A Self-Tuning Navigation Algorithm for a Robotic Vehicle

Abstract

This research deals with the problem of navigation for an industrial automated guided vehicle (AGV). The vehicle navigates by using a combination of odometric dead reckoning (DR) and position measurements from known landmarks. The measurement system could be a video camera, optical or radio frequency triangulation, sonar image mapping or others. A parameter identification system is used to identify unknown or changing model parameters, and is coupled with a Kalman filter. The system provides a better estimate of the vehicle's position and heading than could be obtained from either the measurement or dead reckoning alone. The subject of this paper is the parameter identification algorithm.