Cross-station filtered and remaining area spray trajectory re-planning method and system

Abstract

The application discloses a kind of filtering and remaining area spray trajectory replanning method and system across station has been sprayed area, belong to industrial robot vision and automatic spraying technical field.This system is composed of mobile chassis, six-axis robot arm, terminal RGB-D camera, spray gun, station visual label and supporting function module;Robot generates standard spraying trajectory by visual segmentation, point cloud processing in first station and records sprayed area data, after moving across station, relies on visual label to complete multi-frame fusion relocation, calculates inter-station coordinate transformation matrix, maps historical sprayed area to current coordinate system and eliminates from real-time point cloud, re-plans serpentine spraying trajectory for remaining to be sprayed area, and accurately controls spray gun on-off in combination with trajectory segmentation logic.The application solves the problem that traditional mobile spraying robot is prone to repeated spraying, uneven paint film and repeated teaching when working across station, realizes multi-station continuous automatic spraying, and the trajectory runs smoothly with high spraying consistency.

Description

Technical Field

[0001] This invention relates to the fields of industrial robot control, machine vision, 3D point cloud processing, and automated spraying technology. Specifically, it relates to a method and system for filtering the already sprayed area and replanning the trajectory of the remaining area across workstations for mobile composite spraying robots. It is mainly applied to automated spraying and local touch-up painting scenarios across workshops and workstations for curved workpieces such as automobiles, doors, bumpers, various sheet metal parts, and experimental car bodies. Background Technology

[0002] With the widespread adoption of automated painting technology in the automotive manufacturing industry, mobile composite painting robots, with their advantages of flexibility and mobility, are gradually replacing fixed painting equipment and manual painting operations. Existing painting solutions suffer from the following main drawbacks: First, traditional painting robots rely on offline programming or manual station-by-station teaching. When the workpiece posture or painting area changes, recalibration is required, resulting in a large workload and low automation when operating across stations. Second, conventional 3D vision painting solutions are only suitable for single-station, single-operation tasks, failing to consider coordinate system shifts after robot movement. When operating across stations, they cannot identify previously painted areas, easily leading to repeated painting and quality defects such as excessively thick paint films, paint runs, and color differences. Third, existing trajectory planning schemes do not distinguish between effective painting segments and inter-row transition segments. When the robotic arm turns at the end of a row, the spray gun remains on, resulting in non-target painting and further reducing painting quality. Fourth, existing visual positioning technology is only used for robot navigation to stations, failing to correlate tag positioning results with point cloud data, painting trajectories, and historical operation data, thus hindering cross-station data reuse and intelligent replanning.

[0003] In summary, existing technologies are insufficient to meet the production needs of continuous multi-station operation of mobile spraying robots. The industry urgently needs a technical solution that can achieve cross-station coordinate unification, automatic filtering of sprayed areas, and dynamic replanning of spraying trajectories. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for filtering sprayed areas across workstations and replanning the spraying trajectory of remaining areas. In a car painting composite robot with a mobile chassis equipped with a six-axis robotic arm, spray gun, and end-effector RGB-D camera, after the robot moves across workshops or workstations, how to use the workstation visual label repositioning results to transform the historically sprayed areas to the current workstation coordinate system, remove the sprayed areas from the current visual point cloud, generate a spraying trajectory only for the remaining areas to be sprayed, and control the spray gun switch according to the effective spraying segment, thereby reducing repeated spraying, reducing manual teaching, and improving spraying consistency.

