Ship plane segmented welding multi-robot path planning system based on visual inspection

Abstract

The invention discloses a ship plane segmented welding multi-robot path planning system based on visual inspection, and relates to the technical field of ship manufacturing, the ship plane segmented welding multi-robot path planning system comprises a multispectral visual inspection module which is mounted on a portal frame and comprises a laser scanning sensor and an infrared thermal imager, the system is used for collecting three-dimensional point cloud and temperature distribution data of ship plane segments in real time. According to the ship plane segmented welding multi-robot path planning system based on visual inspection, the problem of path planning misalignment caused by assembly errors, thermal deformation and complex working conditions in large ship plane segmented welding is fundamentally solved through a multispectral visual real-time perception and material self-adaption dynamic correction mechanism. A composite correction vector and space-time grid collaborative pre-judgment technology is adopted, the welding seam track precision is effectively improved, it is ensured that the weld leg size strictly reaches the standard in the high-difficulty processes such as wrap angle welding and vertical position welding of the maximum-size workpiece, the defects such as air holes and cracks are eliminated, and the rework rate and the quality risk are greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of shipbuilding, and in particular to a multi-robot path planning system for ship plane segmented welding based on visual detection. Background Art

[0002] Flat-section welding of ships, such as hull and deck sections, requires completing multiple weld types, including flat fillet welds, vertical fillet welds, and fillet welds, on large workpieces up to 15m x 15m. Traditional robotic welding lines utilize a pre-programmed offline planning approach: weld information is generated using SMARTWELD software, converted and simulated by the KCONG system, and then welds are executed on-site using fixed NCDATA data. However, this approach suffers from poor dynamic adaptability and low multi-machine coordination efficiency. Among these issues: Workpiece assembly errors, gaps between assembly components, and on-site thermal deformation can cause deviations between the actual working conditions and the pre-set model. Existing systems lack real-time perception and path correction capabilities, which can easily lead to gun interference. For example, fillet welds cannot be welded if the Δ is less than 35mm, or weld quality exceeds standards. The permissible weld leg deviation is only 0-0.5mm. Multi-machine coordination is also inefficient: The line requires multiple robots to work together in a shared space, such as a gantry with a Y-axis travel of 11 to 21m. However, the offline program cannot dynamically adjust paths based on real-time task allocation, resulting in robot waiting and trajectory conflicts. In this regard, the urgent problem to be solved is: how to realize real-time dynamic path planning of multiple robots in large-scale, high-precision welding scenarios to cope with on-site working condition deviations and optimize collaborative efficiency, which has become a decisive challenge to improving the quality and efficiency of ship segmented automated welding. Summary of the Invention

[0003] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-robot path planning system for ship plane segmented welding based on visual detection, including a gantry, a welding robot and a control cabinet:

[0004] The multispectral visual inspection module, mounted on the gantry, contains a laser scanning sensor and an infrared thermal imager, which is used to collect 3D point cloud and temperature distribution data of the ship's plane segments in real time;

[0005] The dynamic path planning module is connected to the control cabinet to receive visual inspection data and output the corrected path;

[0006] The multi-robot collaborative control module is electrically connected to the dynamic path planning module to dynamically allocate motion tasks to each robot;

[0007] The dual closed-loop verification module interacts with the dynamic path planning module and the welding robot signal respectively to perform path pre-verification and welding quality monitoring.

[0008] Preferably, the dynamic path planning module includes a deformation compensation unit and a space-time grid generation unit; the deformation compensation unit calculates the assembly error compensation amount based on the three-dimensional point cloud data, predicts the thermal deformation compensation amount in combination with the temperature distribution data, and generates a composite correction vector; the space-time grid generation unit discretizes the ship plane segments according to the preset grid size, marks the weld type feature points and allocates time windows; the preset grid size is set based on the ship segmentation accuracy requirements and the weld feature recognition requirements.

[0009] Preferably, the deformation compensation unit performs the following operations:

[0010] Based on the laser point cloud, the deviation between the actual contour of the workpiece and the preset model is identified and the geometric correction component is output;

[0011] Based on infrared thermal imaging data, the deformation trend is calculated through the heat conduction model and the thermal compensation component is output;

[0012] According to the dynamic weight coefficient matching of ship steel grades, the geometric correction component and the thermal compensation component are integrated to generate a composite correction vector.

[0013] Preferably, the spatiotemporal grid generation unit is provided with a conflict prediction subunit, which activates a dynamic avoidance mechanism when it is detected that the spatiotemporal grid overlaps within a critical distance Δ of the robot path:

[0014] If Δ is less than the minimum allowable distance for fillet welding, the path priority is forced to adjust;

[0015] If Δ is greater than the safe operating distance, maintain the original path planning.

[0016] Preferably, the multi-robot collaborative control module deploys a distributed reinforcement learning algorithm framework, wherein:

[0017] Each welding robot acts as an independent intelligent agent and generates an action strategy based on welding quality parameters and motion efficiency parameters;

[0018] The central decision maker coordinates the action strategies of each intelligent agent based on the spatiotemporal grid model to resolve the conflict of gantry Y-axis movement resources.

