Assembly Success Determination Using 3D Shape Model Residual Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the success of robotic assembly tasks face challenges in accurately assessing assembly accuracy without appropriately setting a target area, due to variations in part dimensions, shapes, and observation viewpoints.
Innovation Solution
An information processing apparatus that obtains images and three-dimensional shape models of assembled components, calculates position and orientation, and evaluates assembly success using attributes added to specific areas of the models, determining success based on residual errors without requiring a predefined target area.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a target area is set to determine assembly success, then the determination can be performed, but the accuracy is affected by part dimensions, shapes, and observation viewpoints
Solution Approach 1:
The patent extracts the determination from a specific target area and applies it to the entire assembled component. Instead of selecting a particular region to check for assembly success, the system evaluates the complete component by comparing its overall position and orientation against the 3D shape model, thereby eliminating the complexity of target area selection while maintaining determination capability
Solution Approach 2:
The patent makes the determination method universal by applying it to the entire assembled component regardless of part dimensions, shapes, or observation viewpoints. The 3D shape model serves as a universal reference that can evaluate any assembled configuration, making the system adaptable to various component geometries and viewing angles without requiring separate target area definitions
2Manufacturing precision
If the residual error of attachment parts is calculated, then assembly accuracy can be evaluated, but the evaluation becomes unreliable without appropriate target area selection
Solution Approach 1:
The patent merges the evaluation of attachment parts into a comprehensive evaluation of the entire assembled component. By combining the position and orientation data of all parts and comparing them against the complete 3D shape model, the system achieves reliable assembly accuracy evaluation that is not dependent on selecting specific attachment regions, thereby ensuring consistency across different part configurations
3Measurement precision
If a three-dimensional shape model is used to calculate position and orientation, then assembly can be evaluated, but the residual error distribution varies with part dimensions and observation viewpoint
Solution Approach 1:
The patent transitions from two-dimensional image analysis to three-dimensional shape model evaluation. By utilizing the full 3D information from the shape model, the system can accurately calculate position and orientation while accounting for variations in part dimensions and observation viewpoints, as the 3D model provides a comprehensive reference framework that captures the complete geometric information of the assembled component
Data Source
AI summary
An apparatus comprises an unit configured to obtain an image of an assembled object that is constituted by first and second objects that have been assembled; an unit configured to obtain a three-dimensional shape model of the assembled object that has at least one area to which an attribute that corresponds to the first object or the second object is added; an unit configured to obtain a position and orientation of the assembled object based on the image; an unit configured to obtain, from the three-dimensional shape model of the position and orientation, first and second evaluation values that are for evaluating a state of assembly in areas that correspond to the first and second objects; and an unit configured to determine whether or not the assembly was successful based on the first and second evaluation values.