Welding control method and system of double-arm robot

Abstract

According to the welding control method and system for the double-arm robot, the double-arm robot is controlled to complete a welding task on a non-standard workpiece, and the welding control method comprises the following steps that (1) structural analysis and disassembly are conducted on 3D data of a target workpiece, and independent parts are generated; 2) based on reinforcement learning, taking a double-arm robot as an intelligent body, training and optimizing the optimal assembling strategy of the parts, and outputting an assembling task plan and a welding task plan as control strategy results; and (3) the motion trail of the mechanical arm of the double-arm robot is generated based on the control strategy result decision, the multiple parts are sequentially assembled and welded according to the assembling sequence, and a finished product of the target workpiece is obtained. Full-automatic welding control of the double-arm robot on complex non-standard workpieces is achieved, automatic analysis, disassembly, assembly and tailor-welding tasks of complex parts are completed, traditional methods such as manual welding and tool pre-clamping are not needed any more, and the welding quality consistency can be further improved.

Description

Technical Field

[0001] This invention belongs to the field of robotics technology, and relates to robotic arms, specifically a welding control method and system for a dual-arm robot. Background Technology

[0002] Currently, laser welding technology is mainly divided into two categories: handheld laser welding and robotic welding. Handheld laser welding relies on the coordinated operation of the operator's two hands and can flexibly handle welding tasks involving complex workpieces and small-batch production. Robotic welding is usually based on a single-arm collaborative robot or a single-arm industrial robot. Its system consists of the robot body, a welding workstation, and matching fixtures. In the robotic welding process, pre-clamping or pre-spot welding positioning using tooling fixtures is necessary for welding preparation. Subsequently, the welding task is achieved through pre-programming. This welding method is mainly suitable for the welding and manufacturing of batch, standardized products.

[0003] Robotic welding is a crucial direction in the development of automation. However, traditional robotic welding methods often fall short when faced with welding complex, non-standard workpieces. This is primarily because the welding process for complex workpieces is intricate, involving component assembly and multiple welding operations. Pre-programming for automated robotic welding is difficult, and the diverse, small-batch nature of non-standard components necessitates the design of different task plans for each type of component when using robots for automated welding. This approach results in long deployment cycles and high costs. Therefore, general-purpose traditional robotic welding methods are no longer suitable for welding complex, non-standard workpieces.

[0004] In recent years, with the rapid development of sensing technology and artificial intelligence, dual-arm robot technology has been more widely applied and developed. Dual-arm robots can not only complete basic tasks such as sorting and palletizing, but also more complex tasks such as folding clothes. However, current applications of dual-arm robots are mostly concentrated in sorting, grasping, palletizing, and assembly, while technical bottlenecks remain for specialized tasks such as welding, especially the assembly and welding of complex non-standard parts. This is because welding tasks require disassembling parts according to the structure of complex workpieces. In addition to welding, the task of assembling the parts must also be addressed, and the two tasks must be combined according to the welding process, requiring detailed assembly and welding task planning. Existing dual-arm robots do not yet possess this capability. Summary of the Invention

[0005] The technical problem to be solved by the present invention is that in existing robotic welding tasks, the robot only focuses on the welding task. However, for workpiece products composed of multiple non-standard parts, the existing dual-arm robot control method cannot automatically identify the workpiece products and output the corresponding assembly and welding task plan.

[0006] The technical solution of this invention is: a welding control method for a dual-arm robot, which controls the dual-arm robot to complete welding tasks on non-standard workpieces, including the following steps:

[0007] 1) Call the 3D data of the target workpiece, and based on the preset welding process parameters of the target workpiece, perform structural analysis and disassembly on the 3D model of the target workpiece to generate the independent parts required to form the welded body, and prepare the corresponding real parts.

[0008] 2) Based on reinforcement learning, the optimal action strategy for assembling and welding parts is learned by using a dual-arm robot as the intelligent agent. One arm of the dual-arm robot performs the assembly task and the other arm performs the welding task. Combining the structural analysis and disassembly results of the 3D model, the assembly task and welding task are planned as the control strategy results. The assembly task planning includes the assembly sequence and pose of the corresponding parts and the assembly action trajectory. The welding task planning includes the welding action trajectory of the pose of the corresponding parts.

[0009] 3) Generate the motion trajectory of the dual-arm robot based on the control strategy results. One arm of the dual-arm robot acquires real parts and adjusts the pose of the real parts according to the assembly task planning. The other arm controls the welding action to perform welding on the current real parts based on the welding task planning. Through the iterative cycle of assembly-welding, multiple real parts are welded in sequence according to the assembly order and pose to obtain the finished product of the target workpiece.

[0010] Furthermore, in step 3), the control strategy results are simulated and verified using the robot operating system ROS platform. After the simulation verification is passed, the parameters and behavioral data obtained from the simulation are deployed and debugged on the real dual-arm robot system. After the debugging is completed, welding commands are issued to the dual-arm robot system to drive the two arms to perform a collaborative assembly and welding operation sequence in the real physical environment.