[0019] Preferably, the double closed-loop verification module includes:

[0020] Dynamic error envelope analysis unit compares the actual value of the weld foot size with the preset threshold in real time, triggering correction or interruption instructions;

[0021] The virtual twin collision rehearsal unit builds a virtual scene based on real-time 3D point cloud and injects random assembly errors to simulate collision risks.

[0022] Preferably, the virtual twin collision rehearsal unit performs Monte Carlo simulation, and the process includes:

[0023] Generate a random assembly error data set that follows a normal distribution;

[0024] Perform multiple rounds of collision detection on each planned path;

[0025] When the collision probability exceeds the set threshold, a re-planning instruction is sent to the dynamic path planning module.

[0026] Preferably, the weld type characteristic points include the flat fillet weld end point, the vertical fillet weld turning point and the wrap fillet weld positioning point; and the time window is calculated based on the robot's maximum motion acceleration and joint rotation angle.

[0027] Preferably, it also includes a process parameter adaptive unit, which is synchronized with the dynamic path planning module data to match the welding current, wire feeding speed and gas flow parameters according to the weld type.

[0028] Preferably, the matching logic of the dynamic weight coefficient is: for AH32 / DH36 high-strength steel, the weight of the thermal compensation component is increased; for ordinary marine grade A steel, the weight of the geometric correction component is increased.

[0029] The present invention provides a multi-robot path planning system for ship planar segmented welding based on visual inspection. It has the following beneficial effects:

[0030] This multi-robot path planning system for ship planar segment welding, based on visual inspection, fundamentally addresses the path planning inaccuracies caused by assembly errors, thermal deformation, and complex working conditions in large ship planar segment welding through real-time multispectral vision perception and a material-adaptive dynamic correction mechanism. It employs composite correction vectors and a spatiotemporal grid-based collaborative prediction technology to effectively improve weld trajectory accuracy, ensuring that weld leg dimensions meet strict standards in challenging processes such as fillet and vertical welding of the largest workpieces, eliminating defects such as porosity and cracks, and significantly reducing rework rates and quality risks.

[0031] This multi-robot path planning system for ship planar section welding, based on visual inspection, utilizes a distributed collaborative decision-making framework and a thousand-level collision probability verification mechanism to achieve efficient multi-robot collaborative operation within limited space resources. Dynamic priority arbitration and real-time synchronization of path parameters effectively avoid mechanical interference accidents, optimize gantry movement resource allocation, shorten welding cycles, and improve overall production line throughput. This dual safety closed-loop design provides reliable support for automated ship section welding and promotes the intelligent manufacturing of large structural components. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a schematic diagram of module interaction of a multi-robot path planning system for ship planar segment welding based on visual inspection according to the present invention;

[0033] Figure 2The figure is a flow chart of the multi-robot path planning method for ship plane segmented welding based on visual inspection of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] See also Figure 1 and Figure 2 The present invention provides a technical solution: a multi-robot path planning system for ship plane segmented welding based on visual detection, including a gantry, a welding robot and a control cabinet:

[0036] The multispectral visual inspection module, mounted on the gantry, contains a laser scanning sensor and an infrared thermal imager, which is used to collect 3D point cloud and temperature distribution data of the ship's plane segments in real time;

[0037] The dynamic path planning module is connected to the control cabinet to receive visual inspection data and output the corrected path;

[0038] The multi-robot collaborative control module is electrically connected to the dynamic path planning module to dynamically allocate motion tasks to each robot;

[0039] The dual closed-loop verification module interacts with the dynamic path planning module and the welding robot signal respectively to perform path pre-verification and welding quality monitoring.

[0040] It should be further explained that, in the specific implementation process, the implementation process of the multi-robot path planning system for ship plane segmented welding based on visual inspection is as follows:

[0041] A gantry-mounted laser scanning sensor acquires a three-dimensional point cloud of the workpiece surface to detect geometric deviations between the actual assembly contour and the preset model, including assembly errors of ≤±3mm and assembly gaps of ≤2mm. An infrared thermal imager monitors the temperature distribution in the welding area to identify local deformation trends caused by thermal deformation.

[0042] The actual weld position offset is calculated based on the laser point cloud data to generate a geometric correction component. The deformation direction from the high-temperature area to the low-temperature area is predicted based on the thermal imaging data, and the thermal compensation component is output. If the workpiece material is AH32 / DH36 high-strength steel, the weight of the thermal compensation component is increased first; if it is ordinary grade A steel, the geometric correction component is emphasized. Finally, a composite correction vector is output to drive the robot's posture adjustment.

[0043] The workpiece is divided into a 50mm×50mm grid, and characteristic locations such as the fillet weld endpoints and vertical weld turning points are marked. When the planned paths of two or more robots overlap within a critical distance Δ, if Δ is less than 35mm, the robot path with the lower priority is forced to detour or delay execution. If Δ is ≥65mm, the original path is maintained. The minimum allowable distance for fillet welds is 35mm.

