A groove cutting trajectory optimization method, system, device and medium

Abstract

The present application relates to the technical field of industrial automation cutting, and discloses a groove cutting track optimization method, system, device and medium. The method comprises the following steps: acquiring an initial global point cloud of a target workpiece by using a structured light 3D camera, and generating a groove cutting track of the target workpiece through the initial global point cloud; acquiring a first local point cloud of the initial global point cloud except for a special area, acquiring a second local point cloud of the special area by using a line laser camera, projecting the first local point cloud and the second local point cloud to a two-dimensional plane where the target workpiece is located after converting the first local point cloud and the second local point cloud to a robot base coordinate system, and obtaining corresponding two-dimensional point sets; registering the two-dimensional point sets of the two, obtaining a two-dimensional transformation matrix, expanding the two-dimensional transformation matrix into a three-dimensional transformation matrix, fusing the second local point cloud and the first local point cloud, and generating a target global point cloud; and regenerating a cutting track of the special area according to the target global point cloud, and optimizing the groove cutting track through the cutting track of the special area regenerated.

Description

Technical Field

[0001] This invention relates to the field of industrial automated cutting technology, and in particular to a method, system, equipment and medium for optimizing bevel cutting trajectories. Background Technology

[0002] In existing automated beveling processes, it is typically necessary to use 3D vision sensors to acquire 3D point cloud data of the workpiece surface, and then use algorithms to calculate and generate the motion trajectory of the cutting robot. Currently, there are two main solutions in industry for acquiring 3D point clouds of workpieces: structured light 3D camera imaging and line laser camera scanning.

[0003] Among them, the advantage of structured light 3D cameras lies in their ability to acquire overall workpiece information in a single photograph. Due to their large field of view and fast data acquisition speed, they can meet high overall cutting cycle times. However, when cutting small-angle bevels and double-sided bevels, the measurement deviations of structured light 3D cameras are amplified and superimposed, ultimately making it difficult to meet high-standard processing requirements for trajectory accuracy. Furthermore, because structured light cameras have a large field of view but relatively limited resolution, when processing small feature areas or reflective areas on the workpiece, the acquired point cloud is prone to distortion or complete loss (e.g., ...). Figure 1 The diagram shows how square corners are identified as rounded corners. Figure 2 The point cloud shown is missing.

[0004] Compared to structured light solutions, line laser cameras can obtain more precise and detailed workpiece point clouds through line-by-line scanning. They can completely reconstruct small-angle bevels, double-sided bevels, and various minute features, thereby calculating higher-precision cutting trajectories. However, line laser solutions require scanning the entire area of ​​the workpiece one by one. For bevel-cut workpieces of a certain size, the full-scale scanning process is very time-consuming, severely slowing down the overall operation cycle and failing to meet the stringent efficiency requirements of modern industrial mass production.

[0005] In summary, existing visual measurement technologies for bevel cutting suffer from the problem of balancing efficiency and accuracy. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, device and medium for optimizing bevel cutting trajectory, which can solve the problem of difficulty in balancing cutting efficiency and accuracy.

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a method for optimizing bevel cutting trajectories, comprising the following steps: The initial global point cloud of the target workpiece is acquired using a structured light 3D camera; According to the preset bevel cutting algorithm, the bevel cutting trajectory of the target workpiece is generated through the initial global point cloud; the bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, which is the small angle bevel, double-sided bevel, small feature area or reflective area; The first local point cloud, excluding special regions, is obtained from the initial global point cloud. The second local point cloud of the special region is obtained through a line laser camera. The first and second local point clouds are transformed from the corresponding camera coordinate system to the robot base coordinate system and then projected onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set. The two-dimensional point sets of the first local point cloud and the second local point cloud are registered to obtain the two-dimensional transformation matrix of the two. The two-dimensional registration transformation matrix is ​​then extended into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud. The cutting trajectory of a special region is regenerated based on the target global point cloud, and the bevel cutting trajectory is optimized using the regenerated cutting trajectory of the special region.

