A visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method

Abstract

This invention relates to the field of automated welding and discloses a visual perception-based method for collaborative welding path planning that integrates wire feeding guidance and collision avoidance. The invention aims to solve the problem of simultaneously achieving high-precision wire feeding guidance and multi-obstacle avoidance within confined spaces. It establishes a local spatial map containing weld morphology and obstacles through visual perception, and uses laser ranging to assist in correcting deformation characteristics, achieving high-precision reconstruction of the working environment. The relationship between the wire feeding mechanism and the weld position is calibrated, and dynamic compensation is performed using resistance feedback and hose curvature monitoring to ensure stable wire feeding. A spatial decoupling control mechanism is adopted to generate a collaborative strategy for avoiding obstacles and pipelines based on dynamic weights while maintaining the coordinates of the wire feeding contact point. A smooth interpolation mechanism drives the motion, and closed-loop optimization is performed based on the correlation analysis of real-time trajectory and material thermal properties. This achieves precise path planning and collaborative collision avoidance within confined spaces, improving operational accuracy and safety.

Description

Technical Field

[0001] This invention relates to the field of automated welding technology, and in particular to a welding path planning method based on visual perception for wire feeding guidance and collision avoidance. Background Technology

[0002] In the current industrial manufacturing field, with the continuous improvement of welding automation, using visual sensors mounted on the end effector of robots for weld seam recognition and path guidance has become the mainstream technical means, mainly used to solve the problem of weld seam positioning and trajectory tracking in the production process of complex components.

[0003] In actual welding operations of large and complex structural components, systems often face the challenges of limited working space and multiple motion constraints. Existing path planning methods typically focus on achieving a single objective, such as maintaining stable alignment of the welding torch relative to the weld. However, during welding, due to the large footprint of the wire feeding mechanism and the actuator, and the presence of cables and gas lines trailing behind, problems easily arise when the actuator adjusts its posture to adapt to the weld morphology or avoids local obstacles. These problems include interference between the wire feeding mechanism and the workpiece, a surge in resistance due to excessive bending of the wire feeding hose, and physical entanglement of the cables. It is difficult to achieve safe and continuous obstacle avoidance in complex and confined spaces while ensuring high-precision wire feeding guidance. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, in order to solve the problem that existing path planning methods are difficult to balance high-precision wire feeding guidance, smooth physical wire feeding, and multiple obstacle avoidance in a limited working space, this invention provides a visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a vision-based method for wire feeding guidance and collision avoidance in collaborative welding path planning, comprising the following steps: S1. Use a vision sensor installed at the end of the execution to obtain spatial information of the work area, and extract weld features to create a local spatial map reflecting the weld groove morphology. S2. The spatial positional relationship between the wire feeding mechanism and the center of the weld is calibrated, and the wire feeding compensation parameters are calculated in combination with the motion state of the execution end to obtain the guiding command to guide the welding wire into the molten pool; S3. Integrate the local spatial map and the guidance command to construct a follow-up safety protection space around the execution end, and generate a collaborative planning strategy that takes into account both wire feeding guidance and obstacle avoidance requirements through the attitude coordination transformation of the execution mechanism. S4. The collaborative planning strategy is converted into motion commands for the robotic arm joints, and the smoothing mechanism is used to drive the execution end to execute a continuous welding path that avoids obstacles while maintaining the wire feeding guidance constraint. S5. Record the trajectory data during the welding process, optimize the parameter system by comparing the degree of consistency between the measured trajectory and the planned trajectory, and output the path optimization scheme for different component characteristics.

[0008] As a preferred embodiment of the visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method of the present invention, the method involves: using a visual sensor installed at the execution end to acquire spatial information of the work area, and extracting weld features to establish a local spatial map reflecting the weld groove morphology. The specific steps are as follows: The process of establishing a local spatial map includes: projecting structured light stripes onto the workpiece surface using the vision sensor; calculating the depth, width, and edge coordinates of the weld seam based on the deformation information of the stripes; adjusting the photosensitive exposure parameters and image signal amplitude of the vision sensor to counteract the influence of ambient light and welding arc light on the brightness of image features; and fusing and reconstructing the extracted weld seam features with the point cloud data of surrounding tooling fixtures. When establishing the local spatial map, a laser ranging module and an infrared imaging module are integrated into the vision sensor to assist in the identification of the structured light stripes. The actual distance values ​​determined by the laser ranging module are used to proportionally correct the deformation information, and the infrared imaging module is used to extract the thermal feature boundary of the weld seam in a dense smoke environment, improving the accuracy of spatial information acquisition in harsh industrial environments.

