Steel structural member welding seam information acquisition and intelligent planning method

Abstract

The invention belongs to the field of weld joint intelligent planning systems, and particularly relates to a steel structural part weld joint information acquisition and intelligent planning method which comprises the following steps: scanning a workpiece, acquiring scanning data and preprocessing the scanning data to obtain preprocessed scanning data; abrupt change points in the Z direction are detected in the scanning data, and the abrupt change points are divided into an upper edge feature point set and a lower edge feature point set to serve as workpiece contours; according to the contour of the workpiece, different types of welding seams meeting the requirements are detected in the mutation point set, and needed welding seam information is obtained; all welding seam information is traversed, and the welding sequence is planned; generating a weld transition point path according to the planning information of the welding sequence; and in combination with a robot operation instruction, according to the generated weld transition point path, issuing the path to the robot to execute operation through a protocol. The collision-free optimal motion path of the tail end of the welding gun is integrally planned according to the welding sequence, the use difficulty and complexity of the welding robot are reduced, and a large amount of labor cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of weld intelligent planning systems, and specifically relates to a method for obtaining and intelligently planning weld information of steel structure parts. Background Art

[0002] For existing steel component welding robots, before the welding operation starts, the robot must be taught, and a series of welding parameters must be configured. That is, the operator manually guides the robot to move along the actual operation trajectory in advance step by step, and records each action and other corresponding working parameters in this process. After the teaching is completed, a start command needs to be given to the welding robot, and the welding robot will accurately reproduce all the taught operations to complete the welding operation. Teaching and reproduction is a common form when the welding structure deviation of the robot is extremely small and the workpieces are mass-produced. However, in the case of large differences in the shapes of components and small batches, the teaching process will take a large amount of time of the operator, and the high training threshold results in too high labor costs. The working efficiency of the robot cannot meet the production requirements, or the accuracy decreases due to frequent teaching.

[0003] To solve the problems of labor cost and efficiency, the teaching-free welding technology has emerged. In the existing technical solutions related to teaching-free, usually, a vision or laser sensor is used to scan the workpiece to identify the weld position, and then it is imported into a digital model or drawing for comparison. After the operation configuration is completed, the welding operation is generated and sent to the welding robot. However, this technical solution still has many problems, such as: ① If the actual workpiece deviates greatly from the digital model or drawing, or there is even no digital model or drawing, the welding operation cannot be directly executed; ② After the sensor scans and identifies the weld position, a large amount of manual participation is still required for the configuration work, and the workload is not much reduced compared with the teaching process; ③ Once any local change occurs to the workpiece, the operation needs to be reconfigured. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for obtaining and intelligently planning weld information of steel structure parts. Through the scanned data of the sensor, the weld information is deeply extracted and intelligently analyzed, so that the system can automatically classify the welds, generate the welding sequence, and overall plan the collision-free optimal movement path of the end of the welding torch according to the welding sequence. This system has strong adaptability. In the case of no digital model or drawing, only a small amount of manual participation is required to output an operation file for the welding robot, reducing labor costs, improving the convenience of the welding robot, and improving production accuracy and production efficiency.

[0005] The technical solution adopted by the present invention to achieve the above purpose is: a method for obtaining and intelligently planning weld information of steel structure parts, including the following steps:

[0006] 1) The robot control cabinet configures the welding parameter file, starts the welding robot system, and the vision sensor scans the workpiece along the moving direction of the gantry to obtain the scan data; after receiving the scan data, the robot control cabinet performs preprocessing to obtain the preprocessed scan data;

[0007] 2) Based on the height change and continuity in the scan data, the robot control cabinet detects the mutation points in the Z direction in the scan data, and divides the mutation points into two types of feature point sets, the upper edge and the lower edge, as the workpiece contour, so as to distinguish the upper edge points and the edge points through the mutation points;

[0008] 3) Detect the weld: According to the workpiece contour, detect different types of welds that meet the requirements in the set of mutation points, and obtain the required weld information based on the weld positions and the points near the welds corresponding to different types of welds;

[0009] 4) Traverse all the weld information and plan the welding sequence;

[0010] 5) Generate the weld transition point path according to the planned information of the welding sequence;

[0011] 6) Combine with the robot operation instructions, and send the position, posture, type of each weld operation task, the preset welding parameters, and the weld transition point path generated in step 5) to the robot through the protocol to execute the operation.

