A MODEL-BASED APPROACH TO ROBOT HAND-EYE COORDINATION SYSTEM

摘要

This paper presents an implementation of PC-based Robot Hand-Eye Coordination System (RHECS) that positions a 4-Fingered Robot Hand using visual information from two cameras through a model-based approach. The RHECS tracks the robot hand's finger and visual information to define target positions. An error based on the visual distance between the finger and the target is defined. To make this error to ZERO required control information is derived such that the robot hand goes to the required object. The control information is obtained by modeling the robot hand and the camera. This information is integrated into RHECS that performs tracking and stereo control with no special-purpose hardware requirements. Experiments with the RHECS have shown that the entire visual control system runs at a rate of 20 Hz. The control software for tracking is written in C++. Some experimental results of pick and place operation are discussed.