[0011] Furthermore, step 2) based on reinforcement learning, using the dual-arm robot as the intelligent agent to learn the optimal strategy for assembling and welding parts, includes:

[0012] Define a state description function, using the TCP pose, velocity, and welding assembly state of the robotic arm end effector as state variables, to construct the robot action state function for the assembly and welding process, which serves as the basis for judging reinforcement learning training.

[0013] Define an assembly reward function to quantitatively evaluate the quality of the assembly status of parts, the TCP pose of the robot end effector, and the speed during the assembly and welding process;

[0014] Using a dual-arm robot as the intelligent agent, the optimal motion strategy for assembling and welding the parts of the dual-arm robot is reinforced through learning, and the assembly task plan and welding task plan are output as the control strategy results.

[0015] Furthermore, the control strategy results, for the assembly task planning, include the following steps:

[0016] Step A1: Obtain the feature information of the virtual model parts and the feature information of the real parts from the 3D model disassembly in Step 1), match the feature information of the virtual model parts with the feature information of the real parts, and establish the correspondence between the virtual model parts and the real parts.

[0017] Step A2: Determine the base components that form the basis for assembly, generate component assembly strategies for the virtual model components, and perform robot kinematic modeling. Mathematically describe the motion relationships of the robotic arm driving the various components during the assembly process, including the joint motion and pose transformation of the robotic arm, to control the robot's motion trajectory.

[0018] Step A3: Optimize the generated component assembly strategy and robot kinematic model using reinforcement learning. Through the interaction between the agent and the environment, adjust the assembly strategy according to the assembly reward function. The reinforcement learning results are verified using Sim2real, applying the learning results in the simulated environment to the real scene. If the verification is successful, the assembly task plan is output for assembling real components. If the verification fails, return to the reinforcement learning stage to further adjust and optimize the strategy until the verification is successful.

[0019] Step A4: Obtain the actual base components, establish a workpiece coordinate system based on the base components to provide reference coordinates for the assembly operation, and combine the correspondence between the simulated model components and the actual components to obtain the assembly motion trajectory of the dual-arm robot from the assembly task planning.

[0020] Furthermore, the control strategy results, for welding task planning, include the following steps:

[0021] Step B1: Obtain information about the dual-arm robot and the parts to be welded, including the size, shape, material of the parts, and motion specifications of the welding robot.

[0022] Step B2: Establish a welding simulation environment to simulate the relative positional relationship and operating conditions of the welding robot and the parts. Then, identify the weld seams of the parts in the simulation environment and determine their position and shape characteristics.

[0023] Step B3: Generate a welding path based on the identified weld, covering the entire length of the weld and conforming to the process specifications. Then, perform forward and inverse kinematics solutions to calculate the pose of the welding torch and the motion commands of the robot joints. After that, use offline simulation to determine whether the welding posture is reasonable. If it is not reasonable, return to the kinematics solution step for adjustment until the simulation results meet the requirements, and output the welding trajectory corresponding to the identified weld.

[0024] Meanwhile, the weld information of the real parts during the assembly process is obtained and matched with the welds identified in the simulation environment. Based on Sim2real, the robot's motion trajectory in the real environment is determined. Combining the kinematic characteristics of the robotic arm and the welding process parameters, the welding process parameter package is called to obtain the planned current welding trajectory.

[0025] Step B4: Based on the current welding trajectory planned in step B3, complete the welding task for the current actual component and proceed to the welding of the next component;

[0026] Step B5: Repeat steps B3 to B4 until all parts are assembled and welded.

[0027] The present invention also provides a welding control system for a dual-arm robot, including a movable leveling platform, a dual-arm robot, a vision perception system, a welding system, and an AI intelligent computing unit. The dual-arm robot includes a left arm and a right arm. The welding system is set at the end of the right arm via an end-effector gripping system. The AI ​​intelligent computing unit stores at least one instruction or at least one program. When the at least one instruction or the program is executed, the above-mentioned welding control method for the dual-arm robot is implemented.

[0028] Furthermore, the AI ​​intelligent computing unit includes the following modules:

[0029] Assembly trajectory planning module: Used to plan and output the motion trajectory of one arm of the dual-arm robot according to the assembly task, to complete the assembly task of parts, including:

[0030] Assembly robot arm trajectory generation unit: Generates the robot arm assembly motion trajectory based on the current real parts and assembly task plan;

[0031] Assembly trajectory optimization unit: Utilizes smoothing, speed planning, and collision detection algorithms to optimize the robotic arm's assembly motion trajectory;

[0032] Welding trajectory planning module: Used to plan and output the motion trajectory of the other arm of the dual-arm robot according to the welding task, to complete the welding task of the parts, including:

[0033] Welding robot arm trajectory generation unit: Generates the welding motion trajectory of the robot arm based on the current real parts and welding task planning;