[0005] In a first aspect, embodiments of this application provide a method for filtering already sprayed areas across workstations and replanning the spraying trajectory of remaining areas, comprising the following steps: S1. The robot moves to the first spraying station. The six-axis robotic arm drives the end-effector RGB-D camera to the acquisition posture to acquire the RGB image and depth map of the workpiece to be sprayed. The coordinate transformation is completed by combining the camera intrinsic parameters and the end-effector extrinsic parameters of the robotic arm to generate the point cloud of the area to be sprayed in the robot base coordinate system. S2. Obtain the mask of the area to be sprayed through visual segmentation, and perform downsampling, filtering and clustering on the point cloud within the mask area to extract the effective spraying point cloud; S3. Establish a local spraying coordinate system based on the centroid and average normal of the effective spraying point cloud, slice and mesh according to preset parameters, generate a serpentine spraying trajectory by combining the local surface normal, and divide the effective spraying segment and the inter-row transition segment. S4. Perform the spraying operation and simultaneously record the current workstation's slice plane, spraying coverage area, trajectory start and end index, and spray gun switch status data as historical sprayed area data. S5. The robot moves to the next spraying station. The end RGB-D camera recognizes the visual label of the station. The multi-frame fusion algorithm is used to obtain the pose of the label of the previous and next stations and calculates the coordinate transformation matrix between the stations. S6. Use the coordinate transformation matrix to map the historical sprayed area data to the current workstation coordinate system, collect the current workpiece RGB-D data and generate real-time point cloud, remove the point cloud corresponding to the sprayed area, and obtain the remaining point cloud to be sprayed. S7. Repeat steps S2-S3 to regenerate the spraying trajectory for the remaining dot cloud to be sprayed. Control the opening and closing of the spray gun according to the trajectory in segments, and perform spraying of the remaining area. Repeat until the entire workpiece is sprayed.

[0006] In one embodiment, the formula for transforming the camera coordinate system to the robot base coordinate system in step S1 is: Pb = Tbase_end * Tend_cam * Pc; Where Pc is a 3D point in the camera coordinate system, Tend_cam is the extrinsic parameter matrix of the camera relative to the end effector of the robot arm, Tbase_end is the pose matrix of the end effector of the robot arm relative to the robot base, and Pb is a 3D point in the base coordinate system.

[0007] In one embodiment, in step S2, visual segmentation is performed by manually selecting positive and negative samples to generate a mask, or by automatically segmenting the area to be sprayed based on a trained deep learning model; depth pixels outside the mask are set to zero, and only the effective depth information corresponding to the mask is retained.

[0008] In one embodiment, in step S3, the local spraying coordinate system takes the centroid of the point cloud as the origin, the average normal of the point cloud as the Z-axis, projects the preset reference direction onto the tangent plane of the curved surface to obtain the Y-axis, and obtains the X-axis by the cross product of the Y-axis and the Z-axis, thus forming a right-handed coordinate system.

[0009] In one embodiment, in step S3, when generating the spraying trajectory, the spray gun position is obtained by offsetting a preset spraying distance in the opposite direction along the local normal. Adjacent spraying rows adopt a serpentine pattern, and the transition segments between rows are smoothly connected by Bezier curves, B-splines, or double circular arcs.

[0010] In one embodiment, in step S5, the workstation visual label uses any one of AprilTag, ArUco, QR code, or circular target; the formula for calculating the transformation matrix between workstations is: Tprev_curr=Tcurr_tag*inv(Tprev_tag); Where Tprev_tag is the pose matrix of the previous workstation tag relative to the robot base; Tcurr_tag is the pose matrix of the current workstation tag relative to the robot base.

[0011] In one embodiment, in step S6, the historically sprayed area is expressed in at least one of the following forms: slice plane, spray bounding box, trajectory envelope, and trajectory start and end index. The sprayed point cloud is removed by spatial location determination.

[0012] In one embodiment, in steps S3 and S7, the trajectory is resampled at a fixed period, the spray gun is turned on within the spraying row, and the spray gun is turned off in the inter-row transition section and the empty movement section.

[0013] Secondly, embodiments of this application also provide a cross-station sprayed area filtering and remaining area spray trajectory replanning system, used to implement the cross-station sprayed area filtering and remaining area spray trajectory replanning method described in the first aspect, characterized in that it includes a mobile chassis, a six-axis robotic arm, an end effector RGB-D camera, a spray gun, station vision labels, and a controller; wherein; The controller integrates a visual segmentation module, a point cloud processing module, a sprayed area recording module, a cross-workstation coordinate transformation module, a remaining area replanning module, a spray gun switch control module, and a trajectory execution module. The mobile chassis carries a six-axis robotic arm to move across workstations; the six-axis robotic arm is equipped with an RGB-D camera and a spray gun to complete image acquisition and spraying operations. The visual segmentation module is used to generate a mask for the area to be sprayed; the point cloud processing module performs point cloud filtering and coordinate transformation; and the sprayed area recording module stores spraying-related data for each station. The cross-station coordinate transformation module calculates the transformation matrix by combining the visual label pose; the remaining area replanning module completes the filtering of the sprayed area and the generation of a new trajectory; the spray gun switch control module and the trajectory execution module control the on / off of the spray gun and the movement of the robotic arm, respectively.

