The application of adaptive extended Kalman filter in mobile robot localization

Abstract

An adaptive extended Kalman filter (AEKF) algorithm is proposed to resolve the problem of the error accumulation in the process of mobile robot localization. We take the Taylor series of sampling time in AEKF and use the Sage-Husa time-varying noise estimator to estimate observation noise in real time. Meanwhile, the convergence and the complexity of operation of AEKF are analyzed and the experiments show that AEKF has a good comprehensive performance in terms of speed and precision. Finally, two kinds of robot localization algorithm are analyzed and the error is compared with the experiment, that shows AEKF has a better performance.