Multi-axis robot cluster cooperative control system and method for self-adaptive repair of vehicle body

Abstract

The invention discloses a multi-axis robot cluster cooperative control system and method for self-adaptive repair of a vehicle body, and relates to the technical field of vehicle body repair. The multi-axis robot cluster cooperative control system comprises a multi-source fusion sensing and double-matching module, a multi-axis robot cluster cooperative control module, a multi-axis robot cluster cooperative control module and a multi-axis robot cluster cooperative control module, damage type and grade analysis and accurate matching of a repair process and a robot cluster are realized; the digital twin mirror image construction and pre-optimization module is used for constructing a real-time synchronous virtual mirror image and completing pre-simulation and parameter debugging; the distributed collaborative decision-making and fault-tolerant control module is used for realizing dynamic splitting and distribution of tasks and rapid takeover of faults; and the force position compliance execution and tool adaptation module drives the robot to carry out compliance operation and adaptively adjust terminal parameters. According to the system, through module cooperation, the damage identification precision and the repairing technology adaptability are greatly improved, the multi-robot cooperation stability is guaranteed, the repairing interruption risk is reduced, the repairing quality and efficiency are improved, and self-adaptive precise repairing of the vehicle body is achieved.

Description

Technical Field

[0001] This invention relates to the field of vehicle body repair technology, specifically to a multi-axis robot cluster collaborative control system and method for adaptive vehicle body repair. Background Technology

[0002] With the rapid development of the automotive industry and the continuous increase in car ownership, damage to vehicle bodies due to collisions, scratches, and other factors during use is becoming increasingly common, making vehicle body repair an important part of the automotive aftermarket. The quality of vehicle body repair directly affects the vehicle's appearance, structural safety, and subsequent performance. Therefore, a multi-axis robot cluster collaborative control system and method for adaptive vehicle body repair are needed.

[0003] Existing technology, such as the invention patent application CN112337677A, discloses a robotic automotive painting control system, relating to the field of automotive manufacturing technology. This invention includes: a painting robot, a car body conveying device, and a vision sensing device; the car body conveying device conveys the car body to the painting area and the inspection area; the vision sensing device performs sensing; the vision sensing device detects the car body position and transmits this information to the analysis module of the painting robot; the analysis module analyzes and calculates the current coordinate information and painting path of the car body based on its position; the painting robot paints the car body according to the current coordinate information and the painting path. This invention improves the painting effect and efficiency by using a vision sensing device to monitor the car body position and transmit this information to the analysis module of the painting robot; the analysis module analyzes and calculates the current coordinate information and the painting path of the car body based on its position; and the painting robot paints the car body according to the current coordinate information and the painting path.

[0004] Regarding the above-mentioned solutions, the applicant of this invention has found that the above-mentioned technologies have at least the following technical problems: 1. In the field of vehicle body repair, existing technologies have significant shortcomings in terms of the accuracy of damage identification and process planning. Most current repair technologies use a single sensing method for damage detection, which makes it difficult to achieve comprehensive coverage identification of surface scratches, dents, and internal hidden defects of the vehicle body, resulting in a one-sided damage assessment; at the same time, the determination of process schemes relies heavily on human experience, lacking a standardized damage level quantification system and an automated process matching mechanism, which not only prolongs the repair preparation cycle, but also easily leads to a mismatch between the repair process and the actual damage due to human judgment bias, affecting the final repair effect.

[0005] 2. The stability and anti-interference capability of multi-device collaborative operation are the main bottlenecks of existing automated vehicle body repair technologies. When using multiple robots or multiple devices for joint repair, existing technologies generally lack mature collaborative scheduling mechanisms, which easily lead to problems such as conflicting work paths and asynchronous action sequences, resulting in low work efficiency. In addition, the fault response system is imperfect. When any work device experiences parameter drift or mechanical failure, it is difficult to quickly complete fault diagnosis and task reconnection, often causing the repair process to be interrupted. This problem is particularly prominent in multi-stage repair scenarios of complex damage, seriously affecting the continuity and reliability of repair operations.

[0006] 3. Insufficient adaptation of actuator parameters and precise control of the work process hinders the improvement of vehicle body repair quality. The parameters of the actuators in existing repair equipment are mostly fixed, making dynamic adjustment difficult based on vehicle body material and damage level. Furthermore, the coordination and control precision between contact force and work position are limited. In core processes such as sanding and paint touch-up, problems such as excessive contact force leading to secondary damage to the vehicle body, or insufficient contact force resulting in incomplete repair, are prone to occur. Moreover, the lack of effective real-time feedback and parameter correction loops fails to guarantee the process stability of the entire repair process. Summary of the Invention

[0007] To address the aforementioned technical shortcomings, the present invention aims to provide a multi-axis robot cluster collaborative control system and method for adaptive vehicle body repair.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: In the first aspect, the present invention provides a multi-axis robot cluster collaborative control system for adaptive vehicle body repair, including the following modules: a multi-source fusion perception and dual matching module: used to deploy a multi-modal sensor array to collect target vehicle body damage data, robot operation data and environmental data, thereby analyzing the vehicle body damage type and level corresponding to the target vehicle body damage, and then matching and adapting the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the current damage.

[0009] Digital twin image construction and pre-optimization module: It is used to construct a virtual image that is synchronized with the physical repair scene in real time based on the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the target vehicle body damage site, and to carry out pre-simulation optimization of the repair process, path conflict simulation and parameter debugging.

[0010] Distributed collaborative decision-making and fault-tolerant control module: It is used to receive the optimal repair process plan and the optimal multi-axis robot cluster type at the target vehicle body damage site, so as to dynamically split and allocate the repair task, and then realize rapid fault response and task takeover through the three-level fault-tolerant architecture of virtual image.

[0011] Force-position compliant execution and tool adaptation module: It is used to drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site based on the issued optimal repair process plan and optimal multi-axis robot cluster type collaborative control instructions, and realize the adaptive adjustment of end effector parameters.