[0008] Further, the step of transforming the first local point cloud and the second local point cloud from their respective camera coordinate systems to the robot base coordinate system, and then projecting them onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set includes: Obtain the normal vectors of the first and second local point clouds in the robot's base coordinate system, respectively; Based on the normal vectors of the first local point cloud and the second local point cloud, determine the rotation matrix of the first local point cloud and the second local point cloud around the Z-axis perpendicular to the ground to rotate into a plane parallel to the X-axis and Y-axis. The first and second local point clouds are transformed according to the corresponding rotation matrix to obtain the projected point clouds of the first and second local point clouds in the Z-axis direction, which serve as the corresponding two-dimensional point sets.

[0009] Further, the registration of the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain their two-dimensional transformation matrix includes: An iterative nearest-point registration algorithm is adopted, with the two-dimensional point set of the first local point cloud as the source point set and the two-dimensional point set of the second local point cloud as the target point set. The translation amount in the X-axis direction, the translation amount in the Y-axis direction, and the rotation angle that minimize the average point-to-point distance between the source point set and the target point set after translation and rotation transformation are obtained to obtain the two-dimensional transformation matrix.

[0010] Furthermore, the step of extending the two-dimensional transformation matrix into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud includes: Fill the X-axis translation, Y-axis translation, and rotation angle from the two-dimensional transformation matrix into the corresponding positions in the preset three-dimensional transformation matrix, and set the Z-axis translation to 0 to obtain the expanded three-dimensional transformation matrix. After spatial transformation of the second local point cloud using the extended 3D transformation matrix, it is fused with the first local point cloud to generate the target global point cloud.

[0011] Furthermore, the first local point cloud and the second local point cloud are transformed from the corresponding camera coordinate system to the robot base coordinate system through the hand-eye matrix of the structured light 3D camera and the line laser camera, respectively.

[0012] Furthermore, the first local point cloud and the second local point cloud are transformed from the corresponding camera coordinate system to the robot base coordinate system using the following formula: ; In the formula, In the eye-in-hand scheme, this is the transformation matrix from the corresponding camera to the robot flange, which serves as the hand-eye matrix. This is the transformation matrix from the robot flange to the robot base coordinates when the camera takes a picture, which is obtained from the robot coordinate transformation during the picture taking.

[0013] Furthermore, before regenerating the cutting trajectory of the specific region based on the target global point cloud, the method further includes: Statistical filtering or voxel filtering is used to preprocess the global point cloud of the target.

[0014] Embodiments of the present invention also provide a bevel cutting trajectory optimization system, comprising the following modules: The global point cloud acquisition module is used to acquire the initial global point cloud of the target workpiece using a structured light 3D camera; The cutting trajectory generation module is used to generate the bevel cutting trajectory of the target workpiece from the initial global point cloud according to the preset bevel cutting algorithm. The bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, which is the small angle bevel, double-sided bevel, small feature area or reflective area. The local point cloud projection module is used to acquire the first local point cloud in the initial global point cloud except for special regions, and to acquire the second local point cloud in the special regions through a line laser camera. After transforming the first local point cloud and the second local point cloud from the corresponding camera coordinate system to the robot base coordinate system, they are projected onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set. The global point cloud optimization module is used to register the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain the two-dimensional transformation matrix of the two. The two-dimensional registration transformation matrix is ​​then extended into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud. The cutting trajectory optimization module is used to regenerate the cutting trajectory of a special region based on the target global point cloud, and optimize the bevel cutting trajectory based on the regenerated cutting trajectory of the special region.

[0015] Embodiments of the present invention also provide a computer device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described bevel cutting trajectory optimization method.

[0016] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described bevel cutting trajectory optimization method.

