Simultaneous localisation and mapping of a mobile robot via interlaced extended Kalman filter

摘要

A crucial task for automatic explorations is the ability for a robot to real-time estimate its position in an unknown environment. To this end, the robot is required to simultaneously localise itself and to build a map of the surroundings (Simultaneous Localisation and Mapping (SLAM) problem). This problem represents an interesting test-bed for non-linear estimator techniques. In this paper we propose to illustrate a solution based on the Extended Kalman Filter (EKF) approach, able to considerably reduce the computational burden and memory occupancy requirements, both of them representing two of the main drawbacks for this class of solutions. Specifically, we adopt the Interlaced Extended Kalman Filter (IEKF) formulation where the whole estimation problem is decomposed into a number of semi-autonomous subproblems. To partially compensate the decoupling errors introduced, process and measurements covariance matrices are suitably augmented. Two different implementations are analysed and compared with traditional EKF-based approaches. Experimental results emphasise that, even if the IEKF formulations suffer for a slight degraded estimation, they dramatically reduce computational burden. In this way, IEKF solutions to SLAM problems appear to be a good trade-off between accuracy and computational requirements, making it suitable for real time implementations.