[0012] The welding robot system includes: a welding robot, a gantry, a sensor, and a robot control cabinet;

[0013] The robot control cabinet is used to receive the scan data detected by the vision sensor during the welding process of the welding robot, process the scan data during the welding process of the welding robot to obtain the weld transition point path, and send the motion and operation instructions according to the path to the gantry and the welding robot respectively in combination with each weld operation task and the preset welding parameters;

[0014] The gantry is provided with a welding robot and is connected to the robot control cabinet, and is used to drive the welding robot to move and perform operations according to the position, posture, type of each weld operation task, the preset welding parameters, and the weld transition point path;

[0015] The welding robot is used to perform welding operations on the workpiece to be processed;

[0016] The vision sensor is arranged at the welding occurrence end of the welding robot and is used to collect the scan data during the welding process of the welding robot and send the scan data to the robot control cabinet for processing.

[0017] The preprocessing of the scan data obtained by the sensor scanning the workpiece by the robot control cabinet includes the following steps:

[0018] 1-1) The sensor is activated to scan the workpiece, obtain the camera coordinates of the workpiece in the scene, and send them to the robot control cabinet;

[0019] 1-2) The robot control cabinet converts the camera coordinates of the workpiece in the camera coordinate system into coordinates in the robot working system;

[0020] 1-3) The robot control cabinet processes the scanning data noise through an integrated filter and interpolates the actual points not scanned by the sensor by the linear interpolation method.

[0021] The specific steps of step 3) are as follows:

[0022] 2-1) Find the inflection points in the workpiece contour, divide the contour into several segments according to the inflection points, and perform straight line detection and circular arc detection on each segment respectively to obtain a set of straight line points or a set of circular arc points that meet the conditions;

[0023] 2-2) For each set of straight line points, find the best-fitting straight line equation through the RANSAC method, project the points on all point sets onto the straight line, find the two endpoints of the straight line, and these two points form a descriptor of a straight line segment; furthermore, obtain the weld information of the straight line weld;

[0024] 2-3) For each set of circular arc points, find the best-fitting spatial circle equation through the RANSAC method, determine the start and end points of the circular arc segment according to the angle between each point and the center of the circle in the spatial circle plane, and take the midpoint of the two points on the circular arc. These three points form a descriptor of a circular arc segment;

[0025] Find the normalized vector Q1 pointing to the center of the circle. The normalized vector in the positive direction of the working plane is Q2. Then the welding torch posture at this point is Q = Q1 + Q2; according to step 2-3), furthermore, obtain the weld information of the straight line weld.

[0026] The weld information includes: the weld information of the straight line weld and the weld information of the circular arc weld;

[0027] Among them, the weld information of the straight line weld includes: the position and posture of the starting point, the position and posture of the ending point, the plane information on both sides of the straight line, and the orientation of the straight line;

[0028] The weld information of the circular arc weld includes: the position and posture of the starting point of the circular arc, the position and posture of the midpoint of the circular arc, the position and posture of the ending point of the circular arc, and the mathematical expression of the circular arc segment.

[0029] The specific steps of traversing all weld information and planning the welding sequence are as follows:

[0030] 3-1) Select the Euclidean distance from the origin of the working plane to the weld ends of all weld information, and take the point corresponding to the minimum value as the starting weld;

[0031] 3-2) Traverse all welds. If there is a paired weld for a weld, select its corresponding weld as the next one; if not, calculate the differences of the six degrees of freedom between the end point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld.