[0044] Each robot acts as an independent intelligent entity, generating an action strategy based on welding quality requirements and motion efficiency. The central decision maker dynamically allocates gantry Y-axis resources to avoid travel conflicts. The welding quality requirements include an allowable deviation of the weld leg size of ±0.5mm.

[0045] The weld leg size is detected in real time. If the actual value exceeds the preset threshold of ±0.5mm three times in a row, welding is interrupted and an alarm is issued. A virtual work scene is constructed based on the real-time point cloud, and random assembly errors are injected to perform thousands of collision rehearsals. If the collision probability exceeds the safety threshold, the current path is immediately terminated and replanned. The welding current and wire feed speed parameters are automatically matched to the weld type, such as flat fillet welds and vertical fillet welds, and the corrected path is updated to the execution terminal simultaneously.

[0046] The dynamic path planning module includes a deformation compensation unit and a space-time grid generation unit; the deformation compensation unit calculates the assembly error compensation amount based on the three-dimensional point cloud data, predicts the thermal deformation compensation amount based on the temperature distribution data, and generates a composite correction vector; the space-time grid generation unit discretizes the ship plane into 50mm×50mm grids, marks the weld type feature points and allocates time windows.

[0047] It should be further explained that in the specific implementation process, the 3D point cloud data from the laser scanning sensor is received, the geometric deviation between the actual contour of the workpiece and the preset model is identified, including assembly errors and assembly gaps, and a geometric correction component is generated; the temperature distribution data from the infrared thermal imager is simultaneously analyzed, the deformation trend of the welding area is predicted through the heat conduction model, and the thermal compensation component is output;

[0048] If the workpiece material is AH32 / DH36 high-strength steel, the weight of the thermal compensation component is increased to a dominant position; if it is ordinary grade A steel, the geometric correction component is used first; finally, a composite correction vector is generated to drive the robot arm's posture adjustment and sent to the welding robot in real time.

[0049] The ship's planar surface is discretized into 50mm x 50mm grids, and weld type feature points, such as fillet weld positioning points and vertical fillet weld turning points, are annotated at the grid nodes. The occupied time window of each grid is calculated based on the robot's maximum motion acceleration and joint rotation angle.

[0050] When it is detected that multiple robot paths overlap in time and space within a critical distance Δ: if Δ is less than 35mm, the low-priority robot path is immediately marked as "forced avoidance", triggering a detour or delayed execution instruction; if Δ is greater than or equal to the safe operating distance: 65mm, the original path planning is maintained unchanged; the gantry Y-axis movement resources are dynamically allocated through the central decision maker to ensure continuous welding of high-priority welds; among which, the critical distance Δ is set to the minimum allowable distance of 35mm for fillet welds.

[0051] The deformation compensation unit performs the following operations:

[0052] Based on the laser point cloud, the deviation between the actual contour of the workpiece and the preset model is identified and the geometric correction component is output;

[0053] Based on infrared thermal imaging data, the deformation trend is calculated through the heat conduction model and the thermal compensation component is output;

[0054] According to the dynamic weight coefficient matching of ship steel grades, the geometric correction component and the thermal compensation component are integrated to generate a composite correction vector.

[0055] It should be further explained that, during the specific implementation process, the laser scanning sensor captures the three-dimensional point cloud of the workpiece surface, aligns the actual contour with the preset model, and identifies the assembly error: ±3mm deviation and group gap: ≤2mm gap; based on the detected deviation, the position offset of the key points of the weld is calculated, and the geometric correction component that drives the welding gun posture adjustment is generated. Among them, the key points of the weld include the end point of the fillet weld and the turning point of the vertical weld.

[0056] The infrared thermal imager monitors the temperature gradient in the welding area in real time and constructs a heat conduction model based on the thermal expansion coefficient of the steel. If a local high-temperature area greater than 300°C is detected, the deformation direction and amplitude toward the low-temperature area are predicted. For AH32 / DH36 high-strength steel, the predicted deformation amplitude is magnified by 1.2 times due to its strong thermal sensitivity. For ordinary Grade A steel, a standard deformation prediction model is used. The welding gun path compensation vector caused by thermal deformation is output as the thermal compensation component.

[0057] The fusion strategy is automatically selected according to the steel grade, including: in the high-strength steel scenario: the weight of the thermal compensation component is increased to more than 70%, and the geometric correction component accounts for ≤30%; in the ordinary steel scenario: the weight of the geometric correction component accounts for ≥80%, and the thermal compensation component is used as an auxiliary adjustment; the fused composite correction vector is sent to the welding robot controller in real time to drive the welding gun to synchronously adjust the spatial posture.

[0058] During the welding process, the weld foot size is detected three times in succession. If the actual value still deviates from the preset threshold by ±0.5mm, the compensation is judged to be invalid and the manual intervention protocol is triggered. If the deviation is stable within the allowable range, dynamic correction is continuously performed.