[0012] The present invention provides a multi-axis robot cluster collaborative control method for adaptive vehicle body repair in a second aspect, comprising the following steps: Step 1, multi-source fusion perception and dual matching: deploying a multi-modal sensor array to collect target vehicle body damage data, robot operation data and environmental data, thereby analyzing the vehicle body damage type and level corresponding to the target vehicle body damage, and then matching and adapting the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the current damage.

[0013] Step 2: Digital Twin Image Construction and Pre-optimization: Based on the optimal repair process and optimal multi-axis robot cluster type corresponding to the target vehicle body damage, a virtual image that is synchronized with the physical repair scene in real time is constructed, and pre-simulation optimization of the repair process, path conflict simulation, and parameter debugging are carried out.

[0014] Step 3: Distributed collaborative decision-making and fault-tolerant control: Receive the optimal repair process plan and the optimal multi-axis robot cluster type for the target vehicle body damage, thereby dynamically splitting and allocating the repair task, and realizing rapid fault response and task takeover through a three-level fault-tolerant architecture of virtual mirror.

[0015] Step 4, Force-position compliant execution and tool adaptation: Based on the issued optimal repair process plan and the optimal multi-axis robot cluster type, the coordinated control instructions generated drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site, and achieve adaptive adjustment of end effector parameters.

[0016] The beneficial effects of this invention are as follows: 1. In this embodiment, the design of multi-source fusion sensing and dual matching modules significantly improves the accuracy of vehicle body damage identification and the rationality of process matching, effectively solving the problems of one-sided damage assessment and reliance on manual experience in traditional repair technologies. The solution uses a multimodal sensor array to achieve comprehensive collection of vehicle body damage, robot operation, and environmental data. Combined with a standardized damage type and level quantitative analysis system, it can accurately identify various types of damage such as scratches, dents, paint peeling, and hidden cracks, avoiding the problem of missed detection of internal defects caused by a single sensing method. At the same time, based on the damage type and level, the optimal repair process scheme and robot cluster type are automatically matched, replacing the manual scheme formulation process. This not only shortens the repair preparation cycle but also ensures the accurate adaptation of the process scheme to the actual damage, laying the foundation for high-quality repair.

[0017] 2. This embodiment of the solution, leveraging a digital twin mirror construction and pre-optimization module and a distributed collaborative decision-making and fault-tolerant control module, significantly improves the stability and efficiency of multi-axis robot cluster collaborative operations, effectively overcoming the bottlenecks of path conflicts, timing asynchrony, and insufficient fault response capabilities in traditional multi-device collaborative repair. By constructing a virtual mirror synchronized with the physical scene in real time, the solution can conduct pre-simulation optimization and path conflict simulation of the repair process in advance, avoiding operational risks and ensuring the coordination of multi-robot motion trajectories. Simultaneously, the dynamic task splitting and allocation mechanism based on robot workspace, load capacity, and operating status can fully utilize the operational efficiency of each robot, improving overall repair efficiency. Furthermore, the three-level fault-tolerant architecture design enables rapid fault verification and task takeover, ensuring seamless continuity of the repair process when a single robot malfunctions, significantly improving operational continuity and reliability.

[0018] 3. This embodiment of the solution significantly improves the execution accuracy and process stability of vehicle body repair through precise control of the force-position compliant execution and tool adaptation module, effectively solving the problems of fixed parameters and insufficient precision in force-position coordination control of traditional repair equipment. The solution uses real-time collected motion, force control, and coordination data to determine the normality of the operation. Combined with an adaptive adjustment mechanism for the end effector parameters, it can dynamically adjust key parameters such as grinding head grit size, nozzle diameter, and operating speed according to different anomalies such as motion accuracy and force control stability. Simultaneously, through deviation compensation adjustment and precision verification closed-loop, it ensures that the end effector parameters are always in an optimal state, avoiding secondary damage or incomplete repairs such as over-grinding or uneven paint application. The application of force-position compliant coordination control technology further ensures precise coordination of contact force and position during the repair process, significantly improving the consistency and stability of repair quality. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the system structure connection of the present invention.

[0021] Figure 2 This is a schematic diagram of the implementation steps of the method of the present invention. Detailed Implementation

[0022] 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.

[0023] Examples of embodiments of the present invention Figure 1 As shown, the multi-axis robot cluster collaborative control system for adaptive vehicle body repair includes the following modules: multi-source fusion perception and dual matching module, digital twin image construction and pre-optimization module, distributed collaborative decision-making and fault-tolerant control module, and force-position compliant execution and tool adaptation module.

[0024] The digital twin image construction and pre-optimization module is connected to the multi-source fusion perception and dual matching module and the distributed collaborative decision-making and fault-tolerant control module, respectively. The distributed collaborative decision-making and fault-tolerant control module is connected to the force-position compliant execution and tool adaptation module.

[0025] Multi-source fusion perception and dual matching module: used to deploy multimodal sensor arrays to collect target vehicle body damage data, robot operation data and environmental data, thereby analyzing the type and level of vehicle body damage corresponding to the target vehicle body damage, and then matching the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the current damage.

[0026] In one specific embodiment, the deployment of the multimodal sensor array collects multi-dimensional data on target vehicle body damage, robot operation, and environment. The specific collection process is as follows: The multimodal sensor array is deployed according to the principle of full coverage of the target vehicle body. Among them, 3D LiDAR and binocular vision sensors are symmetrically deployed around the repair station, 1.5-2m away from the vehicle body; terahertz radar and ultrasonic sensors are deployed next to the robot end effector; UWB positioning sensors are embedded in each robot body; force sensors are integrated into the end effector; and temperature, humidity, and light sensors are deployed at the four corners of the repair station. All sensors are connected to the edge computing node through shielded cables to collect target vehicle body damage data, robot operation data, and environmental data in real time.