[0032] In step 3-2), when there is no pairing, calculate the differences of the six degrees of freedom between the end point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld. Specifically:

[0033] (1) Calculate the differences of the six degrees of freedom between the end point of the current weld and the starting points of the remaining welds respectively, that is:

[0034] Cost 总体 =a1*(X i -X i+1 )+a2*(Y i -Y i+1 )+a3*(Z i -Z i+1 )++a4*(W i -W i+1 )+a5*(P i- -P i+1 )+a6*(R i -R i+1 )

[0035] Among them, a1 to a6 are weight coefficients, XYZ are the coordinates of the three spatial dimensions, and WPR is the axis angle representing the attitude, corresponding to the XYZ axis order;

[0036] (2) Select the point with the lowest Cost in Cost 总体 as the next weld.

[0037] In step 5), generating the weld transition point path according to the welding sequence planning information includes the following steps:

[0038] 4-1) Create a 3D grid map:

[0039] Determine the working plane and the map range according to the original point cloud. According to the upper edge welds and the pairing information between welds, resample the spatial point cloud as the map point cloud, rasterize the map point cloud according to the preset resolution, and set obstacles at the positions where there are points in space;

[0040] 4-2) Calculate the Euclidean distance between the current weld end point and the start point of the next weld. If the Euclidean distance is less than the set minimum value, no transition point is generated; if the Euclidean distance is greater than the maximum value, raise the welding torch to the safe height according to the current pose, move above the start point of the next weld and change the pose to the same as that of this point, and then lower the welding torch vertically;

[0041] 4-3) If the Euclidean distance between the current weld end point and the start point of the next weld is within the set threshold range, in the grid map created in step 4-1), using the 3D-A* algorithm, take the Euclidean distance between the current position and the end point as the heuristic function, search for all path points, then convert the path points into point clouds, and linearly interpolate the pose values of each point according to the number of point clouds to generate the path point poses.

[0042] The specific content of step 4-2) is as follows:

[0043] a. Calculate the Euclidean distance D between the current weld end point P1(X1, Y1, Z1, W1, P1, R1) and the start point P2(X2, Y2, Z2, W2, P2, R2) of the next weld;

[0044] b. Preset two threshold parameters S1 and S2, and S1 < S2, and a torch-raising height parameter Z safe ;

[0045] When D < S1, no transition point is generated;

[0046] When D > S2, generate 2 transition points, which are (X1, Y1, Z1 + Z safe , W1, P1, R1) and (X2, Y2, Z2 + Z safe , W2, P2, R2).;

[0047] Among them, the first transition point is the point above the current weld; the second transition point is: the point above the start point of the next weld; determine the movement trajectory order of the welding robot as follows: the current end point, the first transition point, the second transition point, the start point of the next weld.

[0048] The specific content of step 4-3) is as follows:

[0049] a) When the threshold parameter S1 < Euclidean distance D < threshold parameter S2, convert the current weld end point P1 and the start point P2 of the next weld into grid map coordinates and set the corresponding positions as passable areas;

[0050] b) Start searching from the P1 position in the grid space, move one grid each time, and the search directions are: 6 directions of X+, X-, Y+, Y-, Z+, Z-;

[0051] c) Use the sum of the current path length and the Euclidean distance from the current position to P2 as the search cost, and search for spatial path points in the map space;

[0052] d) Traverse all spatial path points. If multiple path points are on the same straight line segment, then delete the path points except the endpoints, and then inverse-transform the coordinates of the remaining n path points into point cloud coordinates P i (X i ,Y i ,Z i ), and perform linear interpolation on W1 to W2, P1 to P2, and R1 to R2 respectively according to the new number of path points to obtain the pose of the robot at each position;

[0053] The generated path is: point set

[0054] Among them,

[0055] The present invention has the following beneficial effects and advantages:

[0056] 1. Through the scanning data of the sensor, the present invention deeply extracts and intelligently analyzes the weld information, enabling the system to automatically classify the welds, generate the welding sequence, and overall plan the collision-free optimal motion path of the end of the welding torch according to the welding sequence, reducing the use difficulty and complexity of the welding robot and reducing a large amount of labor costs.

[0057] 2. The present invention has strong adaptability. In the case where there are differences between the workpiece and the drawing or there is no drawing, only one scan and a small amount of manual participation are required to directly output the operation file for the welding robot, which is sent to the robot through the protocol to directly start the welding task, improving the convenience of the welding robot, production accuracy, and production efficiency.

