Performance evaluation of robot motion incorporating uncertainties in sensors and motion

Abstract

This paper describes the performance evaluation of mobile robot motion. The evaluation is based on a probabilistic model of robot motion and sensor output. In the simulator, robot motion includes uncertainties in translational velocity and rotational velocity. Among various kinds of sensors, implemented in the simulator is a range sensor which is used widely for obstacle detection, map making, and localization. The sensor model includes measurement noise with Gaussian distribution around true range value, failure of detection, and unexplainable random measurement. The simulator is used to test a collision avoidance algorithm, map making, and motion coordination algorithm. The results from a simulated environment with uncertainty are compared with the results from a simulator with no uncertainty. In practical application of a mobile robot, major problems arise due to the unexpected uncertainty of motion and sensors. The simulator is useful to predict problems of robot motion before the robot is deployed in the field. For wider availability, the simulator can be further developed to include error models in various sensors. © 2009 Springer Berlin Heidelberg.