[0027] It should be noted that the long-distance symmetrical deployment of 3D LiDAR and binocular vision sensors enables full-area scanning of the vehicle body, avoiding missed detection of local damage. Terahertz radar and ultrasonic sensors are deployed next to the end effector, enabling close-range and precise detection of internal damage details. Embedded / integrated deployment of UWB positioning sensors and force sensors can capture the robot's operating status in real time. Temperature, humidity, and light sensors are deployed at the four corners of the workstation to comprehensively perceive environmental interference factors. The deployment positions and distances of various sensors are designed based on the core requirements of full-area coverage, precise detection, and status synchronization. In terms of data transmission assurance, shielded cables are used to connect to edge computing nodes, which can effectively resist electromagnetic interference generated by welding, grinding, and other operations in the repair station, ensuring the integrity and transmission stability of the collected data and laying the foundation for the accuracy of subsequent data processing. In terms of data collaboration value, the three types of data collected simultaneously—vehicle damage, robot operation, and environment—are not isolated but form a collaborative and complementary data source. Vehicle damage data provides the core basis for repair plan formulation, robot operation data provides support for collaborative scheduling and fault diagnosis, and environmental data provides a reference for adaptive adjustment of process parameters. The collaborative collection of these three types of data enables precise perception of all elements of damage, equipment, and environment.

[0028] The target vehicle body damage data includes: three-dimensional coordinates of the damaged area, damage depth, damage area, curvature of the damaged area, and dimensions of internal defects in the vehicle body; the robot operation data includes: robot body pose coordinates, joint rotation speed, joint temperature, end effector contact force, and robot load rate; the environmental data includes: ambient temperature, ambient humidity, and light intensity.

[0029] In a specific embodiment, the analysis of the damage type and level corresponding to the damage to the target vehicle body is carried out as follows: Damage type identification analysis: The quantitative parameters of the damage to the target vehicle body are standardized. If the damage depth of the target vehicle body is <0.5mm and there are no internal defects, it is judged as a scratch; if the damage depth of the target vehicle body is ≥0.5mm, the area is ≥5cm² and there are no internal defects, it is judged as a dent; if the surface coating of the target vehicle body is missing and there is no obvious depth damage, it is judged as paint peeling; if the terahertz radar detects an internal defect size ≥1cm, it is judged as a hidden crack.

[0030] Damage Level Quantitative Analysis: Damage depth is the core quantitative indicator, combined with damage area and internal defect size for level determination. If the target vehicle body meets the following conditions: damage depth ≥ 0.1mm and < 0.3mm, damage area < 2cm², and no internal defects, it is classified as Level 1; if the target vehicle body meets the following conditions: damage depth ≥ 0.3mm and < 0.5mm, damage area ≥ 2cm² and < 5cm², and no internal defects, it is classified as Level 2; if the target vehicle body meets the following conditions: damage depth ≥ 0.5mm and < 0.8mm, damage area ≥ 5cm² and < 10cm², and no internal defects, it is classified as Level 3; damage depth ≥ 0.8mm and < 1.0mm, damage area ≥ 10cm² and < 20cm², and internal defect size < 5cm, it is classified as Level 4; damage depth ≥ 1.0mm, damage area ≥ 20cm², and internal defect size ≥ 5cm, it is classified as Level 5.

[0031] It should be noted that the parameter standardization process specifically adopts a normalization algorithm based on the vehicle body material (such as steel plate, aluminum alloy), which converts the raw data such as damage depth and area collected by different sensing devices into a unified dimension (mm, cm²), and eliminates measurement deviations caused by environmental factors such as temperature and light, ensuring data comparability. The judgment rules clearly distinguish the boundaries of various types of damage through the exclusive design of "core features + exclusion conditions". For example, the limitation of "no obvious depth damage" in the paint peeling judgment excludes misjudgments in the scenario where scratches, dents and paint peeling coexist. The threshold of "internal defect size ≥ 1cm" in the hidden crack judgment is to exclude false defects caused by tiny impurities. The setting of the grading thresholds is closely adapted to the vehicle body repair process capabilities. The thresholds of level 1-2 minor damage correspond to simple manual replacement processes, while the thresholds of level 3-5 moderate to severe damage match complex processes such as pulling and welding by robot clusters. Moreover, there is no overlap between the thresholds of each level, avoiding ambiguity in the grade judgment. At the same time, the introduction of internal defect size makes the grading more in line with the needs of vehicle body structural safety repair.

[0032] It should also be noted that the quantitative parameters of target vehicle body damage are standardized to eliminate differences in parameter dimensions across different vehicle models and damage measurement scenarios. This ensures the comparability and consistency of core parameters such as damage depth, area, and internal defect size, avoiding judgment biases caused by inconsistent parameter measurement standards. From the design basis of the judgment rules, the judgment conditions for various damage types closely align with actual vehicle body damage characteristics—the distinction between scratches and dents focuses on the core differences in depth and area; paint peeling judgment highlights the core characteristic of missing surface coating; and latent cracks rely on the internal detection advantages of terahertz radar to pinpoint internal defects. The classification is based on damage depth as the core, combined with the quantification of area and internal defects. The indicators form a gradient judgment system from shallow to deep and from surface to core, covering both minor damage and serious damage and internal defects, which meets the actual needs of vehicle body repair processes for accurate classification of damage degree. From the perspective of the application value of the analysis results, the accurate analysis of damage type and level is the core basis for matching the optimal repair process scheme and robot cluster type. Different types and levels of damage correspond to different repair processes and equipment requirements. For example, minor scratches are suitable for simple grinding and polishing processes and single-axis robots, while severe dents require complex pull-out correction, welding processes and multi-axis collaborative robots. The analysis results directly determine the rationality and repair efficiency of the repair scheme and are a key preliminary step to achieve adaptive vehicle body repair.

[0033] In a specific embodiment, the matching process is as follows: A1. If the target vehicle body damage type is scratch and the level is 1, then the optimal repair process is: light grinding + local polishing, and the optimal multi-axis robot cluster type is single-axis robot cluster.

[0034] If the target vehicle body damage type is scratch and the level is 2, the optimal repair process is: moderate polishing + local paint touch-up + low temperature curing, and the optimal multi-axis robot cluster type is a three-axis robot cluster.

[0035] If the target vehicle body damage type is dent and the level is 3, the optimal repair process is: dent pull-out correction + fine grinding + paint touch-up + curing, and the optimal multi-axis robot cluster type is a six-axis robot cluster.

