Strategy of perception and temporal representation of beliefs

Abstract

An autonomous system uses sensors in order to update an uncertain representation of its environment, which must remain well-grounded with regards to the risks and the constraints of its mission. The choice of a framework for the representation of beliefs which is adapted to deal with both time and uncertainty is a crucial point. A perception strategy should be founded on an evaluation of the relevance of the beliefs of the system in order to collect the most useful information at the most opportune time. This paper addresses the problem of taking time and ignorance into account in the representation of the uncertain beliefs of the system about the world. The paper proposes a well-adapted approach to the design of a perception strategy for an autonomous surveillance system in a changing environment together with initial results of simulations.