Autonomous Navigation of an Outdoor Mobile Robot in a Cluttered Environment

This paper proposes a modification of HybridSLAM strategy that is used to navigate an outdoor autonomous mobile robot in a cluttered environment. HybridSLAM is combining extended Kalman filter SLAM EKF-SLAM and FastSLAM to take advantage of strengths and to cover shortcomings of both filters. By the use of an unscented version of Kalman filter instead of EKF-SLAM, the formulation of the HybridSLAM is revised. Same as HybridSLAM, the new revised algorithm uses the state distribution capabilities of the unscented Kalman filter to keep the uncertainty of the system to be remembered for a long trajectory, and at each time step, FastSLAM is used to produce local maps. Presented simulations and results evaluate the performance of the proposed approach using Unscented Kalman filter in a cluttered environment.