[0034] Welding trajectory optimization unit: Adjusts the smoothness, continuity, speed, and acceleration parameters of the robotic arm's welding motion trajectory to meet welding process requirements;

[0035] Identification module: used to identify parts and welds, including:

[0036] Component identification unit: Used to identify and classify components by applying image processing and information tagging methods to the acquired component images, shape, and size feature information;

[0037] Weld seam recognition unit: Based on a visual perception system, it detects the position, shape and size of weld seams in real time, identifies weld seam information of real parts, and provides a basis for welding trajectory planning;

[0038] Control module: Used to output control commands to the dual-arm robot based on the control strategy results, including:

[0039] Robot posture control unit: Based on the motion trajectory planned for the assembly and welding tasks and the real-time feedback information from the robot, it controls the posture of the dual-arm robot and performs posture tracking by combining the dynamic model and external disturbance factors;

[0040] Welding process control unit: Sets process parameters such as welding current, voltage, speed, and wire feed speed, and monitors and adjusts them in real time to ensure the stability and consistency of welding process.

[0041] Existing robot welding control methods cannot determine the welding sequence for complex non-standard workpieces, making it difficult to automatically plan welding tasks. Even with pre-setting the welding sequence for non-standard parts, the flexibility and customization of these parts require constant adjustments to the tooling and fixtures to accommodate their assembly. This still makes it difficult to accurately coordinate and plan the motion tasks of a single-arm collaborative robot or a single-arm industrial robot. Furthermore, manual pre-programming is extremely labor-intensive and costly to debug. This invention proposes a welding control method and system for a dual-arm robot, which offers the following advantages:

[0042] 1. End-to-End Welding Task Planning and Execution: This invention automates the entire welding process through reinforcement learning, from task input to welding completion. It automatically plans the optimal path and adjusts welding parameters in real time to ensure welding quality and efficiency. This invention eliminates the need for manual analysis of non-standard workpieces, manual planning of welding schemes, and subsequent programming and deployment to the robot. First, it performs disassembly analysis of non-standard workpieces in a virtual environment and plans the robot's motion trajectory based on reinforcement learning. Then, it matches the strategy from the virtual environment to the real environment, achieving a fully automated welding control method for complex non-standard workpieces using a dual-arm robot: "automatic disassembly - planning assembly + planning welding - collaborative assembly and welding operation." It collaboratively plans the assembly and welding tasks for complex workpieces. For welding tasks of small batches of non-standard workpieces, only a 3D model of the non-standard workpiece needs to be constructed to automatically generate the cyclical collaborative operation task planning for one arm of the dual-arm robot (assembly) and the other arm (welding), completing the automatic analysis and assembly welding of complex components.

[0043] 2. Low-cost intelligent production of non-standard and small-batch products: Utilizing Sim2Real technology, combined with randomized domain enhancement and multimodal perception fusion, seamless migration from simulation training to real-world production is achieved, reducing production costs and improving flexibility and efficiency in small-batch production. This invention generates 3D data of the product through 3D modeling for simulation training, combining assembly task decision-making, fine-tuning model weld seam recognition, and welding trajectory planning technologies. Ultimately, the deployment and implementation of welding tasks for non-standard workpieces are completed. This invention automatically generates a control strategy for the coordinated work of a dual-arm robot; moreover, most of the debugging work for this control strategy is performed in the simulation environment, resulting in low debugging costs. This invention frees up welding labor, eliminating the need for manual welding and tooling pre-clamping, and achieving better welding quality consistency, no longer limited by the skill level of traditional manual welders.

[0044] 3. Low-cost and lightweight deployment: Optimize algorithm design and hardware configuration, adopt the lightweight deep learning model Segformer and modular design to reduce computing resource requirements, lower hardware costs, improve system robustness and reliability, and meet the production needs of modern manufacturing for non-standard and small-batch products. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the overall process of the welding control method of the present invention.

[0046] Figure 2 This is a schematic diagram of a preferred embodiment of the present invention.

[0047] Figure 3 This is a schematic diagram of the assembly task planning process based on reinforcement learning in the welding control method of the present invention.

[0048] Figure 4 This is a schematic diagram of the process of disassembling and matching virtual and physical components based on welding technology in the assembly task planning of this invention.

[0049] Figure 5 This is a flowchart illustrating the welding task planning process in the welding control method of the present invention.

[0050] Figure 6 This is a schematic diagram of the weld identification process, illustrating an embodiment of the welding task planning of the present invention.

[0051] Figure 7 This is a schematic diagram of the structure of the dual-arm robot welding control system of the present invention.

[0052] Figure 8 This is a schematic diagram of the AI ​​intelligent computing unit in the dual-arm robot welding control system of the present invention.

[0053] Figure 9 This is a schematic diagram of a specific embodiment of the present invention, showing the assembly and welding process of the first blade in the assembly and welding of components of a non-standard workpiece "blade".

