Welding path planning method of automobile welding production line, computer program product and system

Abstract

The invention provides a welding path planning method for an automobile welding production line, a computer program product and a system. The method comprises the following steps: detecting actual position data of a welding seam in real time; aligning and registering the actual position data of the welding seam with a preset welding seam model, and calculating the offset of the welding seam; finely adjusting the pose of the welding part in real time according to the welding seam offset; judging whether a preset welding path has an obstacle conflict or not according to the fine-adjusted welding part pose, and if the conflict exists, re-planning a collision-free path which does not collide with the obstacle according to the welding seam position, the fine-adjusted welding part pose and the position of the obstacle; in this way, the control platform can control the welding part to conduct precise displacement welding according to the collision-free path, and the welding seam cannot deviate.

Description

Technical Field

[0001] The present invention relates to the field of welding technology, and particularly relates to a welding path planning method, a computer program product and a system for an automobile welding production line. Background Art

[0002] With the continuous improvement of the welding quality requirements in the automobile manufacturing industry, the application of automated welding technology in the welding production line is becoming increasingly widespread. However, in the actual production process, factors such as workpiece clamping and positioning deviation and welding thermal deformation will cause the actual weld position to deviate from the theoretical weld position, affecting the welding quality. In some high-precision welding occasions, the weld position deviation may lead to defects such as insufficient weld strength and porosity, thereby affecting the structural safety and service life of the whole vehicle. Summary of the Invention

[0003] The technical problem to be solved by the present invention is how to avoid weld deviation during the automobile welding process.

[0004] To solve the above technical problem, the present invention provides a welding path planning method for an automobile welding production line, including the following steps: S1. Real-time detect the actual position data of the weld; S2. Align and register the actual position data of the weld with a preset weld model, and calculate the weld deviation amount; S3. Real-time fine-tune the pose of the welding component according to the weld deviation amount; S4. Judge whether there is an obstacle conflict in the preset welding path according to the fine-tuned pose of the welding component. If there is a conflict, re-plan a collision-free path that does not collide with the obstacle according to the weld position, the fine-tuned pose of the welding component, and the position of the obstacle.

[0005] Further, in the step S1, a vision module is used to obtain an image containing the weld, and then the weld is located based on a deep learning-based weld segmentation model or an edge detection algorithm to generate actual point cloud data containing the actual position of the weld as the actual position data of the weld.

[0006] Further, in the step S2, the actual point cloud data is aligned and registered with the preset weld model, and a rigid body transformation matrix including a rotation offset amount and a translation offset amount is calculated by the least squares method to obtain the weld deviation amount.

[0007] Further, in the step S2, the alignment and registration of the actual point cloud data with the preset weld model includes the following steps S21, S22, S23, S24: S21. Nearest point matching: For each point p in the actual point cloud data i , find the nearest point q in the preset weld model point cloud j , and the specific formula is as follows: ; S22. Calculate the rigid body transformation matrix: Calculate the rigid body transformation matrix including the rotation offset R and the translation offset t by the least squares method, and then minimize the error. The specific formula is as follows: ; Solve for the rotation offset R and the translation offset t using SVD decomposition: ; ; Get , ; S23. Apply the transformation: Apply the rigid body transformation matrix to the actual point cloud to obtain the aligned point cloud ; S24. Iterative optimization: Repeat steps S21~S23 until the error converges.

[0008] Furthermore, in step S3, the pose of the welding component is fine-tuned in real time according to the weld offset using a closed-loop PID control algorithm or a model predictive control algorithm.

[0009] Furthermore, in step S4, to determine whether there is an obstacle conflict in the welding path specifically: Use a bounding box algorithm, a safety box detection algorithm, or a distance field algorithm to calculate the distance between the welding component and the obstacle in real time when it moves along the welding path. If the distance is less than the preset safety distance, it is determined that there is a conflict. If the distance is not less than the preset safety distance, it is determined that there is no conflict. The specific formula is as follows: ; Where d is the distance between the welding component and the obstacle when it moves along the welding path, b is the position of the end of the welding component, o is the position of the obstacle, and O is the set of obstacles.