[0009] As a preferred embodiment of the visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method of the present invention, the following steps are taken: The spatial positional relationship between the wire feeding mechanism and the weld center is calibrated, and the wire feeding compensation parameters are calculated in combination with the motion state of the execution end to obtain the guiding command to guide the welding wire into the molten pool: The process of acquiring the guidance command includes: extracting the spatial trajectory of the weld centerline in the local spatial map, calculating the offset between the wire feed nozzle centerline and the weld centerline; simultaneously monitoring the curvature of the wire feed hose, and if the curvature exceeds a preset threshold, adjusting the spatial position of the execution end to reduce wire feeding resistance, thereby achieving dynamic compensation for the spatial position relationship. When adjusting the spatial position relationship, the feedback value from the physical pressure sensor installed on the wire feed mechanism is obtained. If the feedback value indicates an abnormal increase in the wire advance resistance inside the wire feed hose, the execution end is automatically triggered to perform a micro-translation along the extension direction of the wire feed hose, thereby alleviating wire feeding vibration during the execution of the guidance command through physical displacement.

[0010] As a preferred embodiment of the visual perception-based wire feeding guidance and collision avoidance collaborative welding path planning method of the present invention, the method involves: integrating the local spatial map and the guidance command to construct a follow-up safety protection space around the execution end; and generating a collaborative planning strategy that takes into account both wire feeding guidance and obstacle avoidance requirements through the attitude collaborative transformation of the execution mechanism. The specific steps are as follows: The process of generating a collaborative planning strategy for obstacle avoidance includes: constructing a dynamic virtual flexible protective layer based on the geometry and motion envelope of the end effector; predicting collision risk by detecting the distance between the virtual flexible protective layer and obstacles in the local spatial map; if the collision risk reaches a warning threshold, maintaining the spatial coordinates of the wire feeding contact point unchanged and changing the body posture of the end effector to bypass the obstacle. When generating the collaborative planning strategy, the virtual flexible protective layer is associated with the trajectory information of the cable and air pipe dragged behind the end effector. By marking the positions of the cable and air pipe on the map and reserving a safety gap to prevent the cable and air pipe from entangled with obstacles and tooling fixtures during collaborative posture changes, the strategy is implemented.

[0011] As a preferred embodiment of the visual perception-based wire feeding guidance and collision avoidance collaborative welding path planning method of the present invention, the collaborative planning strategy is converted into motion commands for the robotic arm joints, and a smoothing mechanism is used to drive the end effector to execute a continuous welding path that avoids obstacles while maintaining wire feeding guidance constraints. The specific steps are as follows: The execution process of motion commands includes: subdividing the displacement of the robotic arm joints using an interpolation mechanism, limiting the rotation rate of the execution end to eliminate sudden motion changes, ensuring that the wire feed nozzle stably points to the center of the molten pool during posture changes, and not triggering the overload protection of the robotic arm motor.

[0012] As a preferred embodiment of the visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method of the present invention, the method includes: recording trajectory data during the welding process, optimizing the parameter system by comparing the degree of agreement between the measured trajectory and the planned trajectory, and outputting path optimization schemes for different component characteristics. The specific steps are as follows: The specific process of parameter system optimization includes: extracting trajectory data feedback during the welding process and comparing it with the preset planned trajectory; analyzing the influence of workpiece thermal deformation and mechanical wear on path accuracy using the deviation values ​​generated by the comparison, and accordingly correcting the weight parameters in the collaborative planning strategy to establish the optimal control logic for a specific component. During parameter system optimization, the deviation values ​​are correlated with preset workpiece material thermal expansion and contraction characteristic data. By identifying the dynamic deformation law of the weld caused by welding heat, the initial planned trajectory for the next process and the next workpiece welding is automatically updated, thereby achieving deep collaborative optimization of the welding process and path planning scheme.

[0013] The beneficial effects of this invention are: This invention establishes a local spatial map including weld morphology and tooling obstacles through multimodal visual perception, and uses laser ranging to assist in correcting deformation features, achieving high-precision digital reconstruction of the working environment. By calibrating the spatial relationship between the wire feeding mechanism and the weld center, and combining wire feeding resistance feedback and hose curvature monitoring for dynamic displacement compensation, the stability of the wire feeding physical process under complex postures is ensured. A spatial decoupling control mechanism is employed for coordinated posture transformation, generating path strategies to avoid obstacles and pipeline entanglement based on dynamic weights while maintaining the spatial coordinates of the wire feeding contact point. A smooth interpolation mechanism drives the end effector motion, and parameter closed-loop optimization is performed based on the correlation analysis of real-time trajectory data and material thermal properties. This invention achieves precise planning and coordinated collision avoidance of welding paths in complex and confined spaces, improving the guiding accuracy and operational safety of continuous welding operations, and reducing path deviations caused by thermal deformation and mechanical errors. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0015] Figure 1 This is a flowchart of the overall system process. Figure 2 Flowchart for reconstructing the S1 high-precision local spatial map; Figure 3 Flowchart of wire feeding guidance and physical compensation for S2; Figure 4 This is the core logic diagram for S3 spatial decoupling and collaborative obstacle avoidance. Figure 5 This is a parameter closed-loop optimization diagram for S5 based on trajectory feedback. Detailed Implementation