[0058] 3. For multiple complex welds, the present invention automatically plans the welding sequence according to the different characteristics (position / shape / size, etc.) of the welds, generates welding parameters, and completes the intelligent path planning of the robot at the transition position between the welds according to the welding sequence, making the welding task simple and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 The structural framework diagram of the weld information acquisition and intelligent planning system of the present invention;

[0060] Figure 2 The scanned workpiece point cloud map stitched by combining the sensor scanning data and encoder information in this embodiment;

[0061] Figure 3 The position map of the straight-line type weld and arc type weld extracted in this embodiment;

[0062] Figure 4 is the overall workflow diagram of the present invention;

[0063] Figure 5 is the workflow diagram for creating a 3D grid map and obtaining transition points using the 3D - A* search method in the present invention;

[0064] Figure 6 is the method flowchart for weld detection used in the present invention. Detailed implementation manners

[0065] The following further elaborates on the present invention in conjunction with the accompanying drawings and embodiments.

[0066] The content of the present invention is directed to multiple complex welds. According to the different characteristic attributes of the welds, it realizes weld position recognition, classification, automatically plans the welding sequence, generates welding parameters, and based on the welding sequence, completes the intelligent path planning of the robot at the transition position between welds to generate a welding operation.

[0067] To achieve the above - mentioned purpose, the present invention provides the following technical solutions:

[0068] Install a movable gantry above the workbench, and install a line - structured light scanning sensor and an encoder on the gantry as Figure 1 shown. Utilize the encoder position information to splice each frame of the sensor's scanning data to generate the overall point - cloud data of the workpiece to be welded;

[0069] The weld information acquisition and intelligent planning system includes: a welding robot, a gantry, a sensor, and a robot control cabinet;

[0070] The robot control cabinet is used to receive the scanning data of the welding process detected by the vision sensor for the welding robot, process the scanning data of the welding process of the welding robot to obtain the weld transition - point path, and respectively send the movement - along - the - path and operation instructions to the gantry and the welding robot in combination with the operation tasks of each weld and the preset welding parameters;

[0071] There is a welding robot on the gantry, which is connected to the robot control cabinet and is used to drive the welding robot to move and operate according to the position, posture, type of each weld operation task, the preset welding parameters, and the weld transition - point path;

[0072] The welding robot is used to perform welding operations on the workpiece to be processed;

[0073] The vision sensor is arranged at the welding occurrence end of the welding robot and is used to collect the scanning data of the welding process of the welding robot and send the scanning data to the robot control cabinet for processing.

[0074] Extract and intelligently plan the weld information for the point cloud data. The processing procedure is as follows:

[0075] Preprocess the scanned data before that:

[0076] 1. Respectively from the sensor scanning direction and the gantry moving direction, and execute:

[0077] 1-1) Start the sensor to scan the workpiece, obtain the camera coordinates of the workpiece in the scene, and send them to the robot control cabinet;

[0078] 1-2) The robot control cabinet converts the camera coordinates of the workpiece in the camera coordinate system into the coordinates in the robot working system;

[0079] 1-3) The robot control cabinet processes the scanning data noise through an integrated filter and interpolates the actual points not scanned by the sensor through the linear interpolation method.

[0080] 2. Find the mutation points according to the height change and continuity, and divide the mutation points into two types of feature point sets, the upper edge and the lower edge, as the workpiece contour;

[0081] 3. Detect the inflection points in the workpiece contour, divide the contour into several segments according to the inflection points, and perform straight line detection and circular arc detection on each segment respectively to obtain a set of straight line points / a set of circular arc points / a set of other points that meet the conditions;

[0082] As Figure 6 shown, it is the method flow chart of weld detection used in the present invention;

[0083] 3-1. For each set of straight line points, use the method based on RANSAC to find the best-fitting straight line equation, project the points on all point sets onto the straight line, find the two endpoints of the straight line, and these two points form a descriptor of a straight line segment;