[0010] Furthermore, in step S4, use the rapidly-exploring random tree algorithm or the artificial potential field method to re-plan a collision-free path that does not collide with obstacles, which specifically includes the following steps S41, S42, S43, S44, S45: S41. Random sampling: Randomly sample a sampling point; S42. Find the nearest node: Find the node closest to the sampling point; S43. Expand a new node: Expand a new node from the nearest node towards the sampling point to ensure that the path is collision-free; S44. Check the end point: If the new node is close to the target position of the weld component, the path planning is completed; S45. Iterative optimization: Repeat steps S41~S44 until a collision-free path is found.

[0011] The present invention also provides a computer program product, including a computer program which, when executed by a processor, implements the steps in the above-mentioned method.

[0012] The present invention also provides a welding path planning system for an automobile welding production line, including a control platform, an industrial robot, and a welding component. The control platform is loaded with a robot operating system. The industrial robot has a robotic arm. The welding component is a welding gun installed at the end of the robotic arm of the industrial robot. The control platform is electrically connected to the industrial robot and the welding component respectively. The control platform includes a memory and a processor connected to each other, and the above-mentioned computer program product is stored in the memory.

[0013] Further, it includes a vision module electrically connected to the control platform, and the vision module is a 3D line laser scanner or a binocular vision system installed at the end of the robotic arm of the industrial robot.

[0014] The present invention has the following beneficial effects: If there are obstacle conflicts in the preset welding path, a collision-free path that does not collide with obstacles is re-planned according to the weld position, the fine-tuned pose of the welding component, and the position where the obstacle is located. In this way, the control platform can control the welding component to perform precise displacement welding according to the collision-free path, and the weld will not shift. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic flow chart of a welding path planning method for an automobile welding production line. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The following further elaborates on the present invention in detail in conjunction with specific embodiments.

[0017] This embodiment provides a welding path planning system for an automobile welding production line. The system includes a control platform, an industrial robot, a welding component, and a vision module. The control platform is electrically connected to the industrial robot, the welding component, and the vision module respectively. Among them, the control platform is loaded with a Robot Operating System (ROS). The industrial robot has a robotic arm. The welding component is a welding gun installed at the end of the robotic arm of the industrial robot. The vision module is a 3D line laser scanner or a binocular vision system installed at the end of the robotic arm of the industrial robot, which combines a high-frame-rate industrial camera to achieve sub-millimeter-level precision detection. The control platform includes a memory and a processor connected to each other, and a computer program product is stored in the memory. The computer program product includes a computer program which, when executed by the processor, implements the welding path planning method for an automobile welding production line as Figure 1 shown, specifically including the following steps S1, S2, S3, and S4.

[0018] S1. Real-time detect the actual position data of the weld seam.

[0019] The control platform uses a vision module to obtain an image containing the weld seam, and then locates the weld seam based on a deep learning weld seam segmentation model (U-Net) or an edge detection algorithm to generate actual point cloud data containing the actual position of the weld seam as the actual position data of the weld seam. Among them, the edge detection algorithm uses the combination of the Canny edge detection algorithm and the Hough transform. First, the Canny algorithm is used to detect the edge of the weld seam in the image, and then the Hough transform is applied to the edge image of the weld seam to detect a straight line and locate the weld seam.

[0020] S2. Align and register the actual position data of the weld seam with a preset weld seam model, and calculate the weld seam offset.

[0021] After detecting the actual position data of the weld seam, the control platform aligns and registers the actual position data of the weld seam with a preset weld seam model, and calculates the weld seam offset, that is, aligns and registers the actual point cloud data with the preset weld seam model point cloud. Specifically, a rigid body transformation matrix including a rotational offset and a translational offset is calculated by the least squares method (ICP algorithm) to obtain the weld seam offset. Among them, the actual point cloud data is P={p1,p2,…,pn}, and the preset weld seam model point cloud is Q={q1,q2,…,qm}. Aligning and registering the actual point cloud data with the preset weld seam model point cloud includes the following steps S21, S22, S23, S24: S21. Nearest point matching: For each point p in the actual point cloud data i , find the nearest point q j in the preset weld seam model point cloud. The specific formula is as follows: .

