Feature extraction from infrared scan data based on SM-MS algorithm for autonomous mobile robot

This paper presents a new feature extraction system that is to be used with a 2D infrared (IR) range finder. It consists of three data processing modules: cluster, segment and fit. Taking the accuracy and real time of feature extraction into account, a Most Similar (MS) algorithm is designed, combining the Split and Merge (SM) algorithm, a new data segment method-SM-MS algorithm is proposed, which can detect line, arc and corner accurately and quickly. In order to verify the concept, a computer simulation and an experiment prototype were built. The results proved that the new feature extraction method can provide high performance for autonomous mobile robot (AMR) in an indoor dynamic environment.