A novel landmark tree based self-localization and path-planning method for an intelligent wheelchair

In this paper a novel approach of self-localization for an intelligent wheelchair is presented. Different from previous approaches we adopt a two-stage matching strategy to position control of an intelligent wheelchair, which is necessary if the current location is not known due to re-initialisation or an accident. Based on the landmark tree developed here, we combine a method similar to image-retrieval and a Bayes law based matching algorithm to locate the wheelchair in multiple resolutions, which is related to the requirement of task-completion. A new path-planning algorithm is introduced, which is originated from our landmark tree and is easy to use.