[0036] If the target vehicle body damage type is dent and the level is 4, the optimal repair process is: multi-point pull-out correction + sheet metal shaping + fine grinding + layered paint touch-up + constant temperature curing. The optimal multi-axis robot cluster type is multi-axis collaborative robot cluster.

[0037] If the target body damage type is dent and the level is 5, the optimal repair process is: sheet metal cutting and replacement + welding + grinding + overall paint touch-up + high temperature curing. The optimal multi-axis robot cluster type is multi-axis collaborative robot cluster.

[0038] If the target vehicle body damage type is paint peeling and the level is 1-2, the optimal repair process is: surface cleaning + local paint touch-up + low temperature curing. The optimal multi-axis robot cluster types are single-axis and three-axis robot clusters, respectively.

[0039] If the target vehicle body damage type is paint peeling and the level is 3-5, the optimal repair process is: surface sanding and paint removal + overall paint touch-up + curing. The optimal multi-axis robot cluster types are six-axis, multi-axis collaborative, and multi-axis collaborative robot cluster, respectively.

[0040] If the target vehicle body damage type is a latent crack and the level is 3-5, the optimal repair process is: crack repair agent filling + curing + surface grinding and polishing. The optimal multi-axis robot cluster type is a six-axis or multi-axis collaborative robot cluster.

[0041] It's important to note that the matching system operates on the core logic of determining process requirements based on damage characteristics and matching process requirements with equipment capabilities, achieving precise linkage between damage, process, and equipment. For different damage types and their core repair needs—such as scratches requiring focus on surface smoothness restoration, dents requiring focused morphological correction, and hidden cracks requiring ensuring structural sealing—differentiated process chains are designed accordingly. Simultaneously, based on quantitative indicators of damage levels, robot clusters with corresponding operational precision, degrees of freedom, and load capacity are matched. For example, level 1 minor scratches only require a single-axis robot to complete a single grinding and polishing process, balancing precision and cost; level 5 severe dents involve multiple complex processes, requiring multi-axis collaborative robots to achieve process connection and collaborative operation. The adaptation of processes and equipment also fully considers actual engineering scenarios. For instance, process parameters such as low-temperature curing and constant-temperature curing match the temperature control module capabilities of the robot cluster, and processes such as layered painting and overall painting are adapted to the trajectory planning precision of the robots. Furthermore, the matching process balances repair efficiency and economy, avoiding resource waste from adapting high-level equipment to minor damage and preventing the issue of substandard quality from low-level equipment handling complex damage. The matching results not only provide accurate process and equipment input parameters for subsequent digital twin pre-simulation optimization, but also lay the foundation for task splitting and allocation for distributed collaborative decision-making, which is a key hub for realizing the full automation and precision of the adaptive repair process of the vehicle body.

[0042] Digital twin image construction and pre-optimization module: It is used to construct a virtual image that is synchronized with the physical repair scene in real time based on the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the target vehicle body damage site, and to carry out pre-simulation optimization of the repair process, path conflict simulation and parameter debugging. In a specific embodiment, the construction of the virtual mirror that is synchronized with the physical repair scene in real time is carried out as follows: B1. Basic model import and initialization: Import the matched optimal multi-axis robot cluster 3D model and the target vehicle body 3D basic model, and accurately map the damaged area of ​​the vehicle body model according to the 3D coordinates, damage depth, damage area, curvature of the damaged area and the size of the internal defects of the vehicle body, and complete the initial model construction.

[0043] B2. Scene Element Integration: Integrate the environmental parameters of the repair station, the repair tool model, and the key process parameters in the optimal repair process plan into the virtual scene to achieve full digital mapping of the physical repair scene.

[0044] B3. Real-time synchronization mechanism establishment: A virtual-physical data transmission channel is established through edge computing nodes to synchronize the robot's real-time pose, end-effector contact force, and target vehicle body status data at a 20ms update cycle.

[0045] It should be noted that in the basic model import and initialization stage, accurately mapping the vehicle body damage area is the core key. By importing and matching the robot and vehicle body basic models, combined with the damage quantification parameters collected earlier, the damage area is digitally reproduced, providing a realistic damage benchmark for subsequent pre-simulation of the repair process and avoiding simulation failure due to deviations between the model and actual damage. A comprehensive system linking equipment, tools, processes, and environment is constructed. The collaborative integration of repair station environmental parameters, repair tool models, and key parameters of the optimal repair process enables linked simulation of process parameters with the environment and equipment status, accurately predicting the repair effect under different parameter combinations. Real-time synchronization is also crucial. The 20ms update cycle design fully considers the dynamic response requirements of robot collaborative operations. By building a high-bandwidth, low-latency data transmission channel through edge computing nodes, it can quickly synchronize core operational data such as robot pose and end-effector contact force, ensuring real-time consistency between the virtual image and the physical scene. At the same time, the local data processing capabilities of the edge computing nodes can reduce the transmission pressure on the cloud and improve the stability of data synchronization. This virtual image not only provides a digital platform for path conflict simulation and parameter debugging, but also provides real-time scene data support for subsequent distributed collaborative decision-making and three-level fault-tolerant control. It is the core digital carrier for achieving precise control of the repair process.

[0046] Distributed collaborative decision-making and fault-tolerant control module: It is used to receive the optimal repair process plan and the optimal multi-axis robot cluster type at the target vehicle body damage site, so as to dynamically split and allocate the repair task, and then realize rapid fault response and task takeover through the three-level fault-tolerant architecture of virtual image.

[0047] In a specific embodiment, the analysis of dynamic splitting and allocation of repair tasks and self-optimization of repair parameters is as follows: C1. Task splitting analysis: The overall repair task is split into several sub-tasks, where a single damaged area corresponds to an independent sub-task, and adjacent damaged areas are merged into collaborative sub-tasks.