[0084] 3-2. For each set of circular arc points, use the RANSAC method to find the best-fitting spatial circle equation (sphere equation and a spatial plane equation), determine the start and end points of the circular arc segment according to the angle between each point and the center of the circle in the spatial circle plane, take the midpoint of the two on the circular arc, and these three points form a descriptor of a circular arc segment; find its normalized vector Q1 pointing to the center of the circle, and the normalized vector of the positive direction of the working plane (bottom plate) is Q2, then the torch posture at this point is Q = Q1 + Q2;

[0085] Among them, according to the detected weld positions and the points near the welds, the required weld information is calculated respectively. The straight welds include: the position and attitude of the starting point, the position and attitude of the ending point, the plane information on both sides of the straight line, and the orientation of the straight line; the circular arc welds include: the position and attitude of the starting point of the circular arc, the position and attitude of the midpoint of the circular arc, the position and attitude of the ending point of the circular arc, and the mathematical expressions of the circular arc segment (spatial sphere equation and cross-section equation);

[0086] 4. Traverse all straight-segment welds, and pair the segments two by two according to the direction angle between two segments, the Euclidean distance between the endpoints of two segments, and the Euclidean distance between the midpoints of two segments, based on the set parameters. The unpaired cases are grouped into a separate set;

[0087] 5. Traverse all circular-arc segment welds, calculate the Euclidean distances of three points on each circular-arc segment in pairs, and pair the circular-arc segments two by two in combination with the center position. The unpaired cases are grouped into a separate set;

[0088] 6. Based on the obtained weld information, perform welding operation sequencing. First, select the endpoint of all welds that is closest to the origin of the working plane as the starting point; traverse all welds. If the weld has a paired weld, select its corresponding weld as the next one; if there is no pairing, calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld;

[0089] If there is no pairing, calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld. Specifically:

[0090] (1) Calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, that is:

[0091] Cost 总体 =a1*(X i -X i+1 )+a2*(Y i -Y i+1 )+a3*(Z i -Z i+1 )++a4*(W i -W i+1 )+a5*(P i- -P i+1 )+a6*(R i -R i+1 )

[0092] Among them, a1 to a6 are weight coefficients, XYZ are the coordinates of the three spatial dimensions, and WPR is the axis angle representing the attitude, corresponding to the XYZ axis sequence;

[0093] (2) Select Cost 总体 Select the point with the lowest Cost among them as the next weld seam.

[0094] 7. According to the arranged weld seams, plan the movement trajectory of the welding torch between two weld seams. As shown in Figure 5 , it is the flowchart of creating a 3D grid map and obtaining transition points using the 3D-A* search method in the present invention;

[0095] First, create a 3D grid map: Determine the working plane and map range according to the original point cloud. According to the upper edge weld seam and the pairing information between weld seams, resample the spatial point cloud as the map point cloud, rasterize the map point cloud according to the preset resolution, and set obstacles at the positions where there are points in space. Then calculate the distance between the end point of the current weld seam and the start point of the next weld seam. If it is too small, no transition point is generated; if it is too large, raise the welding torch to a safe height according to the current pose, move above the start point of the next weld seam and change the pose to be the same as that point, and then lower the welding torch vertically; if it is within the set threshold range, in the created grid map, use the 3D-A* algorithm, take the Euclidean distance between the current position and the end point as the heuristic function, search for all path points, and then convert the path points into point clouds, and linearly interpolate the pose values of each point according to the number of point clouds.

[0096] Generate the transition point path: Calculate the Euclidean distance D between the end point P1(X1, Y1, Z1, W1, P1, R1) of the current weld seam and the start point P2(X2, Y2, Z2, W2, P2, R2) of the next weld seam. Preset two threshold parameters S1, S2 (S1 < S2) and a torch lifting height parameter Z safe It is processed in three cases: ① D < S1, no transition point is generated; ② D > S2, generate 2 transition points, which are (X1, Y1, Z1 + Z safe , W1, P1, R1) and (X2, Y2, Z2 + Z safe , W2, P2, R2); ③ S1 < D < S2, convert P1 and P2 into grid map coordinates and set the corresponding positions as passable areas. Start searching from the P1 position in the grid space, move one grid each time, and the search directions are 6 directions of X+, X-, Y+, Y-, Z+, Z-. Take the sum of the current path length and the Euclidean distance from the current position to P2 as the search cost, and search for spatial path points in the map space. Inverse-transform the coordinates of all spatial path points into point cloud position coordinates, and linearly interpolate W1~W2, P1~P2, R1~R2 respectively according to the number of path points to generate the path point poses;

