Robot Vision Alignment Using Learned Target Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robot systems using visual feedback require preparation of a target mark or feature point on a workpiece, necessitating heavy operator workload and complex detection algorithms.
Innovation Solution
A machine learning device with a state observation unit, determination data retrieval unit, and learning unit that calculates movement amounts to align a robot arm end portion with a target image, eliminating the need for pre-defined marks or algorithms by learning the relative positional relationship between a vision sensor and workpiece.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If visual feedback with target marks or feature points is used, then setup time is reduced, but operator workload increases due to mark preparation, detection algorithm configuration, and stable detection know-how
Solution Approach 1:
The robot system performs self-learning by autonomously capturing images at multiple positions, extracting feature points automatically, and generating detection algorithms without human intervention. The learning unit processes images and determines optimal detection parameters independently, eliminating the need for operators to prepare marks or configure detection settings.
Solution Approach 2:
The system performs preliminary image capture and feature point extraction during the learning phase before actual operation. By pre-acquiring images at multiple positions and pre-determining detection algorithms, the system prepares all necessary detection data in advance, enabling rapid setup without operator intervention during production.
2Measurement precision
If target marks or feature points are prepared in advance on workpieces, then detection accuracy is improved, but device complexity increases due to mark preparation processes and detection algorithm requirements
Solution Approach 1:
The system extracts feature points directly from standard workpiece images without requiring pre-prepared target marks. The feature point extraction unit automatically identifies and extracts meaningful features from normal workpiece surfaces, eliminating the need for special mark preparation while maintaining detection accuracy.
Solution Approach 2:
The learning unit creates a digital model by capturing and processing images of the workpiece at multiple positions. This digital representation is stored and used for detection, replacing the need for physical target marks and complex detection algorithms with a simplified image-based reference model.
3Reliability
If conventional visual feedback systems are used, then detection stability is achieved through know-how, but adaptability decreases when dealing with different workpiece types or positions
Solution Approach 1:
The system dynamically adapts to different workpiece types and positions by capturing images at multiple predetermined positions and using the learning unit to determine optimal detection algorithms for each configuration. The detection system can be reconfigured through simple image capture rather than requiring operator know-how adjustment, enabling flexible adaptation to various workpiece variations.
Data Source
AI summary
A machine learning device includes a state observation unit for observing, as state variables, an image of a workpiece captured by a vision sensor, and a movement amount of an arm end portion from an arbitrary position, the movement amount being calculated so as to bring the image close to a target image; a determination data retrieval unit for retrieving the target image as determination data; and a learning unit for learning the movement amount to move the arm end portion or the workpiece from the arbitrary position to a target position. The target position is a position in which the vision sensor and the workpiece have a predetermined relative positional relationship. The target image is an image of the workpiece captured by the vision sensor when the arm end portion or the workpiece is disposed in the target position.