[0054] Figure 10 For the corresponding Figure 9 A schematic diagram of the assembly and welding process for the second blade. Detailed Implementation

[0055] This invention provides a planning and control method and system for handling non-standard welding tasks using a dual-arm robot. Addressing the challenges of welding complex non-standard components, the invention seamlessly integrates a 3D model in a virtual simulation environment with the complex component in the real physical environment using "Sim2Real" technology. Combining the 3D model with a pre-trained weld recognition model, the method quickly and accurately identifies the weld locations and features in complex components, efficiently and precisely disassembling them and generating component information for sub-task execution. This process significantly improves disassembly efficiency and provides accurate component information for subsequent welding task planning. Furthermore, this invention introduces a "brain-like" task-level planning process based on reinforcement learning, achieving full-process planning from complex component disassembly to component assembly and welding, providing an efficient, intelligent, and flexible solution for the automated welding of complex non-standard components.

[0056] At the task planning level, this invention employs a "brain-like" task-layer planning process. This process simulates the decision-making and planning mechanisms of the human brain, starting with the disassembly of complex components and gradually working backward to plan the assembly process. Combined with the actual motion constraints of the dual-arm robot, the robot's motion trajectory is optimized, ultimately resulting in a detailed execution plan for the robot's assembly and welding task. This end-to-end planning method not only considers the complexity and diversity of welding tasks but also dynamically adjusts the task execution strategy based on real-world working conditions, ensuring the efficient completion of welding tasks.

[0057] like Figure 1 The diagram shows the overall flow of the welding control method for a dual-arm welding robot, including the following steps:

[0058] 1) The system retrieves the 3D data of the target workpiece and, based on preset welding process parameters, performs structural analysis and disassembly of the 3D model to generate independent components required to form the welded body. Corresponding real components are also prepared. Specifically, the system identifies parameters of the 3D model to accurately obtain key parameter information for each component, such as size, shape, and position. Next, based on the welding process, the 3D model is disassembled into individual virtual model components, ensuring the integrity and independence of each component. The real components correspond to the virtual model components, awaiting assembly and welding to form the target workpiece.

[0059] 2) Based on reinforcement learning, the optimal action strategy for assembling and welding parts is learned by using a dual-arm robot as the intelligent agent. One arm of the dual-arm robot performs the assembly task and the other arm performs the welding task. Combining the structural analysis and disassembly results of the 3D model, the assembly task and welding task are planned as the control strategy results. The assembly task planning includes the assembly sequence and pose of the corresponding parts and the assembly action trajectory. The welding task planning includes the welding action trajectory of the pose of the corresponding parts.

[0060] Specifically, it includes:

[0061] Define a state description function, using the TCP pose, velocity, and assembly / welding state of the robotic arm end effector as state variables, to construct the robot motion function for the assembly and welding process, which serves as the basis for judging reinforcement learning training.

[0062] Define an assembly reward function to quantitatively evaluate the quality of component assembly status, robot end effector TCP pose, and speed during assembly and welding. As an example, the reward function definition may include factors such as trajectory smoothness, continuity, compliance with robotic arm physical constraints, matching component characteristics and task requirements, and minimizing movement distance. If the trajectory is unreasonable, adjustments are made; if reasonable, the motion data is saved.

[0063] 3) Generate the motion trajectory of the dual-arm robot based on the control strategy results. One arm of the dual-arm robot acquires real parts and adjusts the pose of the real parts according to the assembly task planning. The other arm controls the welding action to perform welding on the current real parts based on the welding task planning. Through the iterative cycle of assembly-welding, multiple real parts are welded in sequence according to the assembly order and pose to obtain the finished product of the target workpiece.

[0064] For the same non-standard workpiece, the method of this invention can be used to perform reinforcement learning once to obtain the control strategy result, which can then be deployed to the robot to complete the mass production task.

[0065] Furthermore, such as Figure 2As shown, this invention also utilizes the Robot Operating System (ROS) platform to perform high-fidelity simulation verification of the planned task strategy. After successful simulation verification, the system is deployed and debugged on a real robot system based on the parameters and behavioral data obtained from the simulation. Once debugging is complete, welding commands are issued to the robot system, driving the dual arms to execute a coordinated assembly and welding sequence in a real physical environment. During the operation, multimodal sensor information, including visual and force sensors, is integrated to achieve real-time closed-loop feedback control of the operation process. Ultimately, this achieves fully automated welding tasks from component assembly to weld formation.

[0066] The following describes the specific implementation of the present invention based on reinforcement learning for planning assembly and welding tasks.

[0067] Reference Figure 3 This is a flowchart for the assembly task planning. The specific process is as follows:

[0068] Step A1: Obtain the feature information of the virtual model parts and the feature information of the real parts from the 3D model disassembly in Step 1), match the feature information of the virtual model parts with the feature information of the real parts, and establish the correspondence between the virtual model parts and the real parts.

[0069] Step A2: Determine the base components that form the basis for assembly, generate component assembly strategies for the virtual model components, and perform robot kinematic modeling. Mathematically describe the motion relationships of the robotic arm driving the various components during the assembly process, including the joint motion and pose transformation of the robotic arm, to control the robot's motion trajectory.