[0059] The spatiotemporal grid generation unit is equipped with a conflict prediction subunit. When it detects that the robot path has a spatiotemporal grid overlap within a critical distance Δ, it activates a dynamic avoidance mechanism:

[0060] If Δ is less than the minimum allowable distance for fillet welding, the path priority is forced to adjust;

[0061] If Δ is greater than the safe operating distance, maintain the original path planning.

[0062] It should be further explained that, during the specific implementation process, the real-time distance between the end points of any two robot welding guns is continuously calculated in the space-time grid model. When it is detected that the distance Δ enters the warning interval, the dynamic avoidance strategy is activated: if a high-precision weld such as fillet welding or vertical welding is currently being performed, the robot is marked as high priority; if an ordinary flat fillet weld is being performed, it is marked as low priority. The warning interval is: 35mm≤Δ<65mm.

[0063] In high-risk interference zones with a delta value less than 35mm, the low-priority robot is forced to perform an evasive maneuver. The detour involves shifting 50mm in the positive Y-axis direction and replanning the path. The delay involves pausing movement until the high-priority robot leaves the critical area. If both robots have high priority, the central decision maker will request arbitration to adjust the path order. For safe operating zones with a delta value greater than 65mm, the original path plan is maintained, and only the time window occupied by the grid is recorded.

[0064] The central decision maker generates a sequence of Y-axis movement instructions based on the occupancy status of the spatiotemporal grid: when multiple robots apply for Y-axis movement at the same time, priority is given to robots corresponding to high-precision welds; for robots that delay avoidance, their waiting time is compensated to the subsequent idle grid window.

[0065] After the avoidance action is executed, the Δ value is re-checked: if Δ is still <35mm, the avoidance is determined to have failed and the secondary avoidance protocol (such as Z-axis lifting) is triggered; if Δ is ≥65mm, the avoidance lock state is released.

[0066] The multi-robot collaborative control module deploys a distributed reinforcement learning algorithm framework, which:

[0067] Each welding robot acts as an independent intelligent agent and generates an action strategy based on welding quality parameters and motion efficiency parameters;

[0068] The central decision maker coordinates the action strategies of each intelligent agent based on the spatiotemporal grid model to resolve the conflict of gantry Y-axis movement resources.

[0069] It should be further explained that, during the specific implementation process, each welding robot acts as an independent intelligent entity, collecting its own welding quality data and motion status data in real time; among them, the welding quality data includes the weld leg size deviation, and the motion status data includes acceleration and joint angle; if high-precision welds such as fillet welding or vertical welding are currently being performed, the weld leg size accuracy is the primary optimization target; if ordinary flat fillet welding is being performed, motion efficiency is prioritized; then an action strategy is generated based on the optimization target: when the risk of path conflict is detected, the impact of the detour path on the welding quality is evaluated; when there is no conflict, the shortest time path is selected.

[0070] Receive space-time grid occupancy applications submitted by all agents and identify the gantry Y-axis resource contention area; when multiple agents apply for the same Y-axis travel section: the high-precision weld task agent automatically obtains priority use; the ordinary weld task agent accepts delay instructions or path segmentation solutions; generates resource allocation instructions and sends them to each robot controller.

[0071] For agents that are forced to wait due to avoidance, their subsequent grid time windows are automatically compensated; if the detour path affects the welding quality, such as interruption of vertical welding, the central decision maker reallocates the task boundaries of adjacent robots.

[0072] For every 10-meter weld completed, the number of task interruptions and the total welding time are counted: if the number of interruptions for high-precision welds exceeds the limit, its subsequent priority weight is increased; if the ordinary weld times out, the path segmentation algorithm is optimized.

[0073] The double closed-loop verification module includes:

[0074] Dynamic error envelope analysis unit compares the actual value of the weld foot size with the preset threshold in real time, triggering correction or interruption instructions;

[0075] The virtual twin collision rehearsal unit builds a virtual scene based on real-time 3D point cloud and injects random assembly errors to simulate collision risks.

[0076] It should be further explained that, in the specific implementation process, the actual weld leg size data is continuously collected during the welding process and compared with the preset threshold in real time; if a single over-limit is detected, the welding current and robot speed are automatically fine-tuned to compensate; if the limit is still exceeded after three consecutive compensations, it is judged as a system-level failure, the welding is immediately interrupted and the sound and light alarm is triggered, and the robot posture is frozen and waits for manual intervention.

[0077] Based on real-time 3D point clouds, a virtual working environment consistent with the physical scene is constructed, and a random assembly error data set that follows a normal distribution is generated to simulate the typical operating condition fluctuations of ship sections. Thousands of independent collision simulations are performed for each planned path: if the collision probability is below the safety threshold, the path execution is approved; if the collision probability exceeds the threshold, an emergency re-planning instruction is sent to the dynamic path planning module.

[0078] The quality monitoring unit shares a data interface with the collision rehearsal unit: when the collision rehearsal requires path replanning, the quality monitoring unit is synchronously notified to suspend detection; after the new path is executed, the quality monitoring unit starts the enhanced detection mode.