[0022] S22. Calculate the rigid body transformation matrix: Calculate a rigid body transformation matrix including a rotational offset R and a translational offset t by the least squares method to minimize the error. The specific formula is as follows: ; Use SVD decomposition to solve for the rotational offset R and the translational offset t: ; ; Get , .

[0023] S23. Apply the transformation: Apply the rigid body transformation matrix to the actual point cloud to obtain the aligned point cloud .

[0024] S24. Iterative optimization: Repeat steps S21 - S23 until the error converges (e.g., the error is less than a preset threshold or the maximum number of iterations is reached).

[0025] S3. Fine - tune the pose of the welding component in real - time according to the weld offset.

[0026] After obtaining the weld offset, the control platform uses a closed - loop PID control algorithm or a model predictive control algorithm (MPC) to fine - tune the pose of the welding component in real - time according to the weld offset. Among them, the closed - loop PID control algorithm is a feedback control strategy that combines three links: proportional (P), integral (I), and derivative (D). By adjusting the system output in real - time, the controlled quantity can accurately track the set value. Model Predictive Control (MPC) is an advanced control strategy based on a process model and is widely used in industrial process control, robotics, autonomous driving, and other fields.

[0027] S4. Determine whether there is an obstacle conflict in the preset welding path according to the fine - tuned pose of the welding component. If there is a conflict, re - plan a collision - free path that does not collide with the obstacle according to the weld position, the fine - tuned pose of the welding component, and the position of the obstacle.

[0028] After fine - tuning the pose of the welding component, the control platform obtains the preset welding path, and then determines whether there is an obstacle conflict in the preset welding path according to the fine - tuned pose of the welding component. Specifically, the control platform uses an Axis - Aligned Bounding Box algorithm (AABB), an Oriented Bounding Box algorithm (OBB), or a Distance Field algorithm to calculate the distance d between the welding component and the obstacle in real - time when the welding component moves along the welding path. If the distance d is less than the preset safety distance, it is determined that there is a conflict. If the distance d is not less than the preset safety distance, it is determined that there is no conflict. The specific formula is as follows: ; where d is the distance between the welding component and the obstacle when the welding component moves along the welding path, b is the position of the end of the welding component, o is the position of the obstacle, and O is the set of obstacles.

[0029] If there is a conflict, the control platform re - plans a collision - free path that does not collide with the obstacle according to the weld position, the fine - tuned pose of the welding component, and the position of the obstacle, using a Rapidly - exploring Random Tree algorithm (RRT) or an Artificial Potential Field method. Specifically, it includes the following steps S41, S42, S43, S44, S45: Step S41. Random sampling: Randomly sample a sampling point qrand in the configuration space; Step S42. Find the nearest node: Find the node qnear in the tree that is closest to the sampling point qrand; Step S43. Expand a new node: Expand a new node qnew from the nearest node qnear in the direction of the sampling point qrand, ensuring that the path is collision-free; Step S44. Check the end point: If the new node qnew is close to the target position qgoal of the weld component, the path planning is completed; Step S45. Iterative optimization: Repeat steps S41 - S44 until a collision-free path is found.

[0030] After re-planning a collision-free path, the control platform can control the welding component to perform precise displacement welding according to the collision-free path, and the weld will not shift.

[0031] As described above, it is only the implementation mode of the present invention, and does not limit the scope of patent protection. Those skilled in the art make non-substantive changes or substitutions based on the present invention, and still fall within the scope of patent protection.

