Abstract
Model-based approaches are essential to the development of robots with high adaptability to various environments. We propose the concept of a Hand-eye simulator: a simulator for a hand-eye coordinating system as a basic tool for off-line programming of visual sensors. The simulator provides useful functions to assist the hand-eye coordinating system in visual recognition and monitoring of the robot environment. Two of these functions, one for eliminating the appearance of the manipulator from the input images to reduce unnecessary image features and another for assisting cameras in occlusion avoidance to get adequate image data without occlusion are described. Solving the problems in effective use of multiple and various kinds of sensors, based on the simulators as described in this paper, will become a key issue for making intelligent hand-eye coordinating systems.
