Robot Cell Calibration Using Virtual-Real Snapshot Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic manufacturing systems face significant positioning errors due to deviations between simulated and real-world robot cell environments, leading to processing errors, which are exacerbated by non-perfect assembly and wear of components over time.
Innovation Solution
A system and method that uses cameras and a robot controller to compare virtual and real snapshots of robot and object positions, adjusting the virtual robot cell calibration to match the real environment, thereby reducing position and orientation errors through continuous refinement during manufacturing runtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If 3D simulation models are used to design and program robotic cells, then the robotic system can be programmed and tested in a virtual environment, but geometric errors between the simulation models and real world positions cause about 90% of total robot cell positioning errors
Solution Approach 1:
The system performs preliminary calibration by capturing images of calibration objects at multiple known poses before actual manufacturing operations. These images are used to pre-determine transformation matrices that map between camera coordinates and robot cell coordinates, establishing an accurate reference framework in advance to eliminate positioning errors during production
Solution Approach 2:
The patent replaces manual mechanical measurement and calibration methods with automated computer vision-based calibration. Cameras capture images of calibration objects, and software automatically computes transformation matrices through image processing and coordinate system transformations, substituting precise mechanical measurement with optical measurement systems
2Productivity
If robot operations are performed without continuous calibration, then productivity is maintained, but position errors accumulate due to robot movement, component wear, and environmental changes
Solution Approach 1:
The calibration system is designed to be continuously operational during manufacturing runtime. The robot periodically captures images of calibration objects at various poses throughout production, and the controller continuously updates transformation matrices based on these new measurements, maintaining accurate calibration without interrupting manufacturing operations
Solution Approach 2:
The system implements feedback by continuously monitoring the positions of calibration objects during manufacturing operations. The controller compares actual captured images with expected positions, detects deviations, and automatically adjusts transformation matrices to compensate for drift caused by robot movement, component wear, and environmental changes
3Measurement precision
If multiple calibration objects at various poses are captured during manufacturing runtime, then calibration accuracy is improved, but the complexity of the calibration process increases
Solution Approach 1:
The calibration object is designed with multiple identifiable features and can be positioned in various poses while serving the same calibration function. The same calibration object structure is used throughout manufacturing runtime for multiple calibration measurements, eliminating the need for different calibration tools for different poses and simplifying the overall system
Solution Approach 2:
The calibration system uses the existing robot and camera infrastructure already present in the manufacturing cell to perform calibration operations. The robot manipulates the calibration object using its own end effector, and the camera that is already part of the manufacturing system captures calibration images, making the calibration process self-contained and eliminating additional complex equipment
Data Source
Figure 1~2
Figure 3~5
Figure 6
AI summary
A calibrating system and method for robot cells including cameras, robots, objects, object movement mechanisms, and robot controllers. The robot controller includes software which generates a virtual robot cell which defines a virtual three dimensional space in which virtual cameras, virtual robots, and virtual objects are positioned and manipulated. The robot controller takes a plurality of virtual snapshots of the virtual robots, in various poses, and virtual objects from different virtual camera angles and positions. The robotic controller also controls the real cameras to take a plurality of real snapshots of the robots and objects in the same robot poses, camera positions as the virtual snapshots. The robot controller compares virtual snapshots to matching real snapshots and determines the closest matching virtual and real snapshots. The detected differences are used to adjust the calibration which will reduce the robot error rate based upon position and orientation errors.