[0097] The generated path is: point set

[0098] Among them,

[0099] 8. Generate a welding operation file based on the positions and postures of the start / end points of all weld segments obtained above, the welding sequence, and the positions and postures of the corresponding transition points, and hand it over to the robot for execution;

[0100] Embodiment 1:

[0101] The working process of the present invention is as Figure 2 shown, and includes the following steps:

[0102] S1: Configure a parameter file, which involves parameters related to scanning, parameters and thresholds of algorithms in weld information extraction, and parameters and thresholds in the intelligent planning process of welds;

[0103] S2: Start the camera to scan the workpiece, obtain the camera coordinates of the workpiece in the scene, and then convert them into coordinates in the robot working system through coordinate system transformation;

[0104] S3: Process the noise problem of the scanned data through a comprehensive filter composed of median filtering, radius filtering, and statistical filtering, and interpolate the points that cannot be scanned by the camera but actually exist through linear interpolation;

[0105] S4: Detect the mutation points in the Z direction (perpendicular to the workbench direction) from the scanned data, and distinguish the upper edge points and edge points;

[0106] S5: Perform weld detection, and detect the required straight welds and circular arc welds from the set of mutation points. The specific process is as Figure 4 shown;

[0107] S6: According to the weld positions detected in step 5 and the points near the welds, calculate the required weld information respectively. For each straight segment weld, perform plane fitting and evaluation on the point clouds on both sides, find the plane equation and the corresponding point cloud that meet the parameter requirements, find the normalized normal vectors P1 and P2 of the two planes on both sides, find the straight line trajectory posture as P = P1 + P2, and assign P to the starting point and ending point of the line segment respectively; for the points on each circular arc segment weld, find the normalized vector Q1 pointing to the center of the circle, and the normalized vector Q2 in the positive direction of the working plane (bottom plate), then the welding torch posture at this point is Q = Q1 + Q2. The specific process is as Figure 3 shown;

[0108] S7: Traverse all welds, calculate the Euclidean distances of all weld endpoints from the origin of the working plane, and take the weld corresponding to the minimum value as the starting weld; traverse all welds. If there is a paired weld for this weld, select its corresponding weld as the next one; if not, calculate the differences Cost 总体 =a1*(X i -X i+1) + a2 * (Y i - Y i+1 ) + a3 * (Z i - Z i+1 ) ++ a4 * (W i - W i+1 ) + a5 * (P i - P i+1 ) + a6 * (R i - R i+1 ) where a1 to a6 are weight coefficients, XYZ are the three-dimensional coordinate values in space, and WPR are the axis angles representing the attitude, corresponding to the XYZ axis sequence; select the point with the lowest Cost as the next weld seam;

[0109] S8: Then generate the path of the transition points. Combine with the robot operation instructions, generate a welding operation file by associating the position, attitude, type, etc. of each weld seam operation task with the pre-set welding parameters, and send it to the robot for execution through the protocol.

[0110] The present invention automatically plans the welding sequence according to different characteristics (position / shape / size, etc.) of the weld seams, generates welding parameters, and completes the intelligent path planning of the robot at the transition positions between weld seams according to the welding sequence, making the welding task simple and efficient.

[0111] The above are only the embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, expansions, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for obtaining weld information and intelligent planning of steel structure parts, characterized in that, Including the following steps: 1) The robot control cabinet configures the welding parameter file, starts the welding robot system, and the vision sensor scans the workpiece along the moving direction of the gantry to obtain scan data; after receiving the scan data, the robot control cabinet performs preprocessing to obtain the preprocessed scan data; 2) Based on the height change and continuity in the scan data, the robot control cabinet detects the mutation points in the Z direction in the scan data, and divides the mutation points into two types of feature point sets, the upper edge and the lower edge, as the workpiece contour to distinguish the upper edge points and the edge points through the mutation points; 3) Detect the welds: According to the workpiece contour, detect different types of welds that meet the requirements in the set of mutation points, and obtain the required weld information based on the weld positions and the points near the welds corresponding to different types of welds; 4) Traverse all the weld information and plan the welding sequence; 5) Generate the weld transition point path according to the planned information of the welding sequence; 6) Combining with the robot operation instructions, send the position, posture, type of each weld operation task, the preset welding parameters, and the weld transition point path generated in step 5) to the robot through the protocol to execute the operation.