[0017] The bevel cutting trajectory optimization method provided by this invention has at least the following beneficial effects: For special areas during beveling, such as small-angle bevels, double-sided bevels, small feature areas, or reflective areas, a line laser camera is used to scan and obtain point clouds. For other regular areas, a structured light 3D camera is used to scan and obtain point clouds. By using a line laser camera, high-precision scanning of point clouds in special areas is ensured while avoiding the redundant time consumption of full measurement by the line laser camera. Furthermore, when fusing the structured light 3D point cloud and the line laser point cloud, a 3D-to-2D dimensionality reduction and registration strategy is adopted. First, dimensionality reduction through projection simplifies the registration calculation, and then precise registration ensures the fusion accuracy. This further guarantees a dual improvement in the efficiency and accuracy of subsequent generation of cutting trajectories based on the fused point cloud. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0019] Figure 1 A schematic diagram of point cloud distortion provided by the present invention; Figure 2 A schematic diagram of point cloud missing data provided by the present invention; Figure 3 A flowchart illustrating a bevel cutting trajectory optimization method provided by the present invention. Figure 1 ; Figure 4 A flowchart illustrating a bevel cutting trajectory optimization method provided by the present invention. Figure 2 ; Figure 5 A schematic diagram of a point cloud provided by the present invention using a structured light 3D camera; Figure 6 A schematic diagram of a cropped structured light 3D camera point cloud provided by the present invention; Figure 7 A schematic diagram of a line laser scanning point cloud provided by the present invention; Figure 8 A schematic diagram of a registered and fused point cloud provided by the present invention; Figure 9 A schematic diagram of a final cutting trajectory provided by the present invention; Figure 10 This is a schematic diagram of a beveling machine system provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] This invention first calculates the cutting trajectory using a global structured light camera. When it is determined that there are special features such as double-sided bevels, small-angle bevels, or small features, the scanning trajectory of the corresponding edge is output simultaneously. During the execution process, only double-sided bevels and small-angle bevels are scanned. The point cloud is fused through the camera calibration relationship, and then a registration algorithm is used for registration. The fused point cloud is then used to recalculate the cutting trajectory, thereby obtaining a high-precision cutting trajectory.

[0022] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] One embodiment of the present invention relates to a method for optimizing bevel cutting trajectories. The flow of the bevel cutting trajectory optimization method in this embodiment can be as follows: Figure 3 As shown, it includes: Step 301: Use a structured light 3D camera to acquire the initial global point cloud of the target workpiece.

[0024] Step 302: Based on the preset bevel cutting algorithm, generate the bevel cutting trajectory of the target workpiece through the initial global point cloud; wherein, the bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, the special area being the small angle bevel, the double-sided bevel, the small feature area or the reflective area.

[0025] Step 303: Obtain the first local point cloud excluding special regions from the initial global point cloud, and obtain the second local point cloud of the special region through a line laser camera. After transforming the first local point cloud and the second local point cloud from the corresponding camera coordinate system to the robot base coordinate system, project them onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set.

[0026] Step 304: Register the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain their two-dimensional transformation matrices, and extend the two-dimensional registration transformation matrix into a three-dimensional transformation matrix so as to fuse the second local point cloud and the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud.

[0027] Step 305: Regenerate the cutting trajectory of the special region based on the target global point cloud, and optimize the bevel cutting trajectory using the regenerated cutting trajectory of the special region.

[0028] The following is based on Figure 4 The execution steps shown provide a detailed explanation of the implementation details of the bevel cutting trajectory optimization method in this embodiment. The following content is only for the convenience of understanding the implementation details and is not necessary for implementing this solution.

[0029] 1. Global point cloud acquisition and coordinate system transformation.

[0030] The workpiece is photographed using a structured light 3D camera to obtain a point cloud in the camera coordinate system. As the initial global point cloud of the target workpiece, it can be like... Figure 5 As shown. Then, the point cloud is transformed into the robot's base coordinate system using a pre-calibrated hand-eye matrix (either eye-in-hand or eye-to-hand is acceptable).

[0031] This embodiment takes eye-in-hand as an example to obtain the point cloud of the robot's base coordinate system. ;in, The hand-eye matrix, in the eye-in-hand scheme, is the transformation matrix from the camera to the robot flange. The transformation matrix from robot flange to robot base coordinates during camera photography can be obtained from the robot coordinate transformation during photography.

[0032] 2. Workpiece identification and feature judgment.

[0033] Based on the existing beveling cutting algorithm, workpiece recognition and beveling cutting trajectory are performed. During the beveling cutting trajectory process, it is determined whether small features are included, whether it is a double-sided beveling, or whether it is a small-angle beveling. If none of these are included, the cutting trajectory is directly output. The cutting process is performed, and the specific cutting trajectory calculation method can be adopted using existing bevel cutting algorithms, which will not be elaborated here.

[0034] If contained, output the corresponding edge scan trajectory. As the cutting trajectory for a special region, the remaining edges are output as cutting trajectories. This serves as the cutting trajectory for a regular region. This embodiment describes a processing solution for the presence of the aforementioned special feature regions.