[0070] Step A3: Optimize the generated component assembly strategy and robot kinematic model using reinforcement learning. Through the interaction between the agent and the environment, adjust the assembly strategy according to the assembly reward function. The reinforcement learning results are verified using Sim2real, applying the learning results in the simulated environment to the real scene. If the verification is successful, the assembly task plan is output for assembling real components. If the verification fails, return to the reinforcement learning stage to further adjust and optimize the strategy until the verification is successful.

[0071] Step A4: Obtain the actual base components, establish a workpiece coordinate system based on the base components to provide reference coordinates for the assembly operation, and combine the correspondence between the simulated model components and the actual components to obtain the assembly motion trajectory of the dual-arm robot from the assembly task planning.

[0072] During the assembly process, the dual-arm robot needs to identify real parts and match them with parts in the simulation environment to determine how to perform actions according to the assembly task plan. The correspondence between virtual model parts and real parts is the foundation of part recognition, such as... Figure 4 The flowchart shown below illustrates the specific implementation process.

[0073] In component identification, for virtual model components, shape and size feature information, including size, shape, and surface texture, is obtained based on the 3D model disassembly and analysis results from step 1) to accurately reflect the shape and position characteristics of the components. Meanwhile, the geometric feature information of real components may differ slightly from the design model due to deviations during manufacturing or wear during use. Therefore, it needs to be acquired and analyzed separately. Feature information of real components, including shape, size, and surface texture, is obtained through scanning or measurement. As an example, images of real components are acquired using a robot-equipped visual sensing device. Real images may be affected by factors such as lighting, angle, and occlusion. When acquiring feature information of real components through images, robust image processing algorithms must be used to ensure the accuracy of the feature information.

[0074] Then, the virtual model components and real components are classified and labeled multiple times according to the acquired feature information. In this embodiment, this step is repeated three times to ensure the accuracy and comprehensiveness of the classification and labeling. Each classification and labeling may be based on different feature dimensions, such as size, shape, function, etc., to more finely distinguish and identify different types of components. Through multiple classifications and labels, the accuracy of component identification can be improved, providing more detailed information for the subsequent matching process. After completing the multiple classifications and labels of the components, the corresponding feature information of the virtual model components and real components is compared and matched to achieve precise alignment and matching between the virtual model and the physical components, establish the correspondence between the virtual model components and the real components, and determine their similarities and differences. This step ensures a high degree of consistency between virtual design and real production, providing a reliable basis for subsequent manufacturing, assembly, or quality control processes.

[0075] Reference Figure 5 The flowchart for welding trajectory planning includes the following steps.

[0076] Step B1: Obtain information about the parts to be welded by the welding robot, including the size, shape, material of the parts, and the specifications of the welding robot.

[0077] Step B2: Establish a welding simulation environment to simulate the relative positional relationship and operating conditions between the welding robot and the components. Then, identify the weld seams of the components in the simulation environment and determine their location and shape characteristics.

[0078] Step B3: Generate a welding path based on the identified weld seam, covering the entire length of the weld seam and conforming to the process specifications. Then, perform forward and inverse kinematics solutions to calculate the welding torch pose and the motion commands for the robot joints. Afterwards, offline simulation is used to determine if the welding posture is reasonable. If it is unreasonable, the process returns to the kinematics solution step for adjustment until the simulation results meet the requirements. Finally, the welding trajectory corresponding to the identified weld seam is output.

[0079] Simultaneously, weld information is acquired from the actual components during the assembly process to ensure its accuracy and completeness. Then, it is matched with the welds identified in the simulation environment. Based on Sim2real, the robot's motion trajectory in the real environment is determined. Combining the kinematic characteristics of the robotic arm and welding process parameters, the welding process parameter package is called to obtain the planned current welding trajectory, which guides the welding operation.

[0080] Step B4: Based on the current welding trajectory planned in step B3, complete the welding task for the current actual component and proceed to the welding of the next component.

[0081] Step B5: Repeat steps B3 to B4 until all components are assembled and welded. The entire process ensures precise planning of the welding trajectory and efficient, stable welding operations.

[0082] Furthermore, in the above process, a lightweight model is used to obtain weld information from real components, as follows: Figure 6 As shown, the specific implementation process is as follows:

[0083] B3.1. Conduct high-quality image acquisition, utilizing professional acquisition equipment to obtain high-resolution images of the weld seam, ensuring that the images clearly reflect the detailed features of the weld seam, providing a high-quality data source for subsequent processing. In this embodiment, a sensing camera is used to acquire images containing the weld seam of the current component to be welded;

[0084] B3.2 Preprocessing of the image includes converting the image to grayscale to simplify the data structure and highlight shape features, while using denoising algorithms to remove random noise from the image and improve the image clarity and usability;

[0085] B3.3 Weld Feature Extraction: Image processing techniques are used to extract key weld features from the preprocessed image, including edge contours and texture patterns. A lightweight Segformer segmentation model, based on deep learning, is then used to segment the image, efficiently and accurately separating the weld region from complex backgrounds and precisely defining its boundaries and extent, providing a clear visual foundation for subsequent detailed analysis. The segmented welds are then identified and analyzed, including quantitative analysis of their geometry, surface quality, and defect features. This assesses whether the welds meet process standards and quality requirements, generating an analysis report to ensure the accuracy and integrity of the welds and determine whether welding should proceed.