[0014] In one embodiment, the end RGB-D camera can be replaced by a binocular camera, a structured light camera, or a ToF camera; the workstation visual labels are placed on the workshop floor, walls, workpiece, or tooling surface.

[0015] The method and system for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations provided by this invention have the following significant advantages compared to existing technologies: 1. Reduce the risk of repeated spraying across workstations: By transforming coordinates across workstations and removing point clouds from already sprayed areas, historical sprayed areas are precisely filtered out, completely solving problems such as paint film overlap, sagging, and color difference in cross-workstation operations; by distinguishing between spraying sections and transition sections and linking the spray gun switch, non-target spraying between rows is eliminated, and the uniformity of the paint surface is greatly improved.

[0016] 2. Reduce manual teaching workload: Relying on visual segmentation, automatic point cloud processing and trajectory replanning technology, there is no need for repeated manual teaching and calibration when switching workpiece shape and workstation. Fully automatic operation can be achieved with only a small number of initial samples for training, reducing labor costs and debugging time.

[0017] 3. Improved adaptability of the spraying area: The local spraying coordinate system and local normal estimation enable the system to adapt to different surface shapes such as the whole vehicle, local touch-up paint, bumpers, doors, sheet metal parts and experimental vehicle body surfaces.

[0018] 4. Improve the quality of inter-row transitions: Close the spray gun at the inter-row transition section and use rounded corners or smooth transition trajectories to reduce end-of-row pauses, jitter, and overspray.

[0019] 5. Improve the continuity of mobile composite robot operations: The mobile chassis can move between workshops or workstations, and the visual label repositioning results directly serve the unification of spraying coordinates and trajectory replanning.

[0020] 6. Improve the reliability of trajectory execution: The trajectory is issued after resampling, tool coordinate transformation, batch inverse kinematics and joint limitation checks, which can reduce the risk of abnormal execution of the robotic arm. Attached Figure Description

[0021] Figure 1 This is a structural diagram of the mobile composite spraying robot system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the overall process of cross-station spraying according to an embodiment of the present invention; Figure 3 This is a schematic diagram of mask-constrained point cloud generation according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the spray trajectory generation in an embodiment of the present invention; Figure 5 This is a schematic diagram of cross-station sprayed area mapping and filtering according to an embodiment of the present invention; Figure 6This is a diagram showing the correspondence between the spray gun switch and the trajectory segment in an embodiment of the present invention; Figure 7 This is a schematic diagram of tag multi-frame fusion and workstation repositioning according to an embodiment of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that when a component is said to be "mounted on" another component, it can be directly mounted on the other component or may be interspersed with a component. When a component is said to be "set on" another component, it can be directly set on the other component or may be interspersed with a component. When a component is said to be "fixed to" another component, it can be directly fixed to the other component or may be interspersed with a component.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or / and" as used herein includes any and all combinations of one or more of the associated listed items.

[0025] Please see Figures 1 to 7 In a first aspect, this embodiment provides a cross-station sprayed area filtering and remaining area spraying trajectory replanning system for implementing the cross-station sprayed area filtering and remaining area spraying trajectory replanning method of the first aspect. The system is characterized by comprising a mobile chassis, a six-axis robotic arm, an end effector RGB-D camera, a spray gun, station visual labels, and a controller. The controller integrates a visual segmentation module, a point cloud processing module, a sprayed area recording module, a cross-station coordinate transformation module, a remaining area replanning module, a spray gun switch control module, and a trajectory execution module. The mobile chassis carries the six-axis robotic arm to achieve cross-station movement. The six-axis robotic arm is equipped with an RGB-D camera and a spray gun to complete image acquisition and spraying operations. The visual segmentation module generates a mask for the area to be sprayed. The point cloud processing module performs point cloud filtering and coordinate transformation. The sprayed area recording module stores spraying-related data for each station. The cross-station coordinate transformation module calculates the transformation matrix based on the visual label pose. The remaining area replanning module completes sprayed area filtering and new trajectory generation. The spray gun switch control module and the trajectory execution module control the on / off state of the spray gun and the movement of the robotic arm, respectively.

