HIDDEN MARKOV MODEL APPROACH TO SKILL LEARNING AND ITS APPLICATION TOTELEROBOTICS

In this paper, we discuss the problem of how human skill can be repres ented as a parametric model using a hidden Markov model (HMM), and how an HMM-based skill model can be used to learn human skill. HMM is fea sible to characterize a doubly stochastic process-measurable action an d immersible mental states-that is involved in the skill learning. We formulated the learning problem as a multidimensional HMM and develope d a testbed for a variety of skill learning applications. Based on ''t he most likely performance'' criterion, the best action sequence can b e selected from all previously measured action data by modeling the sk ill as an HMM. The proposed method has been implemented in the teleope ration control of a space station robot system, and some important imp lementation issues have been discussed. The method allows a robot to l earn human skill in certain tasks and to improve motion performance.