[0079] The virtual twin collision rehearsal unit performs Monte Carlo simulation, which includes:

[0080] Generate a random assembly error data set that follows a normal distribution;

[0081] Perform multiple rounds of collision detection on each planned path;

[0082] When the collision probability exceeds the set threshold, a re-planning instruction is sent to the dynamic path planning module.

[0083] It should be further explained that, during the specific implementation process, the three-dimensional point cloud data of the laser scanning sensor is received in real time to generate a virtual ship segment model that is completely consistent with the physical workpiece; the spatial envelope dimensions of the robot body, welding gun and gantry motion mechanism are accurately marked in the model to ensure that the simulation geometric accuracy reaches ±0.1mm.

[0084] Generate a random error data set that conforms to the typical working conditions of ship sections: for ordinary Grade A steel workpieces, a normally distributed error with a standard deviation of 1.5mm is used; for high-strength steel (AH32 / DH36) workpieces, due to their sensitivity to thermal deformation, the standard deviation is reduced to 1.0mm; before each simulation, a set of error values ​​are randomly selected and loaded into the key assembly points of the virtual model.

[0085] Execute 1,000 consecutive independent simulation tests on a single planned path, with each test randomly rotating the error loading direction, including the X, Y, and Z axes; dynamically simulate the model deformation drift caused by welding thermal deformation; and record interference events between the welding gun and the workpiece, between the robot, and the gantry structure.

[0086] The number of collisions in a thousand tests is counted. If the number of collisions is ≤1, that is, the probability is ≤0.1%, the path is deemed safe and execution is approved. If the number of collisions is ≥2, the path is immediately frozen and a red alert is sent to the dynamic path planning module. For high-risk welds, such as fillet welds with a Δ<35mm, the safety threshold is tightened to zero collision.

[0087] When a red alarm is triggered, the closest feasible solution in the historical safe path library is automatically called as a temporary path; the welding quality monitoring unit is simultaneously notified to start the highest level of detection; after the first section of the weld is completed after the new path is executed, the simulation verification cycle is reinitiated.

[0088] Weld type characteristic points include flat fillet weld endpoints, vertical fillet weld turning points, and wrap fillet weld positioning points; the time window is calculated based on the robot's maximum motion acceleration and joint rotation angle. It should be further explained that, during the specific implementation process, when locating the wrap fillet weld endpoint, if the distance between adjacent components is detected to be less than 65mm, it is automatically marked as a high-interference risk point, and a warning buffer zone with a radius of 35mm is generated; when identifying the vertical weld turning point, if the workpiece height exceeds 800mm or there is a sudden change in curvature, it is marked as a high-precision control point, requiring the robot to slow down to 60% of the standard speed to pass; flat fillet weld endpoints are classified according to the intersection angle of the components: 75°-90° is normally marked, and <75° triggers the process verification protocol.

[0089] Based on the maximum acceleration of the robot body and the rotation angle limit of the turntable, the shortest time window is calculated according to the maximum acceleration for straight weld segments; for arc welds, such as fillet weld R holes, the time window is calculated in combination with the ±185° limit of the rotation mechanism;

[0090] When the weld passes through the joint of the spliced ​​plate: if the height difference ΔH ≤ 3mm, a continuous time window is allocated for a single weld; if ΔH > 3mm, it is divided into two independent welds and allocated separate windows.

[0091] In the edge area of ​​the workpiece, that is, <200mm from the boundary: an additional 20% safety margin is added to the time window to prevent gantry inertial vibration; for multi-layer and multi-pass welding processes: the first welding time window is compressed to 80% of the standard value, and the window length of subsequent layers is increased according to the heat accumulation effect.

[0092] It also includes a process parameter adaptive unit, which synchronizes data with the dynamic path planning module to match welding current, wire feed speed, and gas flow parameters according to the weld type. It should be further explained that in the specific implementation process, the visual inspection module is used to identify the characteristics of the current welding position: if it is a fillet weld, that is, R-hole or SNIP type, the anti-spatter mode is activated: the gas flow is increased to 120% of the standard value, and the welding gun angle is locked at 45°±2°; if it is a vertical fillet weld, that is, the height is greater than 800mm, the multi-layer welding strategy is enabled: the first layer current is reduced by 10%, and the wire feed speed is simultaneously reduced; if it is a flat fold line fillet weld, segmented pulse welding is used: the current of the straight section is constant, and the current at the inflection point increases by 15% to compensate for heat loss.

[0093] Dynamically adjust parameters based on workpiece material: For high-strength steel (AH32 / DH36): the upper limit of welding current is compressed to 90% of the standard value to prevent degradation of the heat-affected zone; for ordinary grade A steel: the current is allowed to be over-matched by 10% to improve penetration efficiency; when the primer thickness is detected to be greater than 20μm, the gas flow is increased by an additional 15% to ensure arc stability.

[0094] When receiving the composite correction vector sent by the dynamic path planning module: if the path offset is greater than 2mm, the welding current is dynamically compensated by increasing or decreasing by 2% for every millimeter of offset; if the robot's avoidance action causes the welding gun angle to change by more than 5°, the gas flow is immediately corrected to the theoretical value; the parameter update instruction and the robot posture adjustment instruction are sent synchronously, with a delay of less than 50ms.