[0035] 3. Preprocessing of raw point clouds.

[0036] Based on the scanning trajectory, The point cloud is cropped according to the scan trajectory, and the point cloud near the scan trajectory is deleted to obtain the result. This allows us to obtain the first local point cloud, excluding special regions, from the initial global point cloud. In this embodiment, deleting the original point cloud before subsequent fusion with the line laser point cloud avoids the line laser point cloud being affected by the original point cloud, thus preventing a decrease in accuracy. Specifically, the point cloud cropped by the structured light 3D camera can be obtained as follows: Figure 6 As shown.

[0037] 4. Line laser local scanning and point cloud generation.

[0038] robot along Perform a fine line laser scan, and combine it with a line laser hand-eye matrix to transform the scanned point cloud to the robot's base coordinate system, obtaining... Refer to point 1 for the transformation method, thus obtaining the first and second local point clouds in the robot's base coordinate system. The line laser scanning point cloud can be obtained as follows: Figure 7 As shown.

[0039] 5. Point cloud fusion and registration.

[0040] For applications involving planar bevel-cut workpieces, to address the issues of insufficient point cloud fusion accuracy caused by hand-eye matrix deviation and low efficiency of conventional 3D registration, a 2D projection dimensionality reduction registration strategy is introduced on top of the basic Iterative Closest Point (ICP) registration framework. This achieves efficient and high-precision point cloud fusion and registration. Specific technical details are as follows: 5.1 Registration core logic.

[0041] Since the bevel-cut workpieces targeted in this embodiment are all planar parts, their core cutting features (such as bevel edges, small-angle bevels, double-sided bevels, etc.) are mainly concentrated in the plane where the workpiece is located, and the dimensional changes in the Z-axis direction are minimal. Based on this characteristic, this embodiment adopts the core logic of "3D point cloud → 2D projection → 2D registration → 3D mapping". First, the registration calculation is simplified by reducing the dimension through projection, and then the fusion accuracy is ensured by precise alignment, ultimately achieving a dual improvement in efficiency and accuracy.

[0042] Specifically, and The two types of 3D point clouds have been transformed to the same robot base coordinate system through corresponding hand-eye matrices, exhibiting good initial pose consistency and allowing for direct superposition for preliminary fusion. However, due to slight deviations in the hand-eye matrices, the preliminarily fused point clouds still have alignment errors, failing to meet sub-millimeter cutting accuracy requirements. Therefore, this embodiment first projects the two types of 3D point clouds onto the workpiece plane, converting them into 2D point sets. 2D registration is then used to quickly correct the hand-eye matrix deviations. Finally, the corrected 2D registration results are mapped back to 3D space, completing the precise fusion of the two types of point clouds and obtaining a unified and accurate fused point cloud. .

[0043] 5.2 Complete registration process.

[0044] Based on the structural characteristics of planar components, the point cloud fusion and registration process in this embodiment is divided into the following four steps, which are fully adapted to the real-time requirements of industrial sites, require no manual intervention, and can be seamlessly integrated into the overall cutting process: 5.2.1 Projecting 3D point clouds onto a 2D plane.

[0045] Calculate the first local point cloud and the second local point cloud in the robot's base coordinate system respectively. and normal vector and Based on the normal vectors of the first and second local point clouds, determine the rotation matrix that rotates the first and second local point clouds around the Z-axis perpendicular to the ground into a plane parallel to the X and Y axes. and Using this matrix and By performing the transformation, we can obtain the projected point clouds of the two point clouds along the Z-axis. and Extract the X and Y coordinates of each point to form the corresponding 2D point set (denoted as Seg_2D and Scan_2D respectively).

[0046] 5.2.2 2D ICP fine registration (quickly corrects hand-eye deviation).

[0047] because and The two sets of points have been transformed to the same robot base coordinate system through the hand-eye matrix and have undergone the same projection transformation. Their corresponding 2D point sets Seg_2D and Scan_2D have good initial pose consistency with an initial deviation of less than 1mm. Therefore, the 2D ICP registration algorithm can be directly used for fine registration to minimize the point-to-point Euclidean distance between the two sets and achieve accurate alignment.