Claims

1. A welding path planning method for an automobile welding production line, characterized in that, It includes the following steps: S1. Real-time detect the actual position data of the weld seam; S2. Align and register the actual position data of the weld seam with a preset weld seam model, and calculate the weld seam offset; S3. Fine-tune the pose of the welding component in real time according to the weld seam offset; S4. Judge whether there is an obstacle conflict in the preset welding path according to the fine-tuned pose of the welding component. If there is a conflict, re-plan a collision-free path that does not collide with the obstacle according to the weld seam position, the fine-tuned pose of the welding component, and the position where the obstacle is located.

2. The welding path planning method for an automobile welding production line according to claim 1, characterized in that, In step S1, a vision module is used to obtain an image containing the weld seam, and then a deep learning-based weld seam segmentation model or edge detection algorithm is used to locate the weld seam, and actual point cloud data containing the actual position of the weld seam is generated as the actual position data of the weld seam.

3. The welding path planning method for an automotive welding production line according to claim 2, characterized in that, In step S2, the actual point cloud data is aligned and registered with the preset weld seam model, and a rigid body transformation matrix including a rotation offset and a translation offset is calculated by the least squares method to obtain the weld seam offset.

4. The welding path planning method for an automobile welding production line according to claim 3, characterized in that, In step S2, aligning and registering the actual point cloud data with the preset weld seam model includes the following steps S21, S22, S23, S24: S21. Nearest point matching: For each point p in the actual point cloud data i , find the nearest point q in the preset weld model point cloud j , and the specific formula is as follows: ； S22. Calculate the rigid body transformation matrix: Calculate the rigid body transformation matrix including the rotation offset R and the translation offset t by the least squares method, and then minimize the error. The specific formula is as follows: ； Use SVD decomposition to solve for the rotation offset R and the translation offset t: ； ； Obtain , ; S23. Apply transformation: Apply the rigid body transformation matrix to the actual point cloud to obtain the aligned point cloud ; S24. Iterative optimization: Repeat steps S21 to S23 until the error converges.

5. The welding path planning method for an automobile welding production line according to claim 1, characterized in that, In step S3, according to the weld seam offset, a closed-loop PID control algorithm or a model predictive control algorithm is used to fine-tune the pose of the welding component in real time.

6. The welding path planning method for an automotive welding production line according to claim 1, characterized in that, In step S4, specifically judging whether there is an obstacle conflict in the welding path: Use the bounding box algorithm, the safety box detection algorithm or the distance field algorithm to calculate the distance between the welding component and the obstacle in real time when the welding component moves along the welding path. If the distance is less than the preset safety distance, it is judged that there is a conflict. If the distance is not less than the preset safety distance, it is judged that there is no conflict. The specific formula is as follows: ； Where d is the distance between the welding component and the obstacle when the welding component moves along the welding path, b is the position of the end of the welding component, o is the position of the obstacle, and O is the set of obstacles.

7. The welding path planning method for an automotive welding production line according to claim 1, characterized in that, In step S4, the rapidly-exploring random tree algorithm or the artificial potential field method is used to re-plan a collision-free path that does not collide with the obstacle, specifically including the following steps S41, S42, S43, S44, S45: S41. Random sampling: Randomly sample a sampling point; S42. Find the nearest node: Find the node closest to the sampling point; S43. Expand a new node: Expand a new node from the nearest node towards the sampling point to ensure that the path is collision-free; S44. Check the end point: If the new node is close to the target position of the weld seam component, the path planning is completed; S45. Iterative optimization: Repeat steps S41 to S44 until a collision-free path is found.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps in the method described in any one of claims 1 to 7.

9. A welding path planning system for an automobile welding production line, characterized in that It includes a control platform, an industrial robot, and a welding component. The control platform is loaded with a robot operating system. The industrial robot has a robotic arm. The welding component is a welding torch installed at the end of the robotic arm of the industrial robot. The control platform is electrically connected to the industrial robot and the welding component respectively. The control platform includes a memory and a processor connected to each other. The computer program product as recited in claim 7 is stored in the memory.

10. The welding path planning system according to claim 9, characterized in that, It includes a vision module electrically connected to the control platform. The vision module is a 3D line laser scanner or a binocular vision system installed at the end of the robotic arm of the industrial robot.

Images