Sensor-based control of AGVs

Abstract

This article summarises work on a number of related sensor-guided mobile robot projects at Oxford University. Two fundamental problems are discussed: navigation and obstacle avoidance. Beacons are central to navigation. Implemented systems are described that make different assumptions about the environment and use different beacons. An implemented system is also described that can sense and avoid obstacles on the fly, without stopping. It is based on a layered architecture. A fully decentralised Kalman filter has been applied to a number of sensor integration tasks, including tracking an object visually as it moves around a room.