[0095] The infrared thermal imager monitors the molten pool shape in real time: if the melt width fluctuation is greater than the preset threshold, the wire feed speed is automatically increased until it stabilizes; if three consecutive adjustments are invalid, it will fall back to the safe parameter template.

[0096] The dynamic weight coefficient matching logic is as follows: for AH32 / DH36 high-strength steel, the thermal compensation component is weighted more heavily; for standard marine Grade A steel, the geometric correction component is weighted more heavily. It should be noted that in the specific implementation, a laser scanning sensor analyzes the workpiece surface reflectance spectrum characteristics: if the characteristic spectral line of AH32 / DH36 high-strength steel is identified, the thermal deformation dominant mode is activated; if the spectrum matches the standard Grade A steel spectrum, the geometric correction dominant mode is enabled. For areas where spectral identification is not possible, the high-strength steel weighting strategy is used by default.

[0097] The weight of the thermal compensation component has been elevated to a core position. If welding heat accumulation > 300°C s is detected, the thermal compensation weight is increased by an additional 10%. In areas where the vertical welding height is > 800mm, the geometric correction component is used to assist in fine-tuning. The geometric correction component is only used to compensate for sudden assembly deviations.

[0098] The geometric correction component dominates, that is, when the geometric correction component exceeds 80%: when the pair gap is greater than 1.5mm, the geometric correction weight is increased to 95%; when the primer thickness is detected to be greater than 20μm, the thermal compensation component intervenes to prevent arc drift; thermal deformation compensation is only used in high-temperature intensive areas, that is: continuous welding at >250℃ for more than 30s.

[0099] After executing the weight allocation, the welding quality is tested three times in a row: if the weld leg size deviation is still greater than 0.4mm, switch to the reverse weighting mode, that is, high-strength steel is geometrically dominated and ordinary steel is thermally dominated; if it is still ineffective after switching, trigger a system-level process inspection.

[0100] It should be further explained that, during the specific implementation process, the system uses a multispectral visual inspection module installed on the gantry to capture the working status of the ship's plane segments in real time. The laser scanning sensor acquires the three-dimensional contour data of the workpiece surface and accurately identifies the geometric deviation between the actual assembly position and the preset model, including assembly errors and assembly gaps. The infrared thermal imager simultaneously monitors the temperature distribution in the welding area and tracks the deformation trend caused by the heat conduction process. When a geometric deviation is detected, the system generates a geometric correction component; when the influence of thermal deformation is predicted, the system generates a thermal compensation component. The correction strategy is automatically selected according to the workpiece material type: for high-strength steel materials, the system prioritizes the thermal compensation component as the dominant correction amount; for ordinary steel materials, the system uses the geometric correction component as the core correction basis. The two components are fused through dynamic weighting to form the final correction instruction, which drives the welding robot to adjust the spatial position of the welding gun in real time.

[0101] The system divides the workpiece surface into uniform grid units, labels each grid with the weld feature type and allocates a time occupancy window. When the planned paths of multiple robots overlap in time and space within a critical distance range, the conflict prediction mechanism is immediately activated: if a high-precision weld task such as fillet welding or vertical welding is detected, its path execution priority is given; if an ordinary flat fillet welding task is detected, it is required to perform an avoidance action. The avoidance strategy includes path detour or delayed waiting, and the specific choice depends on the degree of impact of path deviation on welding quality. The central decision maker dynamically coordinates the gantry movement resources, and when a high-priority task occupies a critical area, it automatically compensates the waiting robot for the subsequent idle time window. After completing every ten-meter weld length, the system counts the number of task interruptions and the total time consumed, and dynamically optimizes the subsequent path allocation strategy.

[0102] During the welding process, the actual weld leg size is continuously monitored and compared with the preset threshold in real time. In the event of a single over-limit, the process parameters are automatically fine-tuned; if three consecutive fine-tunings are invalid, the system is deemed to have failed, welding is immediately interrupted and an alarm is issued. Before the path is executed, a virtual work scene is constructed based on the real-time three-dimensional point cloud, and a random assembly error data set that meets the typical working conditions of the ship segment is injected. Thousands of independent collision simulation tests are performed on each path, and each test randomly rotates the error loading direction and simulates thermal deformation drift. The frequency of collisions is counted: if the frequency exceeds the safety threshold, the path is immediately abandoned and the historical safety plan is called instead; if the frequency meets the standard, it is approved for execution. The quality monitoring unit is interconnected with the collision rehearsal unit data. Quality inspection is suspended during path replanning, and enhanced monitoring mode is started after the new path is executed.

