A two-dimensional vision-based finite element partitioning full-automatic polishing method

Abstract

This invention discloses a fully automated finite element partitioning polishing method based on two-dimensional vision, comprising the following steps: S1: Collecting all defect information acquired by the robot, organizing and analyzing it, and pushing the defect information to the visualization module to select defect points on the 3D model of the vehicle and obtain path coordinates; S2: Decomposing the regions of different parts of the vehicle using finite element optimization; S3: Setting pose transformation points between adjacent regions to avoid collisions; S4: When a defect point appears in a certain region, it is classified and assigned, and the corresponding coordinates and defect type are sent to the polishing robot. The advantages of this invention compared to existing technologies are: existing technologies cannot achieve complete linkage of actions between different regions, while the polishing method of this invention better solves the current situation of inability to achieve linkage and complete automation in polishing, and also handles the collision problem in the continuous processing of real-time random defect points.

Description

Technical Field

[0001] This invention relates to a polishing method, and more particularly to a fully automated finite element partitioning polishing method based on two-dimensional vision, belonging to the field of intelligent robots. Background Technology

[0002] In the existing technology of automatic automotive polishing, Chinese Patent Publication No. CN 108284388 B, entitled "A Vision-Guided Intelligent Force-Controlled Polishing Device," discloses a vision-guided intelligent force-controlled polishing device. This device includes a polishing vision sensor, a moving unit, a force sensor, a polishing head, and a control unit. The moving unit, connected to the force sensor, has two degrees of freedom and is used to move the end-effector polishing head. The force sensor detects the contact force between the end-effector polishing head and the workpiece. The vision sensors are symmetrically arranged on both sides of the moving unit to monitor the three-dimensional structure and spatial position of the workpiece, the position of the polishing head, and the surface processing quality of the workpiece. The control unit is connected to the moving unit, the force sensor, and the vision sensor. Its advantage is that it can perform three-dimensional structure recognition and control of the polishing area. Its disadvantage is that it is considered a single step in the overall processing solution and does not adequately address issues such as robotic arm path and obstacle avoidance, thus failing to perform intelligent zoning planning and fully automated polishing.

[0003] Chinese Patent Publication No. CN 116021366 A, entitled "A Method for Planning Grinding Paths on the External Surface of a Spatial Convex Polyhedron," discloses a method and storage medium for planning grinding paths on the external surface of a spatial convex polyhedron. This method is based on an acquired set of working planes, and for each working plane, it directly obtains the set of surfaces to be ground. The method includes: S1, determining and sorting the working planes to be ground; S2, sorting multiple surfaces to be ground within a single working plane; S3, establishing a path point queue; and S4, calculating the local optimal solution based on the computation time. This invention's method for planning grinding paths on the external surface of a spatial convex polyhedron is based on an acquired set of spatial coordinates of three types of grinding elements and known relationships between topological nodes, performing spatial three-dimensional external surface path planning. Its advantages are: point sorting and preliminary planning are performed. Its disadvantages are: it does not analyze the pose and entry / exit points of the grinding area, resulting in defects in the overall processing, thus preventing intelligent partitioning planning and fully automated grinding.

[0004] Chinese Patent Publication No. CN 113759907 A, entitled "A Method, Apparatus, Device, and Storage Medium for Regional Full-Coverage Path Planning," discloses a method, apparatus, device, and storage medium for regional full-coverage path planning, applied to intelligent tracking devices. It includes: constructing a grid cost map based on the environmental and positioning information of the intelligent tracking device; converting the grid cost map into a regular pixel map; dividing the regular pixel map into multiple sub-regions; determining a first path within each sub-region and a second path between the sub-regions; and determining the optimal path for the intelligent tracking device based on the first and second paths. Its advantage is that it performs block processing. Its disadvantage is that it does not provide detailed planning for actual numerical processing, and therefore cannot achieve fully automated polishing.

[0005] Therefore, developing a fully automated finite element partitioning grinding method based on two-dimensional vision to solve the problem that existing technologies cannot perform intelligent partitioning planning for fully automated grinding has become an urgent problem for those skilled in the art. Summary of the Invention

[0006] To address the aforementioned shortcomings, this invention provides a fully automated finite element partitioning grinding method based on two-dimensional vision.

[0007] The above-mentioned objective of this invention is achieved through the following technical solution: a fully automated finite element partitioning polishing method based on two-dimensional vision, comprising the following steps:

[0008] S1: Collect all defect information obtained by the robot, organize and analyze it, push the defect information to the visualization module, select defect points on the 3D model of the car, and obtain the path coordinates;

[0009] S2: Decompose the regions of different components of the vehicle using finite element optimization;

[0010] S3: Setting pose transition points between two adjacent regions can avoid collisions;

[0011] S4: When a defect appears in a certain area, classify and assign it, and send the corresponding coordinates and defect type to the polishing robot.

[0012] In step S1, the acquisition robot, carrying an acquisition camera, collects and organizes the defect information of the workpiece to be processed, processes the defect information, and pushes it to the visualization module. The visualization module has its own database. Through prior training, it accumulates data, confirms the defect type through comparison, displays and outputs the data through communication, and obtains the two-dimensional spatial coordinates through the acquisition system. Then, it obtains the robot arm pose through digital model reconstruction and pose transformation matrix.