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0017] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0018] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0019] Example 1 Reference Figures 1-5 This is the first embodiment of the present invention, which provides a visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method, including the following steps: S1. Using a vision sensor installed at the end of the execution device to acquire spatial information of the work area, and extracting weld features to create a local spatial map reflecting the weld groove morphology, including the following steps: The process of establishing a local spatial map includes: projecting structured light stripes onto the workpiece surface using a vision sensor; calculating the depth, width, and edge coordinates of the weld seam based on the deformation information of the stripes; adjusting the photosensitive exposure parameters and image signal amplitude of the vision sensor to counteract the influence of ambient light and welding arc light on the brightness of image features; and fusing and reconstructing the extracted weld seam features with the point cloud data of surrounding tooling fixtures. When establishing the local spatial map, a laser ranging module and an infrared imaging module are integrated into the vision sensor to assist in the recognition of structured light stripes. The actual spacing values ​​determined by the laser ranging module are used to proportionally correct the deformation information, and the infrared imaging module is used to extract the thermal feature boundaries of the weld seam in a dense smoke environment, improving the accuracy of spatial information acquisition in harsh industrial environments.

[0020] It should be noted that during the welding operation initiation phase, the system first activates the vision sensor module installed at the execution end to collect spatial information and model the environment of the work area. The specific operation steps are as follows: Initialization of the vision system and collaborative scanning with sensors: The vision sensor (integrating a binocular structured light module, a laser ranging module, and an infrared imaging module) at the end of the system command execution projects a pre-set frequency array of structured light stripes onto the target welding area. The laser ranging module simultaneously emits a detection signal onto the workpiece surface, acquiring the actual distance value between the sensor and the bevel surface in real time.

[0021] Structured light image acquisition and environmental interference suppression: As the sensor moves with the end effector, the industrial camera captures stripe images modulated by the bevel geometry. To cope with strong arc light interference during welding operations, the system adjusts the photosensitive exposure parameters and image signal amplitude of the vision sensor in real time, and uses physical filters to control the feature brightness within the algorithm processing range, effectively offsetting the influence of external ambient light and arc light on image features and ensuring clear stripe boundaries.

[0022] Pixel-level extraction and calculation of bevel features: The system analyzes the acquired stripe images frame by frame to locate the pixel displacement of the stripes at the bevel. To eliminate scaling distortion caused by fluctuations in shooting distance, the system introduces the following physical mapping formula for scaling correction:

[0023] : Represents the actual geometric dimensions (such as depth or width) of the weld bevel.

[0024] : This is the actual distance between the sensor and the workpiece surface measured by the laser ranging module, and the value range is usually from 50mm to 300mm.

[0025] : This represents the pixel displacement caused by the structured light stripes on the image sensor.

[0026] : is the lens focal length constant of the vision sensor.

[0027] : This is the optical system distortion correction factor, with a value ranging from 0.95 to 1.05.

[0028] Multi-source fusion and noise processing of point cloud data: The system fuses and reconstructs the calculated 3D point cloud data with the point clouds of surrounding tooling fixtures. During this process, the system automatically removes noise points caused by metal reflection or welding slag sputtering, and uses the thermal feature boundaries extracted by the infrared imaging module in a dense smoke environment to supplement the data, ensuring that the continuity of the bevel features can still be maintained even when visible light is blocked.

[0029] Digital reconstruction of the local spatial map: The system fills the processed 3D coordinate data into a digital spatial grid to generate a local spatial map reflecting the actual location of the weld, bevel angle, bevel width, and depth distribution. The local spatial map is updated in real time as the execution end moves, retaining only environmental information within the current working area and the preset range ahead, thereby reducing the system's computational load.

