Transformation of Human Hand Positions for Robotic Hand Control

Control of multifingered robotic hands requires complex trajectories, defining simultaneous motion of a large number of joints. To teach a multifingered hand manipulation strategies, traditional point to point training is tedious and fails to embody the underlying purpose and constraints driving motion of the hand. In this paper we describe a method for training robotic hands using the human hand as a multidegree of freedom teaching device. The algorithm is based on a functional analysis of the human hand and results in an algebraic transformation from human hand positions to corresponding positions in the target domain.