A multi-scale navigation method for intelligent wheelchair based on hierarchical POMDPs and DBNS

This paper describes a multi-scale navigation method for intelligent wheelchair based on hierarchical Partially Observable Decision Process (POMDPs) and dynamic Bayesian networks (DBNs) in indoor environment. The intelligent wheelchair and the model of POMDPs are introduced. A hierarchical POMDPs model is given, which decomposes the decision making process into a set of small ones that can be done more efficiently. We explore the advantages of representing hierarchical POMDPs as DBNs. The multi-scale navigation method based on hierarchical POMDPS and DBNs is proposed. The method is implemented and tested onboard a smart wheelchair in the indoor navigation task. The experimental results show the reliability and efficiency of the navigation method.