[0048] In the 2D ICP registration process, Scan_2D (the 2D point set corresponding to the local point cloud of the line laser) is used as the target point set, and Seg_2D (the 2D point set corresponding to the local point cloud of the structured light) is used as the source point set. Only three registration parameters are optimized: translation in the X-axis direction, translation in the Y-axis direction, and rotation angle. Through iterative optimization, the average point-to-point distance between the source point set and the target point set after translation and rotation transformation is minimized (optimal threshold ≤ 0.1mm), and finally the optimal 2D registration transformation matrix is ​​obtained.

[0049] Compared to traditional 3D ICP registration (which requires optimization of 6 parameters: 3 translation parameters + 3 rotation parameters), the computational load of 2D ICP registration in this step is significantly reduced, and the registration speed is increased by more than 50%.

[0050] 5.2.3 Mapping 2D registration results back to 3D space.

[0051] After obtaining the optimal 2D registration transformation matrix, it is expanded into a 3D registration transformation matrix, ensuring no translation or rotation along the Z-axis, retaining only the translation and rotation angles along the X and Y axes, thus achieving accurate mapping of the 2D registration result to 3D space. The 3D registration transformation matrix is ​​based on the identity matrix. The rotation matrix and translation vector from the 2D registration transformation matrix are filled into the corresponding X and Y axis positions of the 3D transformation matrix, with the Z-axis translation set to 0. This matrix is ​​then used to... Transform it to make it the same as Align and overlay to obtain a point cloud. .

[0052] 5.2.4 3D point cloud fusion and post-processing.

[0053] After 3D registration transformation pass By transforming the inverse matrix, the final target point cloud can be obtained. After fusion, if the point cloud contains a small number of duplicate points or minor noise, the quality can be further improved by removing duplicate points, ensuring the accuracy of subsequent cutting trajectory calculations. This post-processing step can be flexibly selected and enabled based on the actual point cloud quality without affecting the overall registration efficiency. The fused point cloud can then be displayed as follows: Figure 8 As shown.

[0054] 6. Point cloud post-processing.

[0055] If noise exists after fusion and registration, it can be processed using statistical filtering, voxel filtering, or other methods.

[0056] 7. Calculation and merging of special cutting trajectories.

[0057] use Calculate the cutting trajectory for cases including double-sided bevels, small angles, and small features. By merging the conventional cutting trajectory with the special cutting trajectory, the final complete cutting trajectory is obtained: = + ; Ultimately, the robot performed the merged... This allows for beveling of the workpiece. The final cutting trajectory can be as follows: Figure 9 As shown, the left figure represents the trajectory of the line laser fusion scheme, and the right figure represents the trajectory of the structured light point cloud scheme. The line laser fusion scheme exhibits better accuracy due to its smaller point spacing and more stable point cloud quality. Ideally, the accuracy of the structured light point cloud scheme is typically 1-1.5 mm, while the line laser scheme can achieve an accuracy of 0.5 mm. Furthermore, this embodiment also provides the mechanical system for achieving the beveling of the workpiece, such as... Figure 10 As shown, the configuration includes a KUKA robot, a line laser camera, a structured light 3D camera, and a Haibao plasma system.

[0058] This invention adopts a hierarchical architecture of structured light global coarse measurement + line laser local fine measurement. It quickly completes workpiece positioning and feature recognition through structured light; introduces feature judgment logic to accurately trigger line laser local scanning to achieve "on-demand measurement"; and is compatible with two hand-eye calibration modes, flexibly adapting to different robot and camera installation layouts, reducing deployment difficulty.

[0059] In the specific optimization process, the global point cloud is first cropped according to the scanning trajectory to remove interference points near the scanning path, laying the foundation for fusion; then, ICP registration is used to eliminate hand-eye matrix deviation and achieve high-precision alignment; finally, a combination of statistical filtering and voxel filtering is used to remove noise and redundancy, improve the quality of the point cloud, and ensure the stability of trajectory calculation.

[0060] Meanwhile, the calculation of trajectories in regular areas and special feature areas is decoupled to avoid error propagation; the two types of trajectories are seamlessly stitched together through an adaptive algorithm, balancing efficiency and accuracy; and line laser scanning is embedded into the process to build a "measurement-correction-cutting" closed loop, further improving cutting accuracy.