[0026]

[0027] In some alternative implementations, the end RGB-D camera can be replaced with a binocular camera, a structured light camera, or a ToF camera; workstation visual tags are placed on the workshop floor, walls, workpiece, or tooling surface.

[0028] Secondly, this embodiment also provides a method for filtering the already sprayed area across workstations and replanning the spraying trajectory of the remaining area, including the following steps: S1. The robot moves to the first spraying station. The six-axis robotic arm drives the end-effector RGB-D camera to the acquisition posture to acquire the RGB image and depth map of the workpiece to be sprayed. The coordinate transformation is completed by combining the camera intrinsic parameters and the end-effector extrinsic parameters of the robotic arm to generate the point cloud of the area to be sprayed in the robot base coordinate system. S2. Obtain the mask of the area to be sprayed through visual segmentation, and perform downsampling, filtering and clustering on the point cloud within the mask area to extract the effective spraying point cloud; S3. Establish a local spraying coordinate system based on the centroid and average normal of the effective spraying point cloud, slice and mesh according to preset parameters, generate a serpentine spraying trajectory by combining the local surface normal, and divide the effective spraying segment and the inter-row transition segment. S4. Perform the spraying operation and simultaneously record the current workstation's slice plane, spraying coverage area, trajectory start and end index, and spray gun switch status data as historical sprayed area data. S5. The robot moves to the next spraying station. The end RGB-D camera recognizes the visual label of the station. The multi-frame fusion algorithm is used to obtain the pose of the label of the previous and next stations and calculates the coordinate transformation matrix between the stations. S6. Use the coordinate transformation matrix to map the historical sprayed area data to the current workstation coordinate system, collect the current workpiece RGB-D data and generate real-time point cloud, remove the point cloud corresponding to the sprayed area, and obtain the remaining point cloud to be sprayed. S7. Repeat steps S2-S3 to regenerate the spraying trajectory for the remaining dot cloud to be sprayed. Control the opening and closing of the spray gun according to the trajectory in segments, and perform spraying of the remaining area. Repeat until the entire workpiece is sprayed.

[0029] To achieve cross-workstation visual label relocation, in some optional implementations, the transformation formula from the camera coordinate system to the robot base coordinate system in step S1 is: Pb = Tbase_end * Tend_cam * Pc; Where Pc is a 3D point in the camera coordinate system, Tend_cam is the extrinsic parameter matrix of the camera relative to the end effector of the robot arm, Tbase_end is the pose matrix of the end effector of the robot arm relative to the robot base, and Pb is a 3D point in the base coordinate system.

[0030] In some of the alternative implementations, in step S2, visual segmentation is performed by manually selecting positive and negative samples to generate a mask, or by automatically segmenting the area to be sprayed based on a trained deep learning model; depth pixels outside the mask are set to zero, and only the effective depth information corresponding to the mask is retained.

[0031] In some optional embodiments, in step S3, the local spraying coordinate system takes the centroid of the point cloud as the origin, the average normal of the point cloud as the Z-axis, projects the preset reference direction onto the tangent plane of the curved surface to obtain the Y-axis, and obtains the X-axis by the cross product of the Y-axis and the Z-axis, thus forming a right-handed coordinate system.

[0032] To achieve the goal of generating point clouds of the area to be sprayed, in some optional implementations, after the end-effector RGB-D camera acquires RGB images and depth maps, the visual segmentation module generates a mask for the area to be sprayed based on manually selected positive and negative cue points. In the initial stage when the amount of training data is small, manual selection can ensure the accuracy of the boundaries of the area to be sprayed; when the historical data accumulates to a preset amount, an automatic recognition model can be trained to directly output the mask for the area to be sprayed.

