Position estimation of mobile robots with internal and external sensors using uncertainty evolution technique

Abstract

A wheeled mobile robot on a flat plane can estimate its current location and orientation from the cumulative wheel rotation (dead reckoning). The method, however, has the drawback that estimation errors are accumulated as the robot moves, and the accuracy of estimation decreases. To address this problem, a positioning technique using an external sensor is needed. The authors propose a more general positioning method whereby the robot can act autonomously and move along a path unlimited by positioning. The robot can modify the position estimated by dead reckoning by repeating low-level sensing asynchronously and intermittently. This method uses not only the estimated position data but also the accuracy of the dead reckoning and the external sensor effectively. >