[0030] S2. The spatial relationship between the wire feeding mechanism and the weld center is calibrated, and the wire feeding compensation parameters are calculated based on the motion state of the execution end to obtain the guiding command for guiding the welding wire into the molten pool. This includes the following steps: The process of acquiring guidance commands includes: extracting the spatial trajectory of the weld centerline from the local spatial map and calculating the offset between the wire feed nozzle centerline and the weld centerline; simultaneously monitoring the curvature of the wire feed hose. If the curvature exceeds a preset threshold, the spatial position of the execution end is adjusted to reduce wire feeding resistance, achieving dynamic compensation for the spatial position relationship. During spatial position adjustment, feedback values ​​from the physical pressure sensor installed on the wire feed mechanism are acquired. If the feedback value indicates an abnormal increase in the wire advance resistance inside the wire feed hose, the execution end is automatically triggered to perform micro-translation along the extension direction of the wire feed hose, mitigating wire feeding vibration during the execution of guidance commands through physical displacement.

[0031] It should be noted that the specific steps for establishing a precise mapping between the end of the wire feeding mechanism and the physical center of the weld groove are as follows: Spatial coordinate extraction of the weld center trajectory: The system retrieves the weld bevel morphology features of the current welding area in real time from the digital local spatial map generated in step S1. Using feature recognition logic, it locks the root of the bevel or a preset alignment centerline and extracts the real-time spatial coordinates of the centerline in the robot's base coordinate system. .

[0032] Real-time sensing of the end-effector's pose: Through a high-precision position sensor installed at the end of the actuator, the system acquires the real-time three-dimensional spatial coordinates of the wire feed nozzle outlet center in the same coordinate system. The system reads the attitude parameters of the robot's end-effector coordinate system and determines the unit vector of the welding wire extension direction.

[0033] Spatial position relationship calculation based on offset vector: The system introduces a wire feeding offset vector calculation model, and calibrates the spatial position relationship between the wire feeding mechanism and the weld center through the following physical mapping formula:

[0034] : Represents the compensation offset vector for wire feeding guidance, used to quantify the offset distance and direction of the wire feeding contact relative to the weld center.

[0035] : The coordinates of the weld center trajectory point identified in the local spatial map.

[0036] : Real-time spatial coordinates of the end of the feed nozzle.

[0037] : This is the preset extension length of the welding wire from the wire feed nozzle outlet to the workpiece surface. In this embodiment, the value ranges from 0mm to 25mm.

[0038] : This is the real-time bending angle of the wire feeding hose in the current robotic arm posture. This parameter is obtained by monitoring the pitch angle of the end effector and ranges from 0° to 45°. It is used to correct wire pointing deviation caused by hose bending.

[0039] Wire feeding hose physical state monitoring and dynamic compensation: During the calculation process, the system monitors the bending curvature of the hose through flexible strain gauges integrated on the wire feeding hose. When it detects that the bending curvature exceeds the physical critical threshold due to the large-scale detour of the robotic arm, the system automatically fine-tunes the spatial position of the execution end to straighten the hose path and reduce the resistance of the welding wire, thereby achieving dynamic correction of the spatial position relationship.

[0040] Offset fine-tuning based on physical resistance feedback: Real-time acquisition of feedback values ​​from a physical pressure sensor installed inside the wire feeding mechanism. When the value shows an abnormal increase in the wire advance resistance inside the wire feeding hose (e.g., exceeding 20% ​​of the preset friction threshold), it indicates that the current spatial position is causing physical interference between the wire and the bevel sidewall. The system automatically triggers the execution end to perform a micro-translation command along the natural extension direction of the wire feeding hose. This physical displacement alleviates wire vibration during the guiding process, ultimately outputting a guiding command that precisely guides the wire into the center of the molten pool.

[0041] S3. Integrating local spatial maps and guidance instructions, a dynamic safety protection space is constructed around the execution end. A collaborative planning strategy that balances wire feeding guidance and obstacle avoidance requirements is generated through the coordinated attitude transformation of the actuator, including the following steps: The process of generating a collaborative planning strategy for obstacle avoidance includes: constructing a dynamic virtual flexible protective layer based on the geometry and motion envelope of the end effector; predicting collision risk by detecting the distance between the virtual flexible protective layer and obstacles in the local spatial map; if the collision risk reaches a warning threshold, maintaining the spatial coordinates of the wire feeding contact point unchanged and changing the body posture of the end effector to bypass the obstacle. When generating the collaborative planning strategy, the virtual flexible protective layer is associated with the trajectory information of the cable and air pipe dragged behind the end effector. The positions of the cable and air pipe are marked on the map, and a safety gap is reserved during the collaborative posture transformation to prevent the cable and air pipe from entangled with obstacles and tooling fixtures.