[0048] C2. Dynamic Allocation Analysis: Based on the optimal multi-axis robot cluster's workspace, load capacity, and current operating status, precise allocation of sub-tasks is achieved. The specific allocation logic is as follows: First, workspace matching: Extract the 3D coordinates of the target vehicle body damage area corresponding to each sub-task, compare them with the workspace range of each robot, and filter out a set of candidate robots whose workspace can cover the damage area, excluding robots whose workspace cannot reach it. Second, load capacity adaptation: Obtain the rated load parameters of the candidate robots, and combine them with the process load requirements of each sub-task. Specifically, the load requirement for polishing tasks = process force + tool weight, and the load requirement for painting tasks = paint gun weight. + Remaining weight of paint, the load requirement for the drawing and correction task = drawing force + weight of drawing tool, the load requirement for the welding task = weight of welding torch + remaining weight of welding wire, the load requirement for the polishing task = polishing force + weight of polishing wheel, the load requirement for the crack repair agent filling task = weight of filling torch + remaining weight of repair agent, screen out the adaptable robots whose rated load is ≥ 1.2 times the process load of the sub-task; the third step, priority ranking of the current operating status: collect the current operating status parameters of the adaptable robots, including the current load rate, remaining power and fault history, and rank the adaptable robots according to the priority rule of lowest load rate > highest remaining power > no fault history; assign the sub-task to the robot with the best ranking.

[0049] In a specific embodiment, the three-level fault-tolerant architecture enables rapid fault response and task takeover. The specific implementation process is as follows: obtain the optimal repair process scheme corresponding to the target vehicle body, collect the actual process parameters of each robot when executing the corresponding sub-task in real time, and the actual process parameters correspond one-to-one with the process parameters required by the optimal repair process scheme of the sub-task; compare the collected actual process parameters with a preset threshold, which is a parameter standard for judging the normality of operation.

[0050] If the actual process parameters of a robot exceed the preset threshold, a fault is triggered and the fault information is immediately reported to the area controller. After receiving the fault information, the area controller completes fault verification within 50ms. If there is an idle and suitable robot in the area, the unfinished sub-tasks of the faulty robot are immediately assigned to that robot, and the task planning in the area is updated. If the area controller cannot complete the task takeover, it immediately reports to the central controller. The central controller updates the global repair plan synchronously and schedules suitable robots from other areas to complete the task takeover.

[0051] Force-position compliant execution and tool adaptation module: It is used to drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site based on the issued optimal repair process plan and optimal multi-axis robot cluster type collaborative control instructions, and realize the adaptive adjustment of end effector parameters.

[0052] In a specific embodiment, the multi-axis robot cluster after drive matching performs force-position compliant collaborative operation on the target vehicle body damage site. The specific execution process is as follows: collecting core analysis data: robot motion-related data includes the pose coordinates of each robot body, joint rotation speed, actual value of motion trajectory and trajectory tracking error; force control-related data includes the real-time value of end effector contact force and contact force stability accuracy; collaboration-related data includes multi-robot operation timing synchronization error, data interaction delay between robots and collaborative operation accuracy; status feedback data includes operation progress, execution status and sub-task completion time.

[0053] It should be noted that in the robot motion-related data, the body pose coordinates are captured in real time by UWB positioning sensors embedded in each robot body, the joint rotation speed is collected by the rotation encoder built into the robot joint, the actual value of the motion trajectory is calculated by fusing joint angle sensor data and UWB positioning data, and the trajectory tracking error is obtained by comparing the difference between the preset trajectory planning value and the actual collected trajectory value by the edge computing node; the force control-related data is collected in real time by high-precision strain gauge force sensors integrated into the end effector to collect dynamic contact force data, and the contact force stability accuracy is quantified by statistically analyzing the standard deviation and fluctuation range of the contact force data within a preset time window; the coordination-related data is obtained through distributed synchronous... The data acquisition module synchronously captures the timing signals of multiple robot operations and calculates the timing synchronization error. The data interaction delay between robots is determined by the difference between the timestamps of data transmission and reception recorded by the edge computing node. The collaborative operation accuracy is quantitatively evaluated based on the overlap of the operation trajectories of each robot in the digital twin mirror and a preset accuracy threshold. The status feedback data is uploaded in real time by the local controller of each robot to show the operation progress. The time taken to complete a subtask is obtained by recording the timestamps of the subtask start and completion by the edge computing node and calculating the difference. All the collected raw data is transmitted to the edge computing node through shielded cables. After noise reduction and standardization, it is synchronized to the digital twin mirror and the collaborative decision-making module to provide accurate data support for judging the normality of the operation.

[0054] Operational normality judgment: Based on the above core analysis data, the operation is judged to be normal if the following conditions are met simultaneously: Motion accuracy meets the standard: trajectory tracking error ≤ ±0.01mm, body posture coordinate deviation ≤ ±0.02mm, joint speed fluctuation ≤ ±5%; Force control stability is: contact force stability accuracy ≤ ±0.2N; Collaborative performance is qualified: timing synchronization error ≤ 10ms, data interaction delay ≤ 5ms, collaborative operation accuracy ≤ ±0.05mm; End effector parameter matching is: the error between the grinding head grit size and the paint gun nozzle diameter and the self-optimized output value of the repair parameters ≤ the threshold, that is, the grinding head grit size error ≤ ±5 mesh and the nozzle diameter error ≤ ±0.02mm.

[0055] If the above conditions are not met simultaneously based on the core analysis data, the operation is deemed abnormal.

[0056] In a specific embodiment, the adaptive adjustment of the end effector parameters is implemented as follows: Adaptive adjustment scheme for end effector parameters in case of abnormality: If the operation is determined to be abnormal and the motion accuracy is not up to standard, the adaptive adjustment of the end effector parameters is as follows: Optimize the grinding head grit size by increasing it by 100-200 mesh, reduce grinding resistance and reduce the nozzle diameter of the touch-up paint gun by 0.02-0.03mm, and at the same time, adjust the center of gravity offset of the end effector to ≤±0.01mm.

[0057] If the operation is determined to be abnormal and the force control is unstable, the end effector parameters will be adaptively adjusted as follows: adjust the hardness of the grinding head and polishing wheel according to the direction of the contact force deviation. If the contact force is too large, replace the grinding head with one or two grades lower in hardness. If the contact force is too small, replace the grinding head with one or two grades higher in hardness. For touch-up painting, adjust the nozzle atomization pressure to ±0.05MPa and simultaneously correct the target value of grinding and polishing force to ±0.1N.