[0086] B3.4 Output the results of weld identification and analysis, including the location, size and quality grade of the weld, to be used in conjunction with the welds identified in the simulation environment to determine the robot's motion trajectory in the real environment based on Sim2real.

[0087] like Figure 7 As shown, this invention also provides a welding control system for a dual-arm robot. The core components of this system include a movable leveling platform 1, an end effector 2, a left-arm industrial robot 3, a vision perception system 4, a right-arm industrial robot 5, a welding system 6, and an AI intelligent computing unit 7. During operation, the movable leveling platform 1 maintains the position of the dual-arm robot, ensuring it operates stably on the field. The vision perception system 4 acquires the status information of the task product, and the AI ​​intelligent computing unit 7 performs decision analysis according to the aforementioned control method, ultimately driving the left-arm industrial robot 3 and the right-arm industrial robot 5 to operate collaboratively. The end effector 2 acquires and grips the parts, and the welding system 6 performs the welding task.

[0088] Reference Figure 8 The AI ​​intelligent computing unit in the dual-arm robot welding control system of the present invention includes the following modules:

[0089] Assembly trajectory planning module: Used to plan and output the motion trajectory of one arm of the dual-arm robot according to the assembly task, to complete the assembly task of parts, including:

[0090] Assembly robot arm trajectory generation unit: Generates the assembly motion trajectory of the robot arm based on the current actual parts and assembly task planning; in this embodiment, the left arm is used as the assembly robot arm.

[0091] Assembly trajectory optimization unit: Utilizes smoothing, speed planning, and collision detection algorithms to optimize the robotic arm's assembly motion trajectory.

[0092] Welding trajectory planning module: Used to plan and output the motion trajectory of the other arm of the dual-arm robot according to the welding task, to complete the welding task of the parts, including:

[0093] Welding robot arm trajectory generation unit: Based on the current real parts and welding task planning, it generates the welding motion trajectory of the robot arm; in this embodiment, the right arm is used as the welding robot arm.

[0094] Welding trajectory optimization unit: Adjusts the smoothness, continuity, speed, and acceleration parameters of the robotic arm's welding motion trajectory to meet welding process requirements.

[0095] Identification module: used to identify parts and welds, including:

[0096] Component identification unit: Used to identify and classify components by applying image processing and information labeling methods to the acquired component images, shape and size feature information.

[0097] Weld seam recognition unit: Based on the visual perception system, it detects the position, shape and size of weld seams in real time, identifies the weld seam information of real parts, and provides a basis for welding trajectory planning.

[0098] Control module: Used to output control commands to the dual-arm robot based on the control strategy results, including:

[0099] Robot posture control unit: Based on the motion trajectory planned for the assembly and welding tasks and the robot's real-time feedback information, it controls the posture of the dual-arm robot and performs posture tracking by combining the dynamic model and external disturbance factors.

[0100] Welding process control unit: Sets process parameters such as welding current, voltage, speed, and wire feed speed, and monitors and adjusts them in real time to ensure the stability and consistency of welding process.

[0101] Figure 9 and Figure 10 This is a schematic diagram illustrating an example of controlling a dual-arm robot to assemble and weld complex non-standard workpieces, specifically fan blades. First, using the 3D model feature information of the fan blade workpiece, and based on the component forming characteristics, the weld seam position is determined according to the welding process of the target workpiece. Using the weld seam as a dividing line, independent virtual model components are separated, and component features are extracted. Simultaneously, corresponding real components are prepared. Then, based on reinforcement learning, a control strategy is generated in a simulation environment based on the dual-arm robot parameters and the virtual model components, planning the assembly and welding tasks, and generating the dual-arm motion trajectory. After determining the welding trajectory's rationality by considering obstacle avoidance and welding process parameters, a robotic arm control program is generated. The operation of the dual-arm robot is further verified on the robot operating system (ROS) platform, deployed to the dual-arm robot system for testing, and then used to assemble and weld batches of real components.

[0102] like Figure 9 As shown, in the assembly task planning, the central component of the fan blade workpiece serves as the base component. According to the assembly task plan, the dual-arm robot identifies and controls its left arm to acquire the first blade, and then guides the first blade to dock with the central component, completing the assembly. The left arm then holds the irregularly shaped first blade in a stable position, awaiting welding by the right arm. In this process, the dual-arm robot uses a visual perception system to identify components based on features such as shape and size. Simultaneously, the welding task planning controls the robot to identify the weld seams of the real components, matching them with the weld seams obtained from the simulation environment. It then calls up the welding process parameter package and controls the right arm to move the welding system according to the welding trajectory, welding the first blade onto the central component, thus completing the assembly and welding task of the first blade. Figure 10 This diagram illustrates the assembly and welding of subsequent blades; the control process is the same as that for the first blade.