[0042] It should be noted that dynamic obstacle avoidance is achieved through multi-source information fusion while ensuring welding accuracy. The specific operation steps are as follows: Dynamic construction of a follow-up virtual flexible protective space: The system integrates the digital local space map generated in step S1 with the guidance instructions obtained in step S2 to obtain the spatial relationship between the distribution of obstacles on the workpiece surface and the ideal wire feeding trajectory. Based on the physical geometry of the execution end, the motion envelope range of the wire feeding mechanism, and the real-time position of the rear drag cable and air pipe, a follow-up virtual flexible protective layer surrounding the entire execution end is constructed in the digital space.

[0043] Real-time collision risk prediction based on local maps: The system detects the nearest spatial distance between the virtual flexible protective layer and obstacles marked on the local spatial map (such as tooling fixtures, non-welded parts of parts, cable bundles, etc.) in real time. The system compares the distance to a preset safety warning threshold. If the distance continues to decrease and reaches the warning threshold, the system determines that there is a risk of collision or cable entanglement.

[0044] Attitude Coordination Based on Spatial Decoupling Algorithm: When a risk is detected, the system initiates a spatial decoupling control program, decomposing the motion of the end effector into wire feeding alignment components and body obstacle avoidance components. While ensuring that the spatial coordinates and alignment accuracy of the wire feeding contact point within the weld remain constant, the system calculates the correction increment for the rotation of the end effector body around the wire feeding contact point. :

[0045] : The attitude maintenance vector required to maintain precise alignment of the wire feeding guide.

[0046] : The detour attitude correction vector required to avoid collisions with obstacles or prevent cable entanglement.

[0047] , : This refers to the dynamic weight parameters in the collaborative planning strategy, and their values ​​are determined based on distance. The changes are adjusted in real time to ensure that obstacle avoidance actions are given higher priority when approaching obstacles. , ∈[0,1] and the sum of the two is 1.

[0048] Path planning strategy generation for associated pipeline constraints: During attitude cooperative transformation, the system marks the dynamic space occupied by cables and air pipes in real time and automatically reserves safety clearances when calculating the detour trajectory. By adjusting the cooperative motion of each axis joint of the robotic arm, a cooperative planning strategy is generated that can guide the welding wire precisely to the center of the molten pool and enable the end effector and its associated pipelines to avoid all physical obstacles in the local spatial map.

[0049] Strategy safety verification and pre-output: The system performs logical verification on the generated cooperative planning strategy to confirm that each axis joint will not enter a kinematically singular pose during the orbital process, and that the amplitude of the attitude transformation will not cause the bending curvature of the wire feeding hose to exceed the critical threshold in step S2. After the verification is passed, the final composite motion path is output to step S4 for interpolation-driven execution.

[0050] S4. Transform the collaborative planning strategy into motion commands for the robotic arm joints, and use a smoothing mechanism to drive the end effector to execute a continuous welding path that avoids obstacles while maintaining wire feeding guidance constraints. This includes the following steps: The execution process of motion commands includes: subdividing the displacement of the robotic arm joints using an interpolation mechanism, limiting the rotation rate of the execution end to eliminate sudden motion changes, ensuring that the wire feed nozzle stably points to the center of the molten pool during posture changes, and not triggering the overload protection of the robotic arm motor.

[0051] It should be noted that this step aims to translate high-level planning strategies into precise actions at the underlying hardware level, ensuring the continuity and stability of the welding process. The specific operational steps are as follows: Discretization of the collaborative planning strategy: The system receives the collaborative planning strategy generated by step S3, which takes into account both wire feeding guidance and obstacle avoidance requirements. According to the preset control cycle, the continuous spatial trajectory in the collaborative planning strategy is decomposed into a series of target pose points containing coordinate positions and attitude angles.

[0052] Command conversion in the joint space of the robotic arm: Using existing industrial control software, the target pose sequence is converted into pulse control commands for the motors of each joint of the robotic arm through inverse kinematics. During the conversion process, the system monitors the physical limits of each joint in real time to ensure that the generated motion commands do not cause the robotic arm to enter kinematic singularities or exceed its physical travel range.

[0053] Smoothing based on interpolation mechanism: The displacement of the robotic arm joints is subdivided into milliseconds using an interpolation mechanism. By limiting the rotation rate and acceleration change rate of the end effector during obstacle avoidance, abrupt motion changes that may occur due to frequent attitude shifts are eliminated. This smoothing process can suppress instantaneous impacts on the mechanical structure, thereby preventing the end effector from generating mechanical vibrations that could interfere with the formation of the molten pool.