[0058] If the operation is determined to be abnormal and the collaborative performance is unqualified, the end effector parameters will be adaptively adjusted as follows: the operation rate parameters of the end effectors of each collaborative robot will be uniformly adjusted—the grinding and polishing trajectory speed will be reduced by 20%-30%, the touch-up painting speed will be reduced by 15%-25%, and the single-process operation interval will be extended by 0.1-0.2s; at the same time, the end tool adaptation parameters will be standardized, and the grinding head particle size deviation will be uniformly ≤±5 mesh and the nozzle diameter deviation will be ≤±0.02mm.

[0059] If the operation is determined to be abnormal and the end effector parameters are mismatched, the end effector parameters will be adaptively adjusted as follows: Based on the target value of the self-optimized output of the repair parameters, deviation compensation will be performed. For grinding head particle size deviation compensation: if the self-optimized output target particle size is 1500 mesh, and the actual particle size deviation is +10 mesh, adjust to 1495 mesh; if the deviation is -10 mesh, adjust to 1505 mesh; if the deviation is +20 mesh, adjust to 1490 mesh; if the deviation is -20 mesh, adjust to 1510 mesh. For touch-up paint gun nozzle diameter deviation compensation: if the self-optimized output target nozzle diameter is 0.2mm, and the actual nozzle diameter deviation is +0.03mm, adjust to 0.17mm; if the deviation is -0.03mm, adjust to 0.23mm; if the deviation is +0.05mm, adjust to 0.15mm; if the deviation is -0.05mm, adjust to 0.25mm.

[0060] Examples of embodiments of the present invention Figure 2 As shown, the multi-axis robot cluster collaborative control method for adaptive vehicle body repair includes the following steps: Step 1, multi-source fusion perception and dual matching: deploy a multi-modal sensor array to collect target vehicle body damage data, robot operation data and environmental data, thereby analyzing the vehicle body damage type and level corresponding to the target vehicle body damage, and then matching the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the current damage.

[0061] Step 2: Digital Twin Image Construction and Pre-optimization: Based on the optimal repair process and optimal multi-axis robot cluster type corresponding to the target vehicle body damage, a virtual image that is synchronized with the physical repair scene in real time is constructed, and pre-simulation optimization of the repair process, path conflict simulation, and parameter debugging are carried out.

[0062] Step 3: Distributed collaborative decision-making and fault-tolerant control: Receive the optimal repair process plan and the optimal multi-axis robot cluster type for the target vehicle body damage, thereby dynamically splitting and allocating the repair task, and realizing rapid fault response and task takeover through a three-level fault-tolerant architecture of virtual mirror.

[0063] Step 4, Force-position compliant execution and tool adaptation: Based on the issued optimal repair process plan and the optimal multi-axis robot cluster type, the coordinated control instructions generated drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site, and achieve adaptive adjustment of end effector parameters.

[0064] The examples described in this invention are not limited to the specific embodiments listed above. The examples are merely illustrative to facilitate understanding of the invention and do not constitute a limitation on the scope of protection of this invention. Any modifications, equivalent substitutions, etc., made within the spirit and principles of this invention should be included within the scope of protection.

[0065] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A multi-axis robot cluster collaborative control system for adaptive vehicle body repair, characterized in that, Includes the following modules: Multi-source fusion perception and dual matching module: used to deploy multimodal sensor arrays to collect target vehicle damage data, robot operation data and environmental data, thereby analyzing the type and level of vehicle damage corresponding to the target vehicle damage, and then matching the optimal repair process and optimal multi-axis robot cluster type corresponding to the current damage; Digital twin image construction and pre-optimization module: It is used to construct a virtual image that is synchronized with the physical repair scene in real time based on the optimal repair process scheme and the optimal multi-axis robot cluster type corresponding to the target vehicle body damage site, and to carry out pre-simulation optimization of the repair process, path conflict simulation and parameter debugging. Distributed collaborative decision-making and fault-tolerant control module: used to receive the optimal repair process plan and the optimal multi-axis robot cluster type at the target vehicle body damage site, thereby dynamically splitting and allocating the repair task, and then realizing rapid fault response and task takeover through the three-level fault-tolerant architecture of virtual image; Force-position compliant execution and tool adaptation module: It is used to drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site based on the issued optimal repair process plan and optimal multi-axis robot cluster type collaborative control instructions, and realize the adaptive adjustment of end effector parameters.

2. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 1, characterized in that, The deployment of the multimodal sensor array collects multi-dimensional data on target vehicle damage, robot operation, and the environment. The specific data collection process is as follows: A multimodal sensor array is deployed to cover the entire target vehicle body. 3D LiDAR and binocular vision sensors are symmetrically deployed around the repair station, 1.5-2 meters from the vehicle body. Terahertz radar and ultrasonic sensors are deployed next to the robot's end effector. UWB positioning sensors are embedded in each robot body, force sensors are integrated into the end effector, and temperature, humidity, and light sensors are deployed at the four corners of the repair station. All sensors are connected to edge computing nodes via shielded cables to collect real-time data on target vehicle damage, robot operation, and environmental data. The target vehicle body damage data includes: three-dimensional coordinates of the damaged area, damage depth, damage area, curvature of the damaged area, and dimensions of internal defects in the vehicle body; the robot operation data includes: robot body pose coordinates, joint rotation speed, joint temperature, end effector contact force, and robot load rate; the environmental data includes: ambient temperature, ambient humidity, and light intensity.

3. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 2, characterized in that, The analysis process for determining the type and level of vehicle body damage at the target vehicle body damage location is as follows: Damage type identification and analysis: The quantitative parameters of the target vehicle body damage are standardized. If the damage depth of the target vehicle body is <0.5mm and there are no internal defects, it is judged as a scratch; if the damage depth of the target vehicle body is ≥0.5mm, the area is ≥5cm² and there are no internal defects, it is judged as a dent; if the surface coating of the target vehicle body is missing and there is no obvious depth damage, it is judged as paint peeling; if the terahertz radar detects an internal defect size ≥1cm on the target vehicle body, it is judged as a hidden crack. Damage Level Quantitative Analysis: Damage depth is the core quantitative indicator, combined with damage area and internal defect size for level determination. If the target vehicle body meets the following conditions: damage depth ≥ 0.1mm and < 0.3mm, damage area < 2cm², and no internal defects, it is classified as Level 1; if the target vehicle body meets the following conditions: damage depth ≥ 0.3mm and < 0.5mm, damage area ≥ 2cm² and < 5cm², and no internal defects, it is classified as Level 2; if the target vehicle body meets the following conditions: damage depth ≥ 0.5mm and < 0.8mm, damage area ≥ 5cm² and < 10cm², and no internal defects, it is classified as Level 3; damage depth ≥ 0.8mm and < 1.0mm, damage area ≥ 10cm² and < 20cm², and internal defect size < 5cm, it is classified as Level 4; damage depth ≥ 1.0mm, damage area ≥ 20cm², and internal defect size ≥ 5cm, it is classified as Level 5.

4. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 3, characterized in that, The matching process involves determining the optimal repair process and the optimal multi-axis robot cluster type corresponding to the current damage. The specific matching process is as follows: A1. If the target vehicle body damage type is scratch and the level is 1, the optimal repair process is: light grinding + local polishing, and the optimal multi-axis robot cluster type is single-axis robot cluster. If the target vehicle body damage type is scratch and the level is 2, the optimal repair process is: moderate polishing + local paint touch-up + low temperature curing, and the optimal multi-axis robot cluster type is a three-axis robot cluster. If the target vehicle body damage type is dent and the level is 3, the optimal repair process is: dent pull-out correction + fine grinding + paint touch-up + curing, and the optimal multi-axis robot cluster type is a six-axis robot cluster. If the target vehicle body damage type is dent and the level is 4, the optimal repair process is: multi-point pull-out correction + sheet metal shaping + fine grinding + layered paint touch-up + constant temperature curing. The optimal multi-axis robot cluster type is multi-axis collaborative robot cluster. If the target body damage type is dent and the level is 5, the optimal repair process is: sheet metal cutting and replacement + welding + grinding + overall paint touch-up + high temperature curing, and the optimal multi-axis robot cluster type is multi-axis collaborative robot cluster. If the target vehicle body damage type is paint peeling and the level is 1-2, the optimal repair process is: surface cleaning + local paint touch-up + low temperature curing. The optimal multi-axis robot cluster types are single-axis and three-axis robot clusters, respectively. If the target vehicle body damage type is paint peeling and the level is 3-5, the optimal repair process is: surface sanding and paint removal + overall paint touch-up + curing. The optimal multi-axis robot cluster types are six-axis, multi-axis collaborative, and multi-axis collaborative robot cluster, respectively. If the target vehicle body damage type is a latent crack and the level is 3-5, the optimal repair process is: crack repair agent filling + curing + surface grinding and polishing. The optimal multi-axis robot cluster type is a six-axis or multi-axis collaborative robot cluster.

5. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 4, characterized in that, The specific construction process for building a virtual image that is synchronized in real time with the physical repair scenario is as follows: B1. Basic Model Import and Initialization: Import the matched optimal multi-axis robot cluster 3D model and the target vehicle body 3D basic model, and accurately map the damaged area of ​​the vehicle body model according to the 3D coordinates, damage depth, damage area, curvature of the damaged area and the size of the internal defects of the vehicle body, to complete the initial model construction. B2. Scene Element Integration: Integrate the environmental parameters of the repair station, the repair tool model, and the key process parameters in the optimal repair process plan into the virtual scene to achieve digital mapping of all elements of the physical repair scene. B3. Real-time synchronization mechanism establishment: A virtual-physical data transmission channel is established through edge computing nodes to synchronize the robot's real-time pose, end-effector contact force, and target vehicle body status data at a 20ms update cycle.

6. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 5, characterized in that, The analysis of the dynamic splitting and allocation of repair tasks and the self-optimization of repair parameters is as follows: C1. Task decomposition analysis: The overall repair task is divided into several sub-tasks, where a single damaged area corresponds to an independent sub-task, and adjacent damaged areas are merged into collaborative sub-tasks. C2. Dynamic Allocation Analysis: Based on the optimal multi-axis robot cluster's workspace, load capacity, and current operating status, precise allocation of sub-tasks is achieved. The specific allocation logic is as follows: First, workspace matching: Extract the 3D coordinates of the target vehicle body damage area corresponding to each sub-task, compare them with the workspace range of each robot, and filter out a set of candidate robots whose workspace can cover the damage area, excluding robots whose workspace cannot reach it. Second, load capacity adaptation: Obtain the rated load parameters of the candidate robots, and combine them with the process load requirements of each sub-task. Specifically, the load requirement for polishing tasks = process force + tool weight, and the load requirement for painting tasks = paint gun weight. + Remaining weight of paint, the load requirement for the drawing and correction task = drawing force + weight of drawing tool, the load requirement for the welding task = weight of welding torch + remaining weight of welding wire, the load requirement for the polishing task = polishing force + weight of polishing wheel, the load requirement for the crack repair agent filling task = weight of filling torch + remaining weight of repair agent, screen out the adaptable robots whose rated load is ≥ 1.2 times the process load of the sub-task; the third step, priority ranking of the current operating status: collect the current operating status parameters of the adaptable robots, including the current load rate, remaining power and fault history, and rank the adaptable robots according to the priority rule of lowest load rate > highest remaining power > no fault history; assign the sub-task to the robot with the best ranking.

7. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 6, characterized in that, The multi-axis robot cluster, after drive matching, performs compliant and coordinated operations on the target vehicle body damage site. The specific execution process is as follows: Collect core analysis data: Robot motion-related data includes the pose coordinates of each robot body, joint rotation speed, actual values ​​of motion trajectory, and trajectory tracking error; Force control-related data includes the real-time value of the end effector contact force and the contact force stability accuracy; Collaboration-related data includes the timing synchronization error of multi-robot operations, data interaction delay between robots, and collaborative operation accuracy; Status feedback data includes operation progress, execution status, and sub-task completion time. Operational normality judgment: Based on the above core analysis data, the operation is judged to be normal if the following conditions are met simultaneously: Motion accuracy meets the standard: trajectory tracking error ≤ ±0.01mm, body posture coordinate deviation ≤ ±0.02mm, joint speed fluctuation ≤ ±5%; Force control stability: contact force stability accuracy ≤ ±0.2N; Collaborative performance is qualified: timing synchronization error ≤ 10ms, data interaction delay ≤ 5ms, collaborative operation accuracy ≤ ±0.05mm; End effector parameter matching: the error between the grinding head grit size and the paint gun nozzle diameter and the self-optimized output value of the repair parameters ≤ the threshold, that is, the grinding head grit size error ≤ ±5 mesh and the nozzle diameter error ≤ ±0.02mm; If the above conditions are not met simultaneously based on the core analysis data, the operation is deemed abnormal.

8. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 7, characterized in that, The three-level fault-tolerant architecture enables rapid fault response and task takeover, and the specific implementation process is as follows: Obtain the optimal repair process plan for the target vehicle body, and collect the actual process parameters of each robot when executing the corresponding sub-task in real time. The actual process parameters correspond one-to-one with the process parameters required by the optimal repair process plan of the sub-task. The actual process parameters collected are compared with preset thresholds, which are determined as parameter standards for judging the normality of operation. If the actual process parameters of a robot exceed the above-mentioned preset threshold, a fault is triggered and the fault information is immediately reported to the area controller. After receiving the fault information, the area controller completes the fault verification within 50ms. If there is an idle and suitable robot in the area, the unfinished sub-tasks of the faulty robot are immediately assigned to that robot, and the task planning in the area is updated. If the regional controller is unable to take over the task, it will immediately report to the central controller. The central controller will then update the global repair plan and dispatch an adapted robot from other regions to take over the task.

9. The multi-axis robot cluster collaborative control system for adaptive vehicle body repair according to claim 8, characterized in that, The adaptive adjustment of the end effector parameters is achieved through the following process: Adaptive adjustment scheme for end effector parameters in case of abnormality: If the operation is determined to be abnormal and the motion accuracy is not up to standard, the end effector parameters will be adaptively adjusted as follows: optimize the grinding head grit size by increasing it by 100-200 mesh, reduce grinding resistance and reduce the nozzle diameter of the touch-up gun by 0.02-0.03mm, and at the same time, adjust the center of gravity offset of the end effector to ≤±0.01mm; If the operation is determined to be abnormal and the force control is unstable, the end effector parameters will be adaptively adjusted as follows: adjust the hardness of the grinding head and polishing wheel according to the direction of the contact force deviation. If the contact force is too large, replace the grinding head with one or two grades lower in hardness. If the contact force is too small, replace the grinding head with one or two grades higher in hardness. For the touch-up painting operation, adjust the nozzle atomization pressure to ±0.05MPa and simultaneously correct the target value of grinding and polishing force to ±0.1N. If the operation is determined to be abnormal and the collaborative performance is unqualified, the end effector parameters will be adaptively adjusted as follows: the operation rate parameters of the end effectors of each collaborative robot will be uniformly adjusted—the grinding and polishing trajectory speed will be reduced by 20%-30%, the touch-up painting speed will be reduced by 15%-25%, and the single-process operation interval will be extended by 0.1-0.2s; at the same time, the end tool adaptation parameters will be standardized, and the grinding head particle size deviation will be uniformly ≤±5 mesh, and the nozzle diameter deviation will be ≤±0.02mm. If the operation is determined to be abnormal and the cause is a mismatch in the end effector parameters, the end effector parameters will be adaptively adjusted as follows: Based on the target value of the self-optimized output of the repair parameters, deviation compensation will be applied. For grinding head particle size deviation compensation: if the self-optimized output target particle size is 1500 mesh, and the actual particle size deviation is +10 mesh, then adjust to 1495 mesh; if the deviation is -10 mesh, adjust to 1505 mesh; if the deviation is +20 mesh, adjust to 1490 mesh; if the deviation is -20 mesh, adjust to 1510 mesh. For touch-up paint gun nozzle diameter deviation compensation: if the self-optimized output target nozzle diameter is 0.2mm, and the actual nozzle diameter deviation is +0.03mm, then adjust to 0.17mm; if the deviation is -0.03mm, adjust to 0.23mm. If the deviation is +0.05mm, adjust to 0.15mm; if the deviation is -0.05mm, adjust to 0.25mm.

10. A method executed using the multi-axis robot swarm cooperative control system for adaptive vehicle body repair according to any one of claims 1-9, characterized in that, Includes the following steps: Step 1: Multi-source fusion perception and dual matching: Deploy a multi-modal sensor array to collect target vehicle damage data, robot operation data and environmental data, thereby analyzing the type and level of vehicle damage corresponding to the target vehicle damage, and then matching the optimal repair process and the optimal multi-axis robot cluster type corresponding to the current damage. Step 2: Digital Twin Image Construction and Pre-optimization: Based on the optimal repair process and optimal multi-axis robot cluster type corresponding to the target vehicle body damage, a virtual image that is synchronized with the physical repair scene in real time is constructed, and pre-simulation optimization of the repair process, path conflict simulation and parameter debugging are carried out. Step 3, Distributed Collaborative Decision Making and Fault Tolerance Control: Receive the optimal repair process plan and the optimal multi-axis robot cluster type for the target vehicle body damage, thereby dynamically splitting and allocating the repair task, and realizing rapid fault response and task takeover through a three-level fault-tolerant architecture of virtual mirrors; Step 4, Force-position compliant execution and tool adaptation: Based on the issued optimal repair process plan and the optimal multi-axis robot cluster type, the coordinated control instructions generated drive the matched multi-axis robot cluster to perform force-position compliant collaborative operations on the target vehicle body damage site, and achieve adaptive adjustment of end effector parameters.

Images