Robot Position Control Using Image-Based Relative Workpiece Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot device controllers face accuracy issues in positioning robots due to measurement errors in calculating the Jacobian matrix, particularly when large movements are involved, leading to low precision in aligning workpieces.
Innovation Solution
A robot device controller that uses a visual sensor to capture images of workpieces, processes these images to detect characteristic amounts, calculates relative position differences, and generates movement commands to adjust the robot's position accurately, eliminating the need for precise calibration of the visual sensor coordinate system and Jacobian matrix calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the Jacobian matrix method is used to calculate robot movement based on characteristic portion position differences, then the robot can be accurately driven in the vicinity of each characteristic portion, but the accuracy decreases as the robot moves away from the characteristic portion, resulting in low accuracy when large movement amounts are involved
Solution Approach 1:
The system captures images before and after robot movement, detects characteristic portions in both images, calculates the position difference, and uses this feedback to correct the robot's actual movement amount. This closed-loop feedback mechanism compensates for the Jacobian matrix accuracy limitations when large movements occur, maintaining high positioning accuracy regardless of movement magnitude.
Solution Approach 2:
The patent replaces the purely mathematical Jacobian matrix calculation with an image-based visual measurement system. By capturing actual images of characteristic portions and calculating positions from image coordinates, the system substitutes theoretical mechanical models with empirical visual measurement, achieving higher accuracy in determining robot movement.
2Measurement precision
If visual sensor coordinate system calibration is performed to detect three-dimensional position of workpieces, then position detection accuracy is improved, but the calibration process increases device complexity and requires precise alignment between visual sensor and robot coordinate systems
Solution Approach 1:
The system uses characteristic portions with distinctive patterns (such as markers or coded symbols) that can be easily identified and copied in images. These characteristic portions serve as visual references that can be detected without complex coordinate transformations, simplifying the measurement process while maintaining accuracy.
Solution Approach 2:
The patent changes the measurement parameter from three-dimensional spatial coordinates requiring calibration to two-dimensional image plane coordinates that can be directly measured. By working with image coordinates and characteristic portion positions in the image plane, the system avoids the complex visual sensor to robot coordinate system calibration while still achieving accurate position detection.
Data Source
AI summary
A first characteristic portion of a first workpiece and a second characteristic portion of a second workpiece are previously determined. A characteristic amount detection unit detects a first characteristic amount related to the position of the first characteristic portion and a second characteristic amount related to the position of the second characteristic portion in an image captured by a camera. A calculation unit calculates, as a relative position amount, the difference between the first characteristic amount and the second characteristic amount. A command generation unit generates a movement command for operating a robot based on a relative position amount in the image captured by the camera and a relative position amount in a predetermined reference image.