[0054] Dynamic maintenance of wire feeding guidance constraints: When the drive actuator executes a detour trajectory to avoid obstacles, the system adjusts the pointing angle of the welding torch in real time through the underlying feedback loop. This ensures that while the robotic arm moves laterally or flips to avoid obstacles, the wire feed nozzle always points stably to the spatial position relationship point calibrated in step S2, maintaining the constant wire feeding guidance constraints.

[0055] Drive execution and real-time safety monitoring: The processed motion command stream is pushed to each joint actuator, driving the end effector to move along the continuous welding path. During execution, the system monitors the current load and feedback speed of each joint motor in real time to ensure smooth movement and prevent triggering of the motor's overload protection mechanism. If the deviation between the actual running trajectory and the command trajectory exceeds a safety threshold, the system will automatically adjust the execution increment to ensure the continuity of the path.

[0056] S5. Record the trajectory data during the welding process, optimize the parameter system by comparing the degree of agreement between the measured trajectory and the planned trajectory, and output the path optimization scheme for different component characteristics, including the following steps: The specific process of parameter system optimization includes: extracting trajectory data feedback during the welding process and comparing it with the preset planned trajectory; analyzing the impact of workpiece thermal deformation and mechanical wear on path accuracy using the deviation values ​​generated by the comparison, and accordingly correcting the weight parameters in the collaborative planning strategy to establish the optimal control logic for specific components. During parameter system optimization, the deviation values ​​are correlated with preset workpiece material thermal expansion and contraction characteristics data. By identifying the dynamic deformation patterns of the weld caused by welding heat, the initial planned trajectory for the next process and the next workpiece welding is automatically updated, thereby achieving deep collaborative optimization of the welding process and path planning scheme.

[0057] It should be noted that the adaptive correction of the path scheme is achieved through closed-loop analysis of welding process data. The specific operation steps are as follows: Real-time trajectory data recording and acquisition: The system utilizes high-precision encoders and vision sensors at each joint of the robotic arm to synchronously record real-time trajectory data during the welding process. This data includes the spatial three-dimensional coordinates of the welding torch tip, attitude deflection angle, welding travel speed, and dynamic response parameters of the wire feeding mechanism.

[0058] Multi-dimensional feature comparison of trajectory matching degree: The recorded real-time trajectory data is fed back to the processing unit and compared with the preset digital planning trajectory in multiple dimensions. The system identifies the path deviation change pattern caused by workpiece thermal deformation, cumulative error of mechanical transmission, or external uncontrollable environmental interference by calculating the geometric deviation between the actual spatial position and the ideal planned position.

[0059] Iterative correction and optimization of the parameter system: By utilizing the deviation values ​​generated from the comparison, the system dynamically corrects the weight parameters in the collaborative planning strategy. Specifically, the system automatically adjusts the incremental amplitude in the attitude collaborative transformation command according to the magnitude of the deviation. By increasing or decreasing the correction ratio, subsequent attitude adjustment actions can compensate for the path offset that has occurred, ensuring that the parameter system is always in the optimal control state for the current operating conditions.

[0060] In-depth collaborative analysis of the thermal properties of related materials: The system further correlates the trajectory deviation value with preset workpiece material thermal expansion and contraction characteristic data. By identifying the dynamic evolution law of a specific weld under heat influence (such as the bulging or contraction displacement of the weld after heating), the system can predict the deformation trend that may occur in subsequent weld passes.

[0061] Output and storage of path optimization schemes: Based on the above feedback and analysis conclusions, the system automatically updates the initial planned trajectory instructions for welding the next weld or the next workpiece, and finally outputs path optimization schemes tailored to different component characteristics. These schemes are stored in the process knowledge base, providing adaptive correction control logic for the automated welding of subsequent similar components.

[0062] Example 2 To further clarify the technical solution of the present invention, the following example, a specific scenario of continuous automatic welding of irregular box-shaped components with complex reinforcing ribs in a heavy machinery manufacturing plant, is used as the second embodiment of the present invention to provide a more detailed description of the technical solution of the present invention.

[0063] In this embodiment, the workpiece to be welded is an L-shaped box with multiple vertical reinforcing ribs and pre-set tooling fixtures. Due to the compact layout of the reinforcing ribs and the fact that the weld is located within a deep bevel, the robot's end effector (welding torch and wire feeding mechanism) is prone to interference with the reinforcing ribs during movement.

[0064] Step S1: Environmental Perception and Local Map Construction Once the welding operation begins, the vision sensor enters the internal space of the housing along with the end effector. At this time, the structured light module integrated into the sensor projects stripes onto the bevel.

[0065] Anti-interference processing: Due to the strong secondary reflected light inside the box and the fact that the electric arc has been ignited, the system reduces the amplitude of the image signal in real time through automatic gain control to prevent overexposure of the striped image.

