A new adaptive smooth variable structure filter SLAM algorithm for unmanned vehicle

Abstract

This paper aims at presenting a solution to the Simultaneous Localization and Mapping (SLAM) problem of Unmanned Ground Vehicles (UGV) by combining information given by an odometer and a laser range finder. The most popular solutions to the SLAM problem are EKF-SLAM and the FAST-SLAM algorithms. The first one solution, have some important limitations which have need of an accurate process and an observation model, be affected by the linearization problem, the second is not suitable for real time implementation. Therefore, a new adaptive approach based on the Smooth Variable Structure Filter (ASVSF) is proposed to solve the UGV SLAM problem. Hence, the adaptive SVSF-SLAM algorithm is proposed with an original formulation. The main contribution of this paper is to introduce a covariance matrix to assess the estimated uncertainty of the SVSF. This new robust algorithm is validated, compared to EKF/SVSF-SLAMalgorithms and the satisfactory values of the UGV position error are obtained. Simulation results demonstrated that the proposed adaptive SVSF-SLAM algorithm is very robust face modeling uncertainties and noises and it has significantly improved the performance of the estimation process.