Hidden Markov model based learning controller

Abstract

Presents a method to learn control strategy by using a hidden Markov model (HMM), i.e., modeling a feedback controller in HMM structure. HMM is a powerful parametric model for non-stationary pattern recognition and is feasible for characterisation of a doubly stochastic process involving observable actions and a hidden decision making process. The control strategy is encoded by HMMs through a training process. The trained model is then employed to control the system. The proposed method has been investigated by simulations of a linear system and an inverted pendulum system. The HMM-based controller provides a novel way to learn control strategy and to model the human decision making process.