Active Visual Feedback Control of Robot Manipulator
Wei-Yun Yau and Han Wang
Nanyang Technological University, Research Lab IV, School of Electrical & Electronic Engineering, Block S1, Nanyang Avenue, Singapore 639798
This paper describes an approach to control the robot manipulator using an active stereo camera system as the feedback mechanism. In the conventional system, increasing the precision of the hand-eye system inevitably reduces its operating range. It is also not robust to perturbations of the vision system which is commonly encountered in real-world applications. The proposed hand-eye system addresses these limitations and shortcomings. In this paper, the concept of pseudo image space which has three dimension is introduced. A relationship between the pseudo image space and the robot space is established via a mapping matrix. Using the mapping matrix together with visual feedback to update it, algorithms are developed to control the robot manipulator. No camera calibration to recover the homogeneous transformation matrix of the stereo vision system is required. Thus, the hand-eye system is robust to changes in the camera orientation. A method to cater for focal length changes in the vision system is also described. This allow the hand-eye system to select the required resolution and accuracy suitable for its task. The proposed method has been tested using actual implementation to verify its robustness and accuracy.