[0103] After welding is completed, the dual-arm robot uses a visual perception system to compare the completed task. Once it confirms that the 3D model of the assembled and welded workpiece matches the target fan blade workpiece, it completes all welding process tasks.

Claims

1. A welding control method for a dual-arm robot, characterized in that: Controlling a dual-arm robot to complete welding tasks on non-standard workpieces includes the following steps: 1) Call the 3D data of the target workpiece, and based on the preset welding process parameters of the target workpiece, perform structural analysis and disassembly on the 3D model of the target workpiece to generate the independent parts required to form the welded body, and prepare the corresponding real parts. 2) Based on reinforcement learning, the optimal action strategy for assembling and welding parts is learned by using a dual-arm robot as the intelligent agent. One arm of the dual-arm robot performs the assembly task and the other arm performs the welding task. Combining the structural analysis and disassembly results of the 3D model, the assembly task and welding task are planned as the control strategy results. The assembly task planning includes the assembly sequence and pose of the corresponding parts and the assembly action trajectory. The welding task planning includes the welding action trajectory of the pose of the corresponding parts. 3) Generate the motion trajectory of the dual-arm robot based on the control strategy results. One arm of the dual-arm robot acquires real parts and adjusts the pose of the real parts according to the assembly task planning. The other arm controls the welding action to perform welding on the current real parts based on the welding task planning. Through the iterative cycle of assembly-welding, multiple real parts are welded in sequence according to the assembly order and pose to obtain the finished product of the target workpiece.

2. The welding control method for a dual-arm robot according to claim 1, characterized in that: In step 3), the control strategy results are simulated and verified using the robot operating system ROS platform. After the simulation verification is successful, the parameters and behavioral data obtained from the simulation are deployed and debugged on the real dual-arm robot system. After the debugging is completed, welding commands are issued to the dual-arm robot system to drive the two arms to perform a collaborative assembly and welding operation sequence in the real physical environment.

3. A welding control method for a dual-arm robot according to claim 1 or 2, characterized in that: Step 2) Based on reinforcement learning, the optimal strategy for assembling and welding parts is learned using a dual-arm robot as the intelligent agent, including: Define a state description function, using the TCP pose, velocity, and welding assembly state of the robotic arm end effector as state variables, to construct the robot action state function for the assembly and welding process, which serves as the basis for judging reinforcement learning training. Define an assembly reward function to quantitatively evaluate the quality of the assembly status of parts, the TCP pose of the robot end effector, and the speed during the assembly and welding process; Using a dual-arm robot as the intelligent agent, the optimal motion strategy for assembling and welding the parts of the dual-arm robot is reinforced through learning, and the assembly task plan and welding task plan are output as the control strategy results.

4. A welding control method for a dual-arm robot according to claim 1 or 2, characterized in that: The control strategy results, for the assembly task planning, include the following steps: Step A1: Obtain the feature information of the virtual model parts and the feature information of the real parts from the 3D model disassembly in Step 1), match the feature information of the virtual model parts with the feature information of the real parts, and establish the correspondence between the virtual model parts and the real parts. Step A2: Determine the base components that form the basis for assembly, generate component assembly strategies for the virtual model components, and perform robot kinematic modeling. Mathematically describe the motion relationships of the robotic arm driving the various components during the assembly process, including the joint motion and pose transformation of the robotic arm, to control the robot's motion trajectory. Step A3: Optimize the generated component assembly strategy and robot kinematic model using reinforcement learning. Through the interaction between the agent and the environment, adjust the assembly strategy according to the assembly reward function. The reinforcement learning results are verified using Sim2real, applying the learning results in the simulated environment to the real scene. If the verification is successful, the assembly task plan is output for assembling real components. If the verification fails, return to the reinforcement learning stage to further adjust and optimize the strategy until the verification is successful. Step A4: Obtain the actual base components, establish a workpiece coordinate system based on the base components to provide reference coordinates for the assembly operation, and combine the correspondence between the simulated model components and the actual components to obtain the assembly motion trajectory of the dual-arm robot from the assembly task planning.

5. The welding control method for a dual-arm robot according to claim 4, characterized in that: The specific steps for establishing the correspondence between virtual model components and real components are as follows: Extract the feature information of virtual model parts decomposed from the 3D model in step 1), including size, shape and surface texture, to represent the shape and position features of the parts; 3D geometric feature information of real parts, including size, shape and surface texture, can be obtained by scanning or measuring. Based on the acquired feature information, the virtual model parts and real parts are classified and labeled multiple times. Each classification and labeling is based on different feature dimensions, including size, shape and part function. Through multiple classifications and labels, the recognition accuracy of individual part features is improved. Finally, based on the categorized and summarized components, the virtual model components are compared with the real components in turn to achieve alignment and matching between the virtual model components and the real components, which is used for the robot to identify the real components.

