An hybridization of global-local methods for autonomous mobile robot navigation in partially-known environments

Abstract

This paper deals with the navigation problem of an autonomous non-holonomic mobile robot in partially-known environment. In this proposed method, the entire process of navigation is divided into two phases: an off-line phase on which a distance-optimal reference trajectory enables the mobile robot to move from an initial position to a desired target which is planned using the B-spline method and the Dijkstra algorithm. In the online phase of the navigation process, the mobile robot follows the planned trajectory using a sliding mode controller with the ability of avoiding unexpected obstacles by the use of fuzzy logic controller. Also, the fuzzy logic and fuzzy wall-following controllers are used to accomplish the reactive navigation mission (path tracking and obstacle avoidance) for a comparative purpose. Simulation results prove that the proposed path planning method (B-spline) is simple and effective. Also, they attest that the sliding mode controller track more precisely the reference trajectory than the fuzzy logic controller (in terms of time elapsed to reach the target and stability of two wheels velocity) and this last gives best results than the wall-following controller in the avoidance of unexpected obstacles. Thus, the effectiveness of our proposed approach (B-spline method combined with sliding mode and fuzzy logic controllers) is proved compared to other techniques.