Supporting real-time update of an environment representation for autonomous mobile robots

Abstract

A model of the environment is a mandatory requirement for the autonomy of a mobile robot. As the environment may change over time, methods are necessary to update the model information, keeping pace with the sensor frame rate. Our approach to cope with this real time problem is to predict virtual sensor images, which are then compared against the real perceptions. Two alternative methods are detailed: model based prediction and history based prediction.<>