[0103] When the system identifies the endpoints of a fillet weld, it automatically detects the spacing between adjacent components. When the spacing is less than the safe operating distance, a circular warning zone is generated and a high interference warning is activated. For vertical welding turning points with a height exceeding 800 mm, the robot's passing speed is forced to be reduced to ensure accuracy. When a height difference in the workpiece joint is detected, the continuity of the weld is automatically determined: height differences within three millimeters are treated as a single weld, and those exceeding three millimeters are divided into independent weld segments. The process parameters dynamically match the weld type: for fillet welding tasks, the shielding gas flow rate is increased and the welding gun inclination angle is locked; for vertical welding tasks, a layered current reduction strategy is adopted; and for broken line fillet welds, current pulse compensation is triggered at the turning point. When the system receives a path correction instruction, the welding parameters are adjusted synchronously: for every one millimeter increase in the path offset, the welding current increases or decreases by a corresponding proportion; when the welding gun angle changes by more than five degrees, the gas flow rate is immediately corrected.

[0104] The system identifies steel grades through spectral analysis. In high-strength steel scenarios, the thermal deformation compensation component dominates, and its weight is further increased when weld heat accumulation exceeds a critical value. Geometric correction is only activated when sudden large-scale assembly deviations are detected. In ordinary steel scenarios, the geometric correction component serves as the core compensation basis. Thermal deformation compensation is locally introduced when the assembly gap or primer thickness exceeds the standard. After weight allocation, welding quality is verified three times in a row: if the deviation continues to exceed the standard, the system switches to reverse weighting mode; if this is still ineffective, a full system process check is initiated. This fault-tolerant design ensures stability under extreme working conditions.

[0105] By evaluating both the steel grade and the working conditions, the system independently selects either geometric correction or thermal deformation compensation as the primary strategy, addressing the inability of a single model to cope with complex working conditions. For high-strength steel, the core objective is to suppress thermal cracking, while for standard steel, the primary task is to overcome assembly errors. A hierarchical avoidance mechanism is implemented based on weld process importance, with high-precision welds receiving path priority. Time window compensation and task boundary reorganization eliminate production line congestion caused by multiple machines waiting.

[0106] Thousands of random error simulations accurately replicate on-site ship operating conditions, ensuring zero collision standards in high-risk areas. A three-level quality control system, comprised of single fine-tuning, three verifications, and system interruptions, eliminates the risk of batch rework. Parameter adjustment commands are synchronized with welding gun position change commands to ensure process stability during dynamic corrections. An offset current compensation algorithm keeps weld leg size fluctuations within limits.

[0107] It should be further explained that the multi-robot path planning method for ship plane segment welding based on visual inspection includes the following steps:

[0108] Step S1: The gantry-mounted laser scanning sensor is used to capture the three-dimensional point cloud of the workpiece surface to accurately identify assembly errors and assembly gaps. An infrared thermal imager is simultaneously used to monitor the temperature gradient distribution in the welding area and record the thermal deformation trend characteristics.

[0109] Step S2: Generate geometric correction components based on laser point cloud data; predict deformation variables based on thermal imaging data and output thermal compensation components; if the workpiece material is high-strength steel, the thermal compensation component is preferentially used as the dominant correction strategy; if it is ordinary steel, the geometric correction component is used as the core correction basis; a composite correction vector is generated by fusion and sent to the robot controller.

[0110] Step S3: Discretize the workpiece surface into fixed-size grids, mark feature locations such as fillet weld endpoints and vertical weld turning points, calculate the time window occupied by each grid, and when multiple robot paths overlap within a critical distance, give priority to high-precision weld tasks and force ordinary weld tasks to execute detours or delay avoidance. The central decision maker dynamically compensates the subsequent time windows of the affected robots.

[0111] Step S4: Each robot generates an action strategy as an independent intelligent agent: high-precision welds prioritize quality, while ordinary welds prioritize efficiency; the central decision maker coordinates gantry movement resources to resolve travel conflicts; and task boundaries are redistributed after path changes.

[0112] Step S5: Continuously detect the deviation between the actual weld foot size and the preset threshold; automatically fine-tune the process parameters if the limit is exceeded once; if the fine-tuning is invalid for three consecutive times, interrupt the welding and trigger a system alarm.

[0113] Step S6: Construct a virtual work scene based on the real-time point cloud; inject normally distributed random assembly errors that conform to the ship's working conditions; perform thousands of independent collision simulations on a single path: if the number of collisions exceeds the safety threshold, call the historical safe path instead; and enforce the zero collision mandatory standard in high-risk weld areas.

[0114] Step S7: Identify the current weld type and match the parameters: for fillet welding, increase the gas flow and lock the welding gun inclination angle; for vertical welding, enable the layered current reduction strategy; when receiving the path correction instruction, adjust the parameters synchronously: the path offset compensates the current according to the linear relationship, and the gas flow is corrected immediately when the welding gun angle changes suddenly.

[0115] Step S8: Identify the steel grade through spectral characteristics. In the high-strength steel scenario, thermal deformation compensation is the main method, and geometric correction is only used for sudden large-scale assembly deviations. In the ordinary steel scenario, geometric correction is the core method, and thermal compensation is locally introduced when the primer exceeds the standard. After weight distribution, the welding quality is verified three times in a row.

