The Dempster-Shafer Approach to Map-Building for an Autonomous Mobile Robot with Fuzzy Controller

Abstract

This paper develops a sensor based navigation method that utilizes fuzzy logic and the Dempster-Shafer evidential theory for mobile robot in uncertain environment. The proposed navigator consists of two behaviors: Obstacle avoidance and goal seeking. To navigate reliably in the environment, we facilitate a map building process before the robot finds a goal position and create a robust fuzzy controller. In this work, the map is constructed on two-dimensional occupancy grids. The sensor values are fused into the map using the Dempster- Shafer inference rule. Whenever the robot moves, it catches new information about the environment and replaces the old map with new one. With that process the robot can wander and find the goal position. The usefulness of the proposed method is verified by a series of simulations.