Integrating elementary functions for autonomous navigation of a mobile robot

Abstract

This paper describes an implementation of autonomous navigation for a mobile robot in an indoor environment. Autonomous navigation requires map building, localization, path planning, obstacle detection, obstacle avoidance, path following, and so on. For the map building, ICP(Iterative Closest Point) builds floor map represented as occupancy grids. MCL(Monte Carlo Localization) updates robot location by comparing range data from LRF(Laser Range Finder) with the range information calculated using the floor map and robot location. Dijkstra algorithm selects way points connecting the current robot location to the goal location with the shortest path length. Two artificial force contribute to obstacle avoidance. Artificial potential field force repulses the robot path away from obstacles while artificial elastic force contracts the robot path. These methods are integrated and implemented in a robot, and resulted in reliable autonomous navigation in a building.