Relative self-localization base on fish-eye lenses and SIFT algorithm for indoor mobile robot

Abstract

In this paper, we consider the problem of mobile robot indoor position estimation using only visual information from a single camera. A camera which has a fish-eye lens is mounted on the top of the mobile robot and pointed to the ceiling. At the beginning of the visual positioning, we assume that we know the initial orientation and position of the mobile robot. Through the key point extraction, non-ceiling key point removal, key point calibration, ellipsoid construction and so on; the robot position and orientation can be determined after a short time moving. © 2011 Springer-Verlag.