the multi-machine collaborative operation digital twin scenario model, for each sub-task of the multiple industrial robots to be collaborated, the Cartesian space initial continuous trajectory of each industrial robot corresponding to the sub-task is generated by combining the kinematic and dynamic parameters of the corresponding industrial robot and the multi-machine collaborative strong coupling constraint system; based on the unified world coordinate system, the Cartesian space initial continuous trajectory is subjected to full-time spatiotemporal matching verification to identify abnormal trajectory segments that do not conform to the multi-machine collaborative strong coupling constraint system; S4: With the optimization objectives of minimizing total operation time, optimizing trajectory smoothness, and minimizing joint drive energy consumption, a multi-machine global collaborative multi-objective optimization function is constructed. The multi-machine collaborative strong coupling constraint system calibrated in step S2 is used as the boundary condition, and the abnormal trajectory segment identified in step S3 is used as the core optimization interval. The initial continuous trajectory in the Cartesian space is then solved through global collaborative optimization to obtain the optimized pre-execution trajectory of each industrial robot. S5: Import the optimized pre-execution trajectory into the multi-machine collaborative operation digital twin scenario model, carry out full-process closed-loop virtual simulation verification, simulate abnormal working conditions during the operation, identify problem trajectory segments that do not conform to the multi-machine collaborative strong coupling constraint system during the simulation, perform local dynamic replanning for the problem trajectory segments, and generate the final multi-machine collaborative operation executable trajectory. S6: The final multi-machine collaborative operation executable trajectory is synchronously sent to the controller of the corresponding industrial robot according to the timing matching window of the synchronous operation calibrated in step S2, driving each industrial robot to perform collaborative operation; during the operation, the actual operation data of each industrial robot is collected in real time and the deviation is compared with the final multi-machine collaborative operation executable trajectory. When the deviation exceeds the preset threshold, local trajectory replanning is triggered to ensure the synchronization of multi-machine collaborative operation.

2. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: After acquiring the kinematic and dynamic parameters of multiple industrial robots to be coordinated, the global static environment information of the work scene, and the total task requirements of multi-robot collaborative operation, the base coordinate system pose data of the multiple industrial robots to be coordinated is also acquired, and the global registration of the multiple industrial robots to be coordinated and the work scene under the unified world coordinate system is completed based on the base coordinate system pose data. The constructed multi-robot collaborative operation digital twin scenario model is a dynamic mapping model that can map the industrial robot's operating posture, changes in the working environment, and the progress of the task execution in real time. The multi-robot collaborative operation digital twin scenario model includes an industrial robot body twin unit, a working environment twin unit, a collaborative task mapping unit, and a simulation verification unit. The industrial robot body twin unit is used to map the kinematics and dynamic characteristics of the multiple industrial robots to be coordinated. The work scene environment twin unit is used to map the global static environment and dynamic changes of the work scene. The collaborative task mapping unit is used to map the total task requirements and the decomposed sub-tasks of the multi-machine collaborative operation. The simulation verification unit is used to realize the full-process virtual simulation of the trajectory and the simulation of abnormal working conditions.

3. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: Step S2 specifically involves: based on the unified world coordinate system and the overall task requirements of the multi-machine collaborative operation, identifying strongly coupled operation nodes of the multi-machine collaborative operation. The strongly coupled operation nodes are operation process nodes where multiple industrial robots need to cooperate synchronously to complete the operation and there is a strong spatial and temporal correlation. The overall task requirements of the multi-machine collaborative operation are broken down into sub-tasks that correspond one-to-one with the multiple industrial robots to be collaborated on and are bound to the strongly coupled operation nodes.

4. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 3, characterized in that: The multi-machine collaborative strong coupling constraint system calibrated in step S2 also includes: joint torque limit constraints of each industrial robot and curvature continuity constraints of Cartesian space trajectory; The safe obstacle avoidance distance threshold between multiple robots is a dynamic safe obstacle avoidance distance threshold that is adjusted in real time according to the operating speed of the industrial robot. The dynamic safe obstacle avoidance distance threshold is positively correlated with the real-time operating speed of the industrial robot; the faster the operating speed, the larger the corresponding safe obstacle avoidance distance threshold. The initial value of the dynamic safe obstacle avoidance distance is calculated according to the following formula: ； In the formula, This is the initial value for the dynamic safe obstacle avoidance distance. Based on the safe distance, For the robot's rated speed, For the density of obstacles in the scene, This is the accuracy coefficient for the operation. , , These are the weighting coefficients corresponding to speed, obstacle density, and operation accuracy, respectively, which are obtained by training a constrained self-learning optimization model.

5. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: In step S3, the generated Cartesian space initial continuous trajectory is a Cartesian space G2 continuous trajectory, which is a trajectory with continuous curvature throughout without abrupt changes.

6. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 5, characterized in that: The full-time spatiotemporal matching verification in step S3 is to perform spatiotemporal dual-dimensional matching verification at any time node in the continuous time dimension of the entire operation cycle for the Cartesian space G2 continuous trajectory of all industrial robots under the unified world coordinate system. The spatiotemporal dual-dimensional matching verification includes: time dimension verification of the matching degree of the operation sequence of each industrial robot, and spatial dimension verification of the relative pose and safe distance between multiple robots. The abnormal trajectory segments that do not conform to the multi-machine collaborative strong coupling constraint system are trajectory segments where, at any time node, the spatial distance between multiple machines is less than the safe obstacle avoidance distance threshold, the timing deviation exceeds the timing matching window of the synchronous operation, and the pose deviation exceeds the end pose synchronization deviation threshold of the collaborative operation.

7. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: In step S4, the constrained non-dominated sorting genetic algorithm NSGA-III is used to complete the global cooperative optimization solution. The multi-machine cooperative strong coupling constraint system calibrated in step S2 is used as the rigid constraint condition of the algorithm, and the spatiotemporal parameters of the initial continuous trajectory in the Cartesian space are used as the optimization decision variables to obtain the Pareto optimal solution set. The optimal solution is selected from the Pareto optimal solution set according to a progressive rule, with the shortest total operation time as the first priority, the best trajectory smoothness as the second priority, and the lowest joint drive energy consumption as the third priority. The progressive rule is as follows: first, select the solution set subset with the best first priority index; then, select the secondary solution set subset with the best second priority index from the solution set subset; finally, select the global optimal solution with the best third priority index from the secondary solution set subset; and generate the optimized pre-execution trajectory for each industrial robot based on the global optimal solution.

8. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: The abnormal working conditions simulated in step S5 include dynamic obstacle intrusion, robot load fluctuation, and joint torque exceeding the limit. During the simulation, the operating posture, timing, and joint torque data of each industrial robot are collected at all times. Based on the multi-machine collaborative strong coupling constraint system, the local dynamic replanning of the problem trajectory segment is completed. The local dynamic replanning is as follows: taking the multi-machine collaborative strong coupling constraint system as the rigid boundary and smooth trajectory connection as the core requirement, the trajectory replanning is only performed on the time interval and the preset transition interval before and after the problem trajectory segment, without changing the non-associated trajectory segment of the unconstrained over-limit problem, and at the same time, it adapts to the real-time abnormal working conditions in the simulation process to complete dynamic adjustment. After completing the local dynamic replanning, a full-process closed-loop virtual simulation verification is carried out again until there are no problems in the simulation process that do not conform to the multi-machine collaborative strong coupling constraint system, and then the final multi-machine collaborative operation executable trajectory is generated.

9. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 1, characterized in that: The actual operating data of each industrial robot collected in real time in step S6 includes the actual operating posture data, joint torque data, and actual operation progress data of each industrial robot. When comparing the deviation with the final multi-machine collaborative work executable trajectory, a multi-dimensional comparison of pose deviation, timing deviation, and torque deviation is carried out simultaneously; the multi-dimensional comparison is judged using a standardized deviation rate, which is calculated according to the following formula: ； In the formula, The deviation rate, This is the actual deviation value. The constraint threshold for the corresponding dimension; The timing deviation is the deviation between the actual operation progress of the industrial robot and the preset operation progress of the final multi-robot collaborative operation executable trajectory.

10. The method for accurate trajectory planning of multi-robot collaborative operation of industrial robots according to claim 9, characterized in that: In step S6, when the deviation exceeds the preset threshold, local trajectory replanning is triggered based on the multi-machine collaborative strong coupling constraint system. The local trajectory replanning refers to using the multi-machine collaborative strong coupling constraint system as a rigid boundary and multi-machine collaborative synchronization as the core requirement, only targeting the trajectory interval corresponding to the deviation exceeding the limit and the preset transition interval before and after, without changing the non-associated trajectory segments without deviation problems, and at the same time adapting to the real-time operation data of the industrial robot to complete dynamic adjustment. The operation trajectories of all associated industrial robots are updated synchronously. The associated industrial robots are all industrial robots that are strongly coupled with the operation nodes of the industrial robot with the deviation exceeding the limit, ensuring that the entire process of multi-machine collaborative operation meets the requirements of the multi-machine collaborative strong coupling constraint system.

Images