Towards a Formal Representation of Driving Behaviors

Abstract

To successfully perform their required tasks, intelligent information agents require accurate and meaningful communication and integration among other agents and information resources. However, the applications and infrastructure of information technology are rife with heterogeneity at many levels. This paper explores applying formal ontologies to address the challenges faced in developing and applying autonomous agents for the purpose of driving unmanned vehicles. Specifically, we are developing a formal ontology to encapsulate the rules of on-road driving, facilitating the ability for an autonomous vehicle to drive on a busy street while following the “rules of the road”. This ontology will constitute a major component of the internal knowledge representation within the Real-Time Control System (RCS) [1], which is the architecture that will control the unmanned vehicle.

This work is being performed under the auspices of the DARPA-funded MARS project [2]. The technology goal of the MARS program is to understand and satisfactorily address the key challenges to the realization of reusable software for autonomous mobile system.