[0033] After the mask is generated, the depth outside the mask is set to zero, and only the pixels with effective depth inside the mask are retained. Let the pixel coordinates be (u, v), the depth be z, and the camera intrinsic parameters be fx, fy, cx, cy. Then, the 3D point in the camera coordinate system is: x=(u-cx)·z / fx, y=(v-cy)·z / fy, z=z.

[0034] Furthermore, based on the robot arm's end-effector pose T_base_end and the extrinsic parameters T_end_cam or T_cam_end from the camera to the end-effector, the point cloud is transformed from the camera coordinate system to the robot's base coordinate system. This process allows for the acquisition of a uniform point cloud of the area to be sprayed, even when the camera is mounted at the end of the robot arm, regardless of the robot arm's different shooting postures.

[0035] To remove background, outlier, and non-target points, this invention performs voxel downsampling, clustering, and statistical filtering on the point cloud of the area to be sprayed, extracting the largest effective spray cluster. Then, a local spray coordinate system is established based on the centroid and average normal of this cluster. The local spray coordinate system is established as follows: the centroid of the point cloud is used as the origin, and the average normal is used as the local z-axis; a preset reference direction is projected onto the tangent plane of the surface to be sprayed as the local y-axis; the local x-axis is obtained by cross-product of the y-axis and z-axis, and the right-handed coordinate system is verified. Through this coordinate system, spray areas with different orientations, installation angles, and surfaces can be unified into a local space that facilitates slicing.

[0036] To avoid trajectory discontinuities caused by the robotic arm abruptly stopping or turning back at the end of a row, this invention generates rounded or smooth transition trajectories between adjacent spraying rows. The transition trajectory can be formed by Bézier curves, B-spline curves, or double circular arc curves, and is generated based on the endpoints, tangential direction, and normal direction of adjacent rows. The spray gun switch control is bound to the trajectory structure: the spray gun is activated during effective spraying segments within a spraying row; the spray gun is deactivated during inter-row transition segments, invalid point cloud segments, and empty line movement segments. After trajectory resampling, the system generates a spray gun switch sequence based on the start and end points or start and end indices of effective spraying segments, thereby avoiding non-target spraying at inter-row transition points.

[0037] In some alternative implementations, the trajectory is resampled at 0.002 s intervals; the spraying speed, row spacing, and spraying distance can be configured by task parameters. The transition speed can be lower than the spraying row speed to improve the continuity of the robotic arm's movement and the stability of the spraying boundaries. In some of the optional implementations, in step S3, when generating the spraying trajectory, the spray gun position is obtained by offsetting a preset spraying distance in the opposite direction along the local normal. Adjacent spraying rows adopt a serpentine pattern, and the transition sections between rows are smoothly connected by Bezier curves, B-splines, or double circular arcs.

[0038] To achieve cross-workstation visual label relocation, in some optional implementations, in step S5, the workstation visual label uses any one of AprilTag, ArUco, QR code, or circular target; the formula for calculating the inter-workstation transformation matrix is: Tprev_curr=Tcurr_tag*inv(Tprev_tag); Where Tprev_tag is the pose matrix of the previous workstation tag relative to the robot base; Tcurr_tag is the pose matrix of the current workstation tag relative to the robot base.

[0039] In some optional implementations, in step S6, the historically sprayed area is expressed in at least one of the following forms: slice plane, spray bounding box, trajectory envelope, and trajectory start and end index. The sprayed point cloud is removed by spatial location determination.

[0040] In some of the alternative implementations, in steps S3 and S7, the trajectory is resampled at a fixed period, the spray gun is turned on within the spraying row, and the spray gun is turned off in the inter-row transition section and the empty movement section.

[0041] To achieve the purpose of filtering the sprayed area and replanning the remaining area, the sprayed area can be represented in at least one form: 1. The slice plane and its normal corresponding to each spraying row; 2. The set of coverage points, bounding box, polygon or bounding box corresponding to each spraying row; 3. The spraying envelope formed by the expansion of the spraying trajectory points along the spraying width direction and the spraying distance direction; 4. The effective trajectory segment and its coverage area corresponding to the spraying start and end index.