[0013] In step S2, by acquiring and statistically calculating the pose data of a large-area robotic arm, the range of multi-curvature pose points is divided, and the same pose and similar poses are sorted and merged. Using the finite element method with curvature as a constraint, the grinding points are divided into similar regions, and the same region type, similar pose type, and overlapping point type are managed in a unified way for tool entry and exit posture.

[0014] In step S3, a one-to-many transition posture training simulation is performed on the regions after finite element division. The transition postures between different regions are compared, optimized, and stored. When the defect points of two different posture regions are obtained, the transition point is selected to achieve a better transition.

[0015] In step S4, when the acquisition system obtains the defect information, it can directly retrieve the corresponding region pose and transition pose coordinates from the database, output them as numerical variables, and transmit them to the polishing robot through the communication protocol. The polishing robot will select polishing parameters (pressure, flexible feedback time, rotation speed, moving speed, polishing action) according to the defect type.

[0016] The advantages of this invention compared to existing technologies are: existing technologies cannot achieve complete linkage of actions in various areas, while the polishing method of this invention better solves the problem of non-linkage and fully automated polishing currently available on the market, and also addresses the collision problem in the continuous processing of random defects in real time. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of the present invention.

[0018] Figure 2 This is a schematic diagram of the polishing scenario of the present invention.

[0019] Figure 3 yes Figure 2 The left view.

[0020] Figure 4 yes Figure 2 Top view. Detailed Implementation

[0021] The invention will now be described in further detail with reference to the accompanying drawings.

[0022] like Figures 1 to 4 As shown, a fully automated finite element partitioning grinding method based on two-dimensional vision includes the following steps:

[0023] S1: The acquisition robot 1 carries an acquisition camera to collect and organize the defect information of the workpiece to be processed, processes and analyzes the defect information, and pushes it to the visualization module. The visualization module has its own database. Through prior training, it accumulates data, confirms the defect type through comparison, and displays and outputs the data through the communication module. It selects the defect points on the 3D model of the car, obtains the 2D spatial coordinates through the acquisition system, and then obtains the robot arm pose through digital model reconstruction and pose transformation matrix.

[0024] S2: The regions of different parts of the vehicle are optimized and decomposed by finite element method. By acquiring and statistically calculating the pose data of the large-area robotic arm, the range of multi-curvature pose points is divided. The same pose and similar poses are sorted and merged. Using the finite element method with curvature as the constraint condition, the grinding points are divided into similar regions. The same region type, similar pose type, and overlapping point type are managed in a unified way for tool entry and exit posture.

[0025] S3: Setting pose transition points between two adjacent regions can avoid collisions. A one-to-many transition posture training simulation is performed on the regions after finite element analysis. The transition postures between different regions are compared, optimized, and stored. When flaw points are obtained in two different posture regions, transition points are selected for better transition.

[0026] S4: When a defect appears in a certain area, it is classified and assigned, and the corresponding coordinates and defect type are sent to the polishing robot 2. That is, when the acquisition system obtains the defect information, it can directly retrieve the corresponding area pose and transition pose coordinates from the database, output them as numerical variables, and transmit them to the polishing robot 2 through the communication protocol. The polishing robot will select polishing parameters (pressure, flexible feedback time, rotation speed, moving speed, polishing action) according to the defect type.

[0027] This polishing method involves two modules: machine hardware driver and data processing. The hardware includes: microcontroller, driver board, robot, polishing head, constant force polishing device, acquisition camera, and industrial control computer for data acquisition and processing. The software includes: image acquisition section, camera calibration module, image processing module, and path planning module.

[0028] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A fully automated finite element partitioning grinding method based on two-dimensional vision, characterized in that: Includes the following steps: S1: Collect all defect information obtained by the robot, organize and analyze it, push the defect information to the visualization module, select defect points on the 3D model of the car, and obtain the path coordinates; S2: Decompose the regions of different components of the vehicle using finite element optimization; S3: Setting pose transition points between two adjacent regions can avoid collisions; S4: When a defect appears in a certain area, classify and assign it, and send the corresponding coordinates and defect type to the polishing robot; In step S2, by acquiring and statistically calculating the pose data of the large-area robotic arm, the range of multi-curvature pose points is divided, and the same pose and similar poses are sorted and merged. Using the finite element method, with curvature as the constraint condition, the grinding points are divided into similar regions, and the same region type, similar pose type, and overlapping point type are managed in a unified way for tool entry and exit posture. In step S3, a one-to-many transition posture training simulation is performed on the regions after finite element division. The transition postures between different regions are compared, optimized, and stored. When the defect points of two different posture regions are obtained, the transition point is selected to achieve a better transition.

2. The full-automatic polishing method based on two-dimensional vision and finite element partitioning according to claim 1, characterized in that: In step S1, the acquisition robot, carrying an acquisition camera, collects and organizes the defect information of the workpiece to be processed, processes the defect information, and pushes it to the visualization module. The visualization module has its own database. Through prior training, it accumulates data, confirms the defect type through comparison, displays and outputs the data through communication, and obtains the two-dimensional spatial coordinates through the acquisition system. Then, it obtains the robot arm pose through digital model reconstruction and pose transformation matrix.

3. The fully automated finite element partitioning grinding method based on two-dimensional vision according to claim 1, characterized in that: In step S4, when the acquisition system obtains the defect information, it can directly retrieve the corresponding region pose and transition pose coordinates from the database, output them as numerical variables, and transmit them to the polishing robot through the communication protocol. The polishing robot will select polishing parameters according to the defect type.

Images