A RAPIDLY CONVERGING RECURSIVE METHOD FOR MOBILE ROBOT LOCALIZATION

This paper proposes a simple method for estimating the position of a r obot from relatively few sensor readings. Our algorithms are intended for applications where sensor readings are expensive or otherwise limi ted, and the readings that are taken are subject to considerable error s or noise. This method exhibits faster convergence with fewer measure ments and greater accuracy than that exhibited by the discrete Kalman filter in this type of application. Our approach is validated with a m obile robot, on which a camera is used to obtain bearing information w ith respect to landmarks in the environment.