[0116] Step S9: If the quality deviation continues to exceed the standard after weight allocation, switch to reverse weight mode; if it is still ineffective, freeze the system and start the full process verification; reset the compensation parameters after the verification is passed.

[0117] Step S10: Count the number of task interruptions and the total time consumed for each ten-meter weld completed; dynamically optimize conflict prediction rules and resource allocation strategies based on statistical data; and update the historical safety path library.

[0118] Through real-time multispectral vision perception and adaptive dynamic correction mechanisms for materials, the system fundamentally addresses the path planning inaccuracies caused by assembly errors, thermal deformation, and complex working conditions during the flat section welding of large ships. The system employs composite correction vectors and collaborative prediction technology with a spatiotemporal grid to effectively improve weld trajectory accuracy, ensuring that weld leg dimensions meet strict standards in challenging processes such as fillet and vertical welding of the largest workpieces, eliminating defects such as porosity and cracks, and significantly reducing rework rates and quality risks.

[0119] A distributed collaborative decision-making framework and a thousand-level collision probability verification mechanism enable efficient multi-robot collaborative operation within limited space resources. Dynamic priority arbitration and real-time synchronization of path parameters effectively avoid mechanical interference accidents, optimize gantry movement resource allocation, shorten welding cycles, and improve overall production line throughput. This dual safety closed-loop design provides reliable support for automated welding of ship sections, driving the intelligent manufacturing of large structural components.

[0120] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0121] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-robot path planning system for ship plane segment welding based on visual inspection, characterized by: It includes a gantry, a welding robot and a control cabinet, and is characterized by: The multispectral visual inspection module, mounted on the gantry, contains a laser scanning sensor and an infrared thermal imager, which is used to collect 3D point cloud and temperature distribution data of the ship's plane segments in real time; The dynamic path planning module is connected to the control cabinet to receive visual inspection data and output the corrected path; The multi-robot collaborative control module is electrically connected to the dynamic path planning module to dynamically allocate motion tasks to each robot; The dual closed-loop verification module interacts with the dynamic path planning module and the welding robot signal respectively to perform path pre-verification and welding quality monitoring.

2. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 1 is characterized in that: The dynamic path planning module includes a deformation compensation unit and a space-time grid generation unit; the deformation compensation unit calculates the assembly error compensation amount based on the three-dimensional point cloud data, predicts the thermal deformation compensation amount based on the temperature distribution data, and generates a composite correction vector; the space-time grid generation unit discretizes the ship plane segments according to the preset grid size, marks the weld type feature points and allocates time windows.

3. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 2 is characterized in that: The deformation compensation unit performs the following operations: Based on the laser point cloud, the deviation between the actual contour of the workpiece and the preset model is identified and the geometric correction component is output; Based on infrared thermal imaging data, the deformation trend is calculated through the heat conduction model and the thermal compensation component is output; According to the dynamic weight coefficient matching of ship steel grades, the geometric correction component and the thermal compensation component are integrated to generate a composite correction vector.

4. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 2 is characterized in that: The spatiotemporal grid generation unit is provided with a conflict prediction subunit, which activates a dynamic avoidance mechanism when it detects that the spatiotemporal grid overlaps within a critical distance Δ between robot paths: If Δ is less than the minimum allowable distance for fillet welding, the path priority is forced to adjust; If Δ is greater than the safe operating distance, maintain the original path planning.

5. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 1 is characterized in that: The multi-robot collaborative control module deploys a distributed reinforcement learning algorithm framework, where: Each welding robot acts as an independent intelligent agent and generates an action strategy based on welding quality parameters and motion efficiency parameters; The central decision maker coordinates the action strategies of each intelligent agent based on the spatiotemporal grid model to resolve the conflict of gantry Y-axis movement resources.

6. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 1 is characterized in that: The double closed-loop verification module includes: Dynamic error envelope analysis unit compares the actual value of the weld foot size with the preset threshold in real time, triggering correction or interruption instructions; The virtual twin collision rehearsal unit builds a virtual scene based on real-time 3D point cloud and injects random assembly errors to simulate collision risks.

7. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 6 is characterized in that: The virtual twin collision rehearsal unit performs Monte Carlo simulation, and the process includes: Generate a random assembly error data set that follows a normal distribution; Perform multiple rounds of collision detection on each planned path; When the collision probability exceeds the set threshold, a re-planning instruction is sent to the dynamic path planning module.

8. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 2 is characterized in that: The weld type characteristic points include the flat fillet weld end point, the vertical fillet weld turning point and the wrap fillet weld positioning point; the time window is calculated based on the robot's maximum motion acceleration and joint rotation angle.

9. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 1 is characterized in that: It also includes a process parameter adaptation unit, which synchronizes data with the dynamic path planning module to match welding current, wire feed speed and gas flow parameters according to the weld type.

10. The multi-robot path planning system for ship plane segment welding based on visual inspection according to claim 3, characterized in that: The matching logic of the dynamic weight coefficient is: for AH32 / DH36 high-strength steel, the weight of the thermal compensation component is increased; for ordinary marine grade A steel, the weight of the geometric correction component is increased.

Images