[0061] In addition, this invention achieves full automation from point cloud acquisition, feature recognition, trajectory decision-making, scanning measurement to cutting execution without human intervention; the core algorithm can be reused in existing systems, requiring only the addition of a line laser module and feature judgment logic, and supports interface with mainstream industrial robots, 3D cameras, and line laser sensors, reducing equipment modification costs and improving engineering adaptability.

[0062] The above-mentioned hierarchical sensing, which combines global coarse measurement with local fine measurement using structured light and local laser measurement, along with a feature-driven adaptive scanning strategy, has the following significant advantages: (1) For fine structures such as small angles, double-sided bevels, and small features, local line laser scanning improves the measurement accuracy to the sub-millimeter level, solving the problem of insufficient global point cloud accuracy of traditional structured light; (2) Line laser scanning is performed only on areas with special features, avoiding the redundant time consumption of full measurement and significantly improving work efficiency while ensuring accuracy; (3) By optimizing point cloud cropping, fusion and registration, hand-eye matrix deviation and noise interference are eliminated, and the robustness and cutting stability of the system under complex working conditions are improved. (4) The automated closed-loop process and strong compatibility design reduce the cost of engineering deployment and maintenance, and facilitate its application in bevel cutting scenarios in multiple industries.

[0063] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of this invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, without changing the core design of the algorithm and process, are also within the protection scope of this invention.

[0064] Another embodiment of the present invention relates to a bevel cutting trajectory optimization system. The implementation details of this bevel cutting trajectory optimization system are described below. The following content is for ease of understanding and is not essential for implementing this solution. The bevel cutting trajectory optimization system of this embodiment includes: The global point cloud acquisition module is used to acquire the initial global point cloud of the target workpiece using a structured light 3D camera; The cutting trajectory generation module is used to generate the bevel cutting trajectory of the target workpiece from the initial global point cloud according to the preset bevel cutting algorithm. The bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, which is the small angle bevel, double-sided bevel, small feature area or reflective area. The local point cloud projection module is used to acquire the first local point cloud in the initial global point cloud except for special regions, and to acquire the second local point cloud in the special regions through a line laser camera. After transforming the first local point cloud and the second local point cloud from the corresponding camera coordinate system to the robot base coordinate system, they are projected onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set. The global point cloud optimization module is used to register the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain the two-dimensional transformation matrix of the two. The two-dimensional registration transformation matrix is ​​then extended into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud. The cutting trajectory optimization module is used to regenerate the cutting trajectory of a special region based on the target global point cloud, and optimize the bevel cutting trajectory based on the regenerated cutting trajectory of the special region.

[0065] It is not difficult to see that this embodiment is a system embodiment corresponding to the above method embodiments, and this embodiment can be implemented in conjunction with the above method embodiments. The relevant technical details and technical effects mentioned in the above embodiments are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0066] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0067] Another embodiment of the present invention relates to a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the bevel cutting trajectory optimization method of the above embodiments.

[0068] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0069] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0070] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0071] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0072] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for optimizing bevel cutting trajectory, characterized in that, The method includes: The initial global point cloud of the target workpiece is acquired using a structured light 3D camera; According to the preset bevel cutting algorithm, the bevel cutting trajectory of the target workpiece is generated through the initial global point cloud; the bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, which is the small angle bevel, double-sided bevel, small feature area or reflective area; The first local point cloud, excluding special regions, is obtained from the initial global point cloud. The second local point cloud of the special region is obtained through a line laser camera. The first and second local point clouds are transformed from the corresponding camera coordinate system to the robot base coordinate system and then projected onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set. The two-dimensional point sets of the first local point cloud and the second local point cloud are registered to obtain the two-dimensional transformation matrix of the two. The two-dimensional registration transformation matrix is ​​then extended into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud. The cutting trajectory of a special region is regenerated based on the target global point cloud, and the bevel cutting trajectory is optimized using the regenerated cutting trajectory of the special region.

