On the tips of one's toes: self-localization in a dynamic environment

This paper deals with the self-localization of an autonomous mobile robot within a dynamic environment. The concept of dynamic environments is extended along the vertical direction: this paper assumes that above our heads exists a free zone where also highly dynamic environments become almost "static". The localization process, if managed into the free zone, does not suffer from feature occlusions and other related issues. The feature matching process is carried out using a typical split and merge algorithm which supplies an extended Kalman filter with observations used to perform position tracking.