Component Posture Determination Using Distance-Mediated Region Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition methods for determining the posture of components struggle when components are in contact or overlap, leading to inaccurate feature amount association and reduced posture determination accuracy due to insufficient surface shape information.
Innovation Solution
A component posture information acquiring device that uses a distance measure and hardware processor to determine the posture of components by identifying a region of interest based on measured distances, allowing for accurate posture determination even in cases of contact or overlap.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image recognition processing is performed using conventional image capture methods, then outer shape information can be obtained, but surface shape information is insufficient leading to increased recognition error
Solution Approach 1:
A distance measuring sensor is introduced as an intermediary device between the camera and the component. This sensor measures the distance from the camera to the component surface, providing additional depth information that complements the 2D image data. By combining image recognition with distance measurement, the system can accurately determine surface shape information even when the component is in contact with or overlaps other components, resolving the information loss problem.
2Reliability
If conventional image recognition methods are used to determine component posture, then processing can be performed, but accuracy decreases when components are in contact or overlap
Solution Approach 1:
The distance measuring sensor acts as a mediator that provides reliable distance data even in complex scenarios where components are in contact or overlap. This distance information serves as a reliable reference for associating feature amounts with specific components, enabling accurate posture determination despite the difficulty of detecting and measuring features in overlapping configurations.
Solution Approach 2:
The system segments the component detection process into two independent parts: image-based feature amount extraction and distance-based component identification. By separating these functions, the system can accurately associate features with components even when they are in contact or overlap, as the distance measure provides unambiguous component identification regardless of spatial relationships between multiple components.
3Measurement precision
If feature amounts are acquired based on outer shape information, then processing can be performed, but feature amount association with components becomes inaccurate when components are in contact
Solution Approach 1:
The distance measuring sensor provides an intermediary measurement that directly relates to component boundaries and spatial relationships. By measuring the distance from the camera to the component surface, the system can accurately identify component boundaries and associate feature amounts with the correct components, preventing the loss of boundary information that occurs with conventional image-based methods alone.
Data Source
AI summary
A component posture information acquiring device acquires information of a posture of a component from a hand that holds the component arranged on a picking table and places the component at a supply position. The component posture information acquiring device includes a distance measure that is provided in a vicinity of a component holding part of the hand and measures a distance from the distance measure to the component, and a hardware processor that determines the posture of the component, in which the hardware processor determines a position of a region of interest to be used for determining the posture of the component based on information of an arrangement position of the distance measure when the distance to the component measured by the distance measure becomes a predetermined distance, and determines the posture of the component based on a feature amount in the region of interest.