2. The method for obtaining and intelligently planning the weld seam information of a steel structure member according to claim 1, wherein The welding robot system includes: a welding robot, a gantry, a sensor, and a robot control cabinet; The robot control cabinet is used to receive the scan data detected by the vision sensor during the welding process of the welding robot, process the scan data of the welding process of the welding robot to obtain the weld transition point path, and send the motion and operation instructions according to the path to the gantry and the welding robot respectively in combination with each weld operation task and the preset welding parameters; The gantry is provided with a welding robot and is connected to the robot control cabinet, and is used to drive the welding robot to move and perform operations according to the position, posture, type of each weld operation task, the preset welding parameters, and the weld transition point path; The welding robot is used to perform welding operations on the workpiece to be processed; The vision sensor is arranged at the welding occurrence end of the welding robot and is used to collect the scan data during the welding process of the welding robot and send the scan data to the robot control cabinet for processing.

3. A method for obtaining and intelligently planning weld seam information of a steel structure member according to claim 1, characterized in that The preprocessing of the scan data obtained by the sensor scanning the workpiece by the robot control cabinet includes the following steps: 1-1) Start the sensor to scan the workpiece, obtain the camera coordinates of the workpiece in the scene, and send them to the robot control cabinet; 1-2) The robot control cabinet converts the camera coordinates of the workpiece in the camera coordinate system into the coordinates in the robot working system; 1-3) The robot control cabinet processes the scan data noise through a comprehensive filter and interpolates the actual points not scanned by the sensor by the linear interpolation method.

4. A method for obtaining and intelligently planning the weld seam information of a steel structure member according to claim 1, characterized in that, The specific content of step 3) is: 2-1) Find the inflection points in the workpiece contour, divide the contour into several segments according to the inflection points, and perform straight line detection and circular arc detection on each segment respectively to obtain the set of straight line points or circular arc point sets that meet the conditions; 2-2) For each set of linear points, find the best-fitting linear equation based on the RANSAC method, project the points on all point sets onto the line, find the two endpoints of the line, and these two points form a descriptor of a line segment; furthermore, obtain the weld information of the linear weld seam; 2-3) For each set of circular arc points, find the best-fitting spatial circle equation by the RANSAC method, determine the start and end points of the circular arc segment according to the angle between each point and the center of the circle in the plane of the spatial circle, and take the midpoint of the two points on the circular arc. These three points form a descriptor of a circular arc segment; Find the normalized vector Q1 pointing to the center of the circle. The normalized vector in the positive direction of the working plane is Q2. Then the torch posture at this point is Q = Q1 + Q2; according to step 2-3), further obtain the weld information of the linear weld seam.

5. The method for obtaining and intelligently planning the weld seam information of a steel structure part according to claim 1 or 4, characterized in that, The weld information includes: the weld information of the linear weld seam and the weld information of the circular arc weld seam; Among them, the weld information of the linear weld seam includes: the position and posture of the starting point, the position and posture of the ending point, the plane information on both sides of the line, and the orientation of the line; The weld information of the circular arc weld seam includes: the position and posture of the starting point of the circular arc, the position and posture of the midpoint of the circular arc, the position and posture of the ending point of the circular arc, and the mathematical expression of the circular arc segment.

6. The method for obtaining and intelligent planning of steel structure weld information according to claim 1, characterized in that, Traverse all the weld information and plan the welding sequence. Specifically: 3-1) Select the Euclidean distance from the weld ends of all weld information to the origin of the working plane, and take the point corresponding to the minimum value as the starting weld; 3-2) Traverse all the welds. If there is a paired weld for this weld, select its corresponding weld as the next one; if not, calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld.