[0066] Data correction: The laser ranging module reports that the current actual distance value is 120.5mm. Based on this, the system scales the bevel pixel features extracted from the image and calculates the actual bevel depth to be 12mm.

[0067] Reconstruction result: The system merges the identified weld trajectory with the point cloud of a reinforcing rib plate 50mm ahead, and establishes a local spatial map in the digital grid that includes the deep bevel path and the nearby rib plate obstacle.

[0068] Step S2: Obtaining and dynamically calibrating wire feeding guide parameters The system calibrates the relative position of the wire feed nozzle and the center of the weld.

[0069] Offset calculation: The system calculates that the current deviation of the wire feed nozzle centerline from the weld center vector is 1.2mm to the left.

[0070] Physical feedback compensation: Because the robotic arm is entering the housing, the wire feeding hose is bent at a large angle. The pressure sensor mounted on the wire feeder detects that the wire advance resistance is 15% higher than the calibrated state.

[0071] Compensation execution: The system automatically calculates the wire feeding compensation parameters and instructs the execution end to make a slight adjustment of 3.5mm to the right and rear. This physical displacement relieves the tension of the hose and ensures that the guiding command can guide the welding wire to accurately point to the center of the molten pool.

[0072] Step S3: Construction of dynamic protection space and generation of collaborative strategies The system integrates local map data and finds that the distance between the outer edge of the welding torch and the reinforcing rib plate is only 15mm, triggering an early warning.

[0073] Follow-up protection: The system constructs a virtual flexible protective layer that extends to the rear cable, with the welding torch as the center.

[0074] Attitude transformation: To avoid the stiffeners, the system performs spatial decoupling calculations. This is while maintaining the wire feed contact. With the coordinates unchanged, by increasing the obstacle avoidance weight coefficient The instruction execution end body tilts outward at a 25-degree angle around the contact point.

[0075] Cable management: The system marks the position of the dragged protective air tube in real time to ensure that the air tube will not be caught on the edge of the clamp when passing around the stiffening plate.

[0076] Step S4: Path Execution and Smooth Control Drive The system translates the generated tilting and circling strategy into motion commands for six joints.

[0077] Smoothing: At the instant of bypassing the edge of the stiffener, the interpolation mechanism refines the motion step to 0.5ms. The end-effector rotation rate is limited to prevent the feed nozzle from wobbling due to drastic posture changes.

[0078] Synergistic effect: The end of the process, with a twisting posture, avoids the side reinforcing ribs while keeping the wire feed nozzle accurately pointing to the root of the bevel, thus completing a continuous obstacle avoidance welding path.

[0079] Step S5: Parameter System Optimization and Feedback After welding is completed, the system extracts the recorded real-time trajectory data.

[0080] Feature comparison: The comparison revealed that when passing through the heat-affected zone in the middle section of the box, due to the thermal expansion of the metal, the actual weld center shifted to the right by 0.8mm compared to the planned path.

[0081] Strategy optimization: The system analyzes the thermal expansion and contraction characteristics of the material and automatically corrects the planning algorithm for the next workpiece. In the next round of welding, the system increases the compensation weight of the wire feeding guide by 5% in advance and presets the corresponding thermal deformation offset displacement, thereby achieving deep collaborative optimization for this box-shaped component.

[0082] In summary, this invention establishes a local spatial map including weld morphology and tooling obstacles through multimodal visual perception, and uses laser ranging to assist in correcting deformation features, achieving high-precision digital reconstruction of the working environment. By calibrating the spatial relationship between the wire feeding mechanism and the weld center, and combining wire feeding resistance feedback and hose curvature monitoring for dynamic displacement compensation, the stability of the wire feeding physical process under complex postures is ensured. A spatial decoupling control mechanism is adopted for posture cooperative transformation, generating path strategies to avoid obstacles and pipeline entanglement based on dynamic weights while maintaining the spatial coordinates of the wire feeding contact point unchanged. A smooth interpolation mechanism is used to drive the end effector motion, and parameter closed-loop optimization is performed based on the correlation analysis of real-time trajectory data and material thermal properties. This invention achieves precise planning and cooperative collision avoidance of welding paths in complex and confined spaces, improves the guiding accuracy and operational safety of continuous welding operations, and reduces path deviations caused by thermal deformation and mechanical errors.

