Title: Experimental comparison of Bayesian positioning methods based on multi-sensor data fusion

Authors: Dominique Gruyer; Alain Lambert; Mathias Perrollaz; Denis Gingras

Addresses: IFSTTAR IM, LIVIC, F-78000 Versailles, France ' IFSTTAR IM, LIVIC, F-78000 Versailles, France ' INRIA Grenoble Rhone-Alpes, 38334 Saint Ismier, France ' Université de Sherbrooke, 2500 Boulevard de l'Université, Sherbrooke, QC J1K 2R1, Canada

Abstract: Localising a vehicle consists in estimating its position state by merging data from proprioceptive sensors (inertial measurement unit, gyrometer, odometer, etc.) and exteroceptive sensors (GPS sensor). A well-known solution in state estimation is provided by the Kalman filter. However, due to the presence of nonlinearities, the Kalman estimator is applicable only through some recursive variants among which the Extended Kalman filter (EKF), the Unscented Kalman Filter (UKF) and the Divided Differences of first and second order (DDI and DD2). We have compared these filters using the same experimental data. The results obtained aim to rank these approaches by their performances in terms of accuracy and consistency.

Keywords: robot localisation; mobile robots; Kalman filter; EKF; UKF; DDI; DD2; sensor fusion; vehicle positioning; Bayesian positioning; accuracy; consistency.

DOI: 10.1504/IJVAS.2014.057852

International Journal of Vehicle Autonomous Systems, 2014 Vol.12 No.1, pp.24 - 43

Received: 19 Dec 2011
Accepted: 09 Oct 2012

Published online: 09 Oct 2013 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article