7. A method for obtaining and intelligently planning weld seam information of a steel structure member according to claim 5, characterized in that In step 3-2), when it is stated that if there is no pairing, calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, and select the point with the lowest overall cost as the next weld. Specifically: (1) Calculate the differences of the six degrees of freedom between the ending point of the current weld and the starting points of the remaining welds respectively, that is: Cost 总体 = a1*(X i - X i+1 ) + a2*(Y i - Y i+1 ) + a3*(Z i - Z i+1 ) + a4*(W i - W i+1 ) + a5*(P i- -P i+1 ) + a6*(R i -R i+1 ) Among them, a1 to a6 are weight coefficients, XYZ are the coordinates of the three spatial dimensions, and WPR is the axis angle representing the posture, corresponding to the XYZ axis sequence; (2) Select Cost 总体 The point with the lowest Cost in 总体 is used as the next weld seam.

8. A method for obtaining and intelligently planning the weld information of a steel structure member according to claim 1, characterized in that In step 5), generating the weld transition point path according to the planned information of the welding sequence includes the following steps: 4-1) Create a 3D grid map: Determine the working plane and the map range according to the original point cloud. According to the upper-edge weld seam and the pairing information between weld seams, perform spatial point cloud resampling as the map point cloud, rasterize the map point cloud according to the preset resolution, and set obstacles at the positions where there are points in space; 4-2) Calculate the Euclidean distance between the ending point of the current weld and the starting point of the next weld. If the Euclidean distance is less than the set minimum value, no transition point is generated; if the Euclidean distance is greater than the maximum value, raise the torch to the safe height according to the current posture, move to above the starting point of the next weld and change the posture to be the same as that of this point, and then lower the torch vertically; 4-3) If the Euclidean distance between the current weld endpoint and the starting point of the next weld is within the set threshold range, in the grid map created in step 4-1), using the 3D-A* algorithm, with the Euclidean distance between the current position and the endpoint as the heuristic function, all path points are searched for. Then the path points are converted into point clouds, and the pose values of each point are linearly interpolated according to the number of point clouds to generate the path point poses.

9. A method for obtaining and intelligently planning weld seam information of a steel structure member according to claim 8, characterized in that, The specific content of step 4-2) is as follows: a. Calculate the Euclidean distance D between the current weld endpoint P1(X1, Y1, Z1, W1, P1, R1) and the starting point P2(X2, Y2, Z2, W2, P2, R2) of the next weld; b. Preset two threshold parameters S1 and S2, where S1 < S2, and a gun-lifting height parameter Z safe ; When D < S1, no transition points are generated; When D > S2, generate two transition points, namely (X1, Y1, Z1 + Z safe , W1, P1, R1) and (X2, Y2, Z2 + Z safe , W2, P2, R2).; Among them, the first transition point is the point above the current weld; the second transition point is the point above the starting point of the next weld; the order of the movement trajectory of the welding robot is determined as follows: the current endpoint, the first transition point, the second transition point, and the starting point of the next weld.

10. A method for obtaining and intelligent planning of steel structure weld information according to claim 8, characterized in that The specific content of step 4-3) is as follows: a) When the threshold parameter S1 < Euclidean distance D < threshold parameter S2, convert the current weld endpoint P1 and the starting point P2 of the next weld into grid map coordinates and set the corresponding positions as passable areas; b) Start searching from the P1 position in the grid space, moving one grid each time, and the search directions are the 6 directions of X+, X-, Y+, Y-, Z+, Z-; c) Take the sum of the current path length and the Euclidean distance from the current position to P2 as the search cost, and search for spatial path points in the map space; d traverses all spatial path points. If multiple path points are on the same straight line segment, delete the path points except the endpoints, and then inverse-transform the coordinates of the remaining n path points into point cloud coordinates P i (X i ,Y i ,Z i ), and perform linear interpolation on W1~W2, P1~P2, R1~R2 respectively according to the new number of path points to obtain the pose of the robot at each position; The generation path is: point set Among them,

Images