Vision-based Control of Multi-fingered Robot Hands using Neural Networks

Abstract

Most research so far on of multi-fingered robot control has assumed that the kinematics is known exactly. However, in many applications of multi-fingered robot hands, the kinematics is uncertain. In this paper, a vision based control problem for multi-fingered robot hands with uncertain kinematics, dynamics and camera model is addressed. Adaptive neural network control law is proposed and it is shown that the stability can be achieved in the presence of the uncertainties. Sufficient conditions for choosing the feedback gains are presented to guarantee the stability.