6. A welding control method for a dual-arm robot according to claim 1 or 2, characterized in that: The control strategy results, for welding task planning, include the following steps: Step B1: Obtain information about the dual-arm robot and the parts to be welded, including the size, shape, material of the parts, and motion specifications of the welding robot. Step B2: Establish a welding simulation environment to simulate the relative positional relationship and operating conditions of the welding robot and the parts. Then, identify the weld seams of the parts in the simulation environment and determine their position and shape characteristics. Step B3: Generate a welding path based on the identified weld, covering the entire length of the weld and conforming to the process specifications. Then, perform forward and inverse kinematics solutions to calculate the pose of the welding torch and the motion commands of the robot joints. After that, use offline simulation to determine whether the welding posture is reasonable. If it is not reasonable, return to the kinematics solution step for adjustment until the simulation results meet the requirements, and output the welding trajectory corresponding to the identified weld. Meanwhile, the weld information of the real parts during the assembly process is obtained and matched with the welds identified in the simulation environment. Based on Sim2real, the robot's motion trajectory in the real environment is determined. Combining the kinematic characteristics of the robotic arm and the welding process parameters, the welding process parameter package is called to obtain the planned current welding trajectory. Step B4: Based on the current welding trajectory planned in step B3, complete the welding task for the current actual component and proceed to the welding of the next component; Step B5: Repeat steps B3 to B4 until all parts are assembled and welded.

7. A welding control method for a dual-arm robot according to claim 6, Its characteristic is that step B3, which involves obtaining weld information for the actual components during the assembly process, specifically includes: B3.

1. Acquire images containing the weld seams of the actual components to be welded using a sensing camera; B3.2 Preprocess the image, including converting the image to grayscale and using a denoising algorithm to remove random noise from the image; B3.3 Weld Feature Extraction: Using image processing technology, key features of the weld, including edge contours and texture patterns, are extracted from the preprocessed image. The Segformer segmentation model is then used to segment the image, and the segmented welds are identified and analyzed. This includes quantitative analysis of the weld geometry, surface quality, and defect features to assess whether the weld meets the process standards and quality requirements, and to determine whether welding should proceed. B3.4 Output the results of weld identification and analysis, including the location, size and quality grade of the weld, to be used in conjunction with the welds identified in the simulation environment to determine the robot's motion trajectory in the real environment based on Sim2real.

8. A welding control method for a dual-arm robot according to claim 1 or 2, characterized in that the three-dimensional geometric data of the target workpiece is retrieved from the process database of the target workpiece, or established online through three-dimensional reconstruction technology.

9. A welding control system for a dual-arm robot, characterized in that: The invention includes a movable leveling platform, a dual-arm robot, a vision perception system, a welding system, and an AI intelligent computing unit. The dual-arm robot includes a left arm and a right arm. The welding system is located at the end of the right arm via an end-effector. The AI ​​intelligent computing unit stores at least one instruction or at least one program. When the at least one instruction or the program is executed, it implements the welding control method of the dual-arm robot according to any one of claims 1-8.

10. A welding control system for a dual-arm robot according to claim 9, characterized in that: The AI ​​intelligent computing unit includes the following modules: Assembly trajectory planning module: Used to plan and output the motion trajectory of one arm of the dual-arm robot according to the assembly task, to complete the assembly task of parts, including: Assembly robot arm trajectory generation unit: Generates the robot arm assembly motion trajectory based on the current real parts and assembly task plan; Assembly trajectory optimization unit: Utilizes smoothing, speed planning, and collision detection algorithms to optimize the robotic arm's assembly motion trajectory; Welding trajectory planning module: Used to plan and output the motion trajectory of the other arm of the dual-arm robot according to the welding task, to complete the welding task of the parts, including: Welding robot arm trajectory generation unit: Generates the welding motion trajectory of the robot arm based on the current real parts and welding task planning; Welding trajectory optimization unit: Adjusts the smoothness, continuity, speed, and acceleration parameters of the robotic arm's welding motion trajectory to meet welding process requirements; Identification module: used to identify parts and welds, including: Component identification unit: Used to identify and classify components by applying image processing and information tagging methods to the acquired component images, shape, and size feature information; Weld seam recognition unit: Based on a visual perception system, it detects the position, shape and size of weld seams in real time, identifies weld seam information of real parts, and provides a basis for welding trajectory planning; Control module: Used to output control commands to the dual-arm robot based on the control strategy results, including: Robot posture control unit: Based on the motion trajectory planned for the assembly and welding tasks and the real-time feedback information from the robot, it controls the posture of the dual-arm robot and performs posture tracking by combining the dynamic model and external disturbance factors; Welding process control unit: Sets process parameters such as welding current, voltage, speed, and wire feed speed, and monitors and adjusts them in real time to ensure the stability and consistency of welding process.

Images