[0083] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for collaborative welding path planning based on visual perception of wire feeding guidance and collision avoidance, characterized in that, Includes the following steps: S1. Use a vision sensor installed at the end of the execution to obtain spatial information of the work area, and extract weld features to create a local spatial map reflecting the weld groove morphology. S2. The spatial positional relationship between the wire feeding mechanism and the center of the weld is calibrated, and the wire feeding compensation parameters are calculated in combination with the motion state of the execution end to obtain the guiding command to guide the welding wire into the molten pool; S3. Integrate the local spatial map and the guidance command to construct a follow-up safety protection space around the execution end, and generate a collaborative planning strategy that takes into account both wire feeding guidance and obstacle avoidance requirements through the attitude coordination transformation of the execution mechanism. S4. The collaborative planning strategy is converted into motion commands for the robotic arm joints, and the smoothing mechanism is used to drive the execution end to execute a continuous welding path that avoids obstacles while maintaining the wire feeding guidance constraint. S5. Record the trajectory data during the welding process, optimize the parameter system by comparing the degree of consistency between the measured trajectory and the planned trajectory, and output the path optimization scheme for different component characteristics.

2. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, The process of establishing the local spatial map in step S1 includes: projecting structured light stripes onto the workpiece surface using the vision sensor, calculating the depth, width, and edge coordinates of the weld seam based on the deformation information of the stripes; adjusting the photosensitive exposure parameters and image signal amplitude of the vision sensor to offset the influence of external ambient light and welding arc light on the brightness of image features, and fusing and reconstructing the extracted weld seam features with the point cloud data of the surrounding tooling fixtures.

3. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 2, characterized in that, When establishing the local spatial map, a laser ranging module and an infrared imaging module are integrated into the visual sensor to assist in the identification of the structured light stripes. The actual spacing value determined by the laser ranging module is used to proportionally correct the deformation information, and the infrared imaging module is used to extract the thermal feature boundary of the weld in a dense smoke environment, thereby improving the accuracy of spatial information acquisition in harsh industrial environments.

4. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, The process of obtaining the guiding command in step S2 includes: extracting the spatial trajectory of the weld centerline in the local spatial map, calculating the offset between the wire feeding nozzle centerline and the weld centerline; and simultaneously monitoring the curvature of the wire feeding hose. If the curvature exceeds a preset threshold, the wire feeding resistance is reduced by adjusting the spatial position of the execution end, thereby achieving dynamic compensation for the spatial position relationship.

5. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 4, characterized in that, When adjusting the spatial position relationship, the feedback value of the physical pressure sensor installed on the wire feeding mechanism is obtained. If the feedback value shows that the wire feeding resistance inside the wire feeding hose has increased abnormally, the execution end is automatically triggered to perform micro-translation along the extension direction of the wire feeding hose, thereby alleviating the wire feeding vibration during the execution of the guiding command through physical displacement.

6. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, The process of generating the collaborative planning strategy for obstacle avoidance requirements in step S3 includes: constructing a follow-up virtual flexible protective layer based on the geometric dimensions and motion envelope range of the execution end; predicting the collision risk by detecting the distance between the virtual flexible protective layer and obstacles in the local spatial map; if the collision risk reaches the warning threshold, maintaining the spatial coordinates of the wire feeding contact point unchanged, and changing the body posture of the execution end to bypass the obstacle.

7. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, When generating the collaborative planning strategy, the virtual flexible protection layer associates the trajectory information of the cable and air pipe dragged at the rear of the execution end. By marking the position of the cable and air pipe on the map, and reserving a safe gap to prevent the cable and air pipe from getting entangled with the obstacle and the tooling fixture when performing attitude collaborative transformation.

8. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, The execution process of the motion command in step S4 includes: subdividing the displacement of the robotic arm joint using an interpolation mechanism, limiting the rotation rate of the execution end to eliminate sudden motion changes, ensuring that the wire feeding nozzle stably points to the center of the molten pool during the attitude change process, and not triggering the overload protection of the robotic arm motor.

9. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 1, characterized in that, The specific process of parameter system optimization in step S5 includes: extracting trajectory data feedback during the welding process and comparing it with the preset planned trajectory; using the deviation value generated by the comparison to analyze the influence factors of workpiece thermal deformation and mechanical wear on path accuracy, and accordingly correcting the weight parameters in the collaborative planning strategy to establish the optimal control logic for a specific component.

10. The visual perception-based wire feeding guidance and anti-collision collaborative welding path planning method as described in claim 9, characterized in that, When optimizing the parameter system, the deviation value is correlated with the preset thermal expansion and contraction characteristics data of the workpiece material. By identifying the dynamic deformation law of the weld caused by the heat effect of welding, the initial planning trajectory of the next process and the next workpiece welding is automatically updated, thereby realizing the deep collaborative optimization of welding process and path planning scheme.

Images