2. The bevel cutting trajectory optimization method according to claim 1, characterized in that, The process of transforming the first and second local point clouds from their respective camera coordinate systems to the robot base coordinate system and then projecting them onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set includes: Obtain the normal vectors of the first and second local point clouds in the robot's base coordinate system, respectively; Based on the normal vectors of the first local point cloud and the second local point cloud, determine the rotation matrix of the first local point cloud and the second local point cloud around the Z-axis perpendicular to the ground to rotate into a plane parallel to the X-axis and Y-axis. The first and second local point clouds are transformed according to the corresponding rotation matrix to obtain the projected point clouds of the first and second local point clouds in the Z-axis direction, which serve as the corresponding two-dimensional point sets.

3. The bevel cutting trajectory optimization method according to claim 2, characterized in that, The registration of the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain their two-dimensional transformation matrices includes: An iterative nearest-point registration algorithm is adopted, with the two-dimensional point set of the first local point cloud as the source point set and the two-dimensional point set of the second local point cloud as the target point set. The translation amount in the X-axis direction, the translation amount in the Y-axis direction, and the rotation angle that minimize the average point-to-point distance between the source point set and the target point set after translation and rotation transformation are obtained to obtain the two-dimensional transformation matrix.

4. The bevel cutting trajectory optimization method according to claim 3, characterized in that, The step of extending the two-dimensional transformation matrix into a three-dimensional transformation matrix, and fusing the second local point cloud with the first local point cloud using the three-dimensional transformation matrix to generate the target global point cloud, includes: Fill the X-axis translation, Y-axis translation, and rotation angle from the two-dimensional transformation matrix into the corresponding positions in the preset three-dimensional transformation matrix, and set the Z-axis translation to 0 to obtain the expanded three-dimensional transformation matrix. After spatial transformation of the second local point cloud using the extended 3D transformation matrix, it is fused with the first local point cloud to generate the target global point cloud.

5. The bevel cutting trajectory optimization method according to claim 1, characterized in that, The first local point cloud and the second local point cloud are respectively transformed from the corresponding camera coordinate system to the robot base coordinate system through the hand-eye matrix of the structured light 3D camera and the line laser camera.

6. The bevel cutting trajectory optimization method according to claim 5, characterized in that, The first local point cloud and the second local point cloud are transformed from the corresponding camera coordinate system to the robot base coordinate system using the following formula: ； In the formula, In the eye-in-hand scheme, this is the transformation matrix from the corresponding camera to the robot flange, which serves as the hand-eye matrix. This is the transformation matrix from the robot flange to the robot base coordinates when the camera takes a picture, which is obtained from the robot coordinate transformation during the picture taking.

7. The bevel cutting trajectory optimization method according to claim 1, characterized in that, Before regenerating the cutting trajectory of the specific region based on the target global point cloud, the method further includes: Statistical filtering or voxel filtering is used to preprocess the global point cloud of the target.

8. A bevel cutting trajectory optimization system, characterized in that, The system includes: The global point cloud acquisition module is used to acquire the initial global point cloud of the target workpiece using a structured light 3D camera; The cutting trajectory generation module is used to generate the bevel cutting trajectory of the target workpiece from the initial global point cloud according to the preset bevel cutting algorithm. The bevel cutting trajectory includes the cutting trajectory of the regular area and the cutting trajectory of the special area, which is the small angle bevel, double-sided bevel, small feature area or reflective area. The local point cloud projection module is used to acquire the first local point cloud in the initial global point cloud except for special regions, and to acquire the second local point cloud in the special regions through a line laser camera. After transforming the first local point cloud and the second local point cloud from the corresponding camera coordinate system to the robot base coordinate system, they are projected onto the two-dimensional plane where the target workpiece is located to obtain the corresponding two-dimensional point set. The global point cloud optimization module is used to register the two-dimensional point sets of the first local point cloud and the second local point cloud to obtain the two-dimensional transformation matrix of the two. The two-dimensional registration transformation matrix is ​​then extended into a three-dimensional transformation matrix to fuse the second local point cloud with the first local point cloud through the three-dimensional transformation matrix to generate the target global point cloud. The cutting trajectory optimization module is used to regenerate the cutting trajectory of a special region based on the target global point cloud, and optimize the bevel cutting trajectory based on the regenerated cutting trajectory of the special region.

9. A computer device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the bevel cutting trajectory optimization method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the bevel cutting trajectory optimization method as described in any one of claims 1 to 7.

Images