[0042] After the robot arrives at the new workstation and acquires the current point cloud, it transforms the already sprayed area of ​​the previous workstation to the coordinate system of the current workstation using Tprev_curr. Then, it checks whether each point in the current point cloud falls within the sprayed area. If a point is within the sprayed area, it is removed from the current point cloud; if the point is outside the sprayed area, it is retained as part of the remaining area to be sprayed.

[0043] In a simplified implementation, the directional distance to the current point cloud is determined using the slice plane of the last row or several rows of the previous station, retaining the unsprayed point cloud on one side of the slice plane. More generally, multiple slice planes, trajectory envelopes, polygon boundaries, or combinations of point cloud registration results can be used to form filtering conditions.

[0044] After filtering the already sprayed areas, the system re-executes effective point cloud extraction, local coordinate system establishment, slicing, local normal estimation, normal offset, serpentine trajectory generation, inter-row smoothing, and spray gun on / off control for the remaining areas to be sprayed. This achieves anti-overspraying and automatic completion of remaining areas in continuous cross-station spraying.

[0045] It should be noted that after the spraying trajectory is generated, the pose of the spray gun tool coordinate system is converted into the pose of the robotic arm end effector coordinate system. The system converts the end effector pose sequence into the pose format required by the robotic arm controller, calls the batch inverse kinematics service to obtain the joint angle sequence, and checks the trajectory validity based on the joint velocity, joint acceleration, and inverse kinematics success flag.

[0046] For trajectory segments that are successfully reverse engineered, the system issues offline or continuous trajectory execution commands; for trajectory segments that fail to reverse engineer or exceed motion limits, partial trajectory deletion, speed reduction, attitude adjustment, or replanning can be performed. The spray gun switch signal is issued synchronously with the trajectory start and end index, ensuring that the spray gun is only turned on during effective spraying segments.

[0047] It should be noted that, preferably, the visual label can be AprilTag, ArUco, QR code, dot target, or other artificially created identifiers whose pose can be estimated.

[0048] Preferably, the RGB-D camera can be replaced with a structured light camera, a binocular camera, a ToF camera, or a laser profilometer.

[0049] Preferably, the segmentation of the area to be sprayed can be achieved by manual point selection segmentation, traditional image segmentation, deep learning instance segmentation, or semantic segmentation.

[0050] Preferably, the filtering of the sprayed area can be achieved based on the slice plane, directed bounding box, trajectory envelope, voxel occupancy map, point cloud nearest neighbor distance or multi-site point cloud registration.

[0051] Preferably, the inter-row transition can be achieved using Bézier curves, B-spline curves, circular arc transitions, or velocity look-ahead interpolation methods.

[0052] Preferably, the present invention can be used for whole vehicle painting, partial touch-up painting, bumpers, car doors, sheet metal parts, experimental vehicle body curved surfaces, and other curved workpieces that require painting by a mobile composite robot.

[0053] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0054] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for filtering already sprayed areas and replanning spraying trajectories for remaining areas across workstations, characterized in that, Includes the following steps: S1. The robot moves to the first spraying station. The six-axis robotic arm drives the end-effector RGB-D camera to the acquisition posture to acquire the RGB image and depth map of the workpiece to be sprayed. The coordinate transformation is completed by combining the camera intrinsic parameters and the end-effector extrinsic parameters of the robotic arm to generate the point cloud of the area to be sprayed in the robot base coordinate system. S2. Obtain the mask of the area to be sprayed through visual segmentation, and perform downsampling, filtering and clustering on the point cloud within the mask area to extract the effective spraying point cloud; S3. Establish a local spraying coordinate system based on the centroid and average normal of the effective spraying point cloud, slice and mesh according to preset parameters, generate a serpentine spraying trajectory by combining the local surface normal, and divide the effective spraying segment and the inter-row transition segment. S4. Perform the spraying operation and simultaneously record the current workstation's slice plane, spraying coverage area, trajectory start and end index, and spray gun switch status data as historical sprayed area data. S5. The robot moves to the next spraying station, the end RGB-D camera recognizes the visual label of the station, and the multi-frame fusion algorithm is used to obtain the pose of the label of the previous and next stations, and calculates the coordinate transformation matrix between the stations. S6. Use the coordinate transformation matrix to map the historical sprayed area data to the current workstation coordinate system, collect the current workpiece RGB-D data and generate real-time point cloud, remove the point cloud corresponding to the sprayed area, and obtain the remaining point cloud to be sprayed. S7. Repeat steps S2-S3 to regenerate the spraying trajectory for the remaining dot cloud to be sprayed. Control the opening and closing of the spray gun according to the trajectory in segments, and perform spraying of the remaining area. Repeat until the entire workpiece is sprayed.

2. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S1, the formula for transforming the camera coordinate system to the robot base coordinate system is: Pb = Tbase_end * Tend_cam * Pc; Where Pc is a 3D point in the camera coordinate system, Tend_cam is the extrinsic parameter matrix of the camera relative to the end effector of the robot arm, Tbase_end is the pose matrix of the end effector of the robot arm relative to the robot base, and Pb is a 3D point in the base coordinate system.

3. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S2, visual segmentation is performed by manually selecting positive and negative samples to generate a mask, or by automatically segmenting the area to be sprayed based on a trained deep learning model; depth pixels outside the mask are set to zero, and only the effective depth information corresponding to the mask is retained.

4. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S3, the local spraying coordinate system takes the centroid of the point cloud as the origin, the average normal of the point cloud as the Z-axis, projects the preset reference direction onto the tangent plane of the curved surface to obtain the Y-axis, and obtains the X-axis by the cross product of the Y-axis and the Z-axis, thus forming a right-handed coordinate system.

5. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S3, when generating the spraying trajectory, the spray gun position is obtained by offsetting the preset spraying distance in the opposite direction along the local normal. Adjacent spraying rows adopt a serpentine pattern, and the transition sections between rows are smoothly connected by Bézier curves, B-splines, or double circular arcs.

6. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S5, the visual label for the workstation can be any one of AprilTag, ArUco, QR code, or circular target; the formula for calculating the transformation matrix between workstations is: Tprev_curr=Tcurr_tag*inv(Tprev_tag); Where Tprev_tag is the pose matrix of the previous workstation tag relative to the robot base; Tcurr_tag is the pose matrix of the current workstation tag relative to the robot base.

7. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In step S6, the historically sprayed area is represented by at least one of the following forms: slice plane, spray bounding box, trajectory envelope, and trajectory start and end index. The sprayed point cloud is removed by spatial location determination.

8. The method for filtering already sprayed areas and replanning spraying trajectories in remaining areas across workstations according to claim 1, characterized in that, In steps S3 and S7, the trajectory is resampled at a fixed period, the spray gun is turned on within the spraying row, and the spray gun is turned off in the inter-row transition section and empty movement section.

9. A cross-station sprayed area filtering and remaining area spray trajectory replanning system, used to implement the cross-station sprayed area filtering and remaining area spray trajectory replanning method as described in any one of claims 1-8, characterized in that, This includes a mobile chassis, a six-axis robotic arm, an end effector RGB-D camera, spray guns, workstation vision labels, and a controller; among which, The controller integrates a visual segmentation module, a point cloud processing module, a sprayed area recording module, a cross-workstation coordinate transformation module, a remaining area replanning module, a spray gun switch control module, and a trajectory execution module. The mobile chassis carries a six-axis robotic arm to move across workstations; the six-axis robotic arm is equipped with an RGB-D camera and a spray gun to complete image acquisition and spraying operations. The visual segmentation module is used to generate a mask for the area to be sprayed; the point cloud processing module performs point cloud filtering and coordinate transformation; and the sprayed area recording module stores spraying-related data for each station. The cross-station coordinate transformation module calculates the transformation matrix by combining the visual label pose; the remaining area replanning module completes the filtering of the sprayed area and the generation of a new trajectory; the spray gun switch control module and the trajectory execution module control the on / off of the spray gun and the movement of the robotic arm, respectively.

10. The cross-station sprayed area filtering and remaining area spray trajectory replanning system according to claim 9, characterized in that, The terminal RGB-D camera can be replaced by a binocular camera, a structured light camera, or a ToF camera; Workstation visual labels are placed on the workshop floor, walls, workpieces, or tooling surfaces.

Images