ECAI 2004 Conference Paper

[full paper]

Indirect and Conditional Sensing in the Event Calculus

Jeremy Forth, Murray Shanahan

Controlling the sensing of an environment by an agent has been accepted as necessary for effective operation within most practical domains. Usually, however, agents operate in a partially observable domains where not all parameters of interest are accessible to direct sensing. In such circumstances, sensing actions must be chosen for what they will reveal indirectly, through an axiomatized model of the domain causal structure. This article shows how sensing can be chosen so as to acquire and use indirectly obtained information to meet goals not otherwise possible. This paper presents Event Calculus extended with a knowledge formalism and causal ramifications, and used to show how inferring unknown information about a domain leads to conditional sensing actions. Such an extended theory is able to reason about knowledge for the purposes of describing and controlling sensing actions in a partially observable causal domain. Partial observability is usual in most practical circumstances, and restricts an agent in its range of directly sensed parameters of the environment. In cases where the state of an unobservable environment fluent is required, this information must be inferred using the causal domain model along with other known fluents. Determining which fluents to sense, and when, is the problem to be solved at plan time.

Keywords: Cognitive Robotics, Autonomous Agents

Citation: Jeremy Forth, Murray Shanahan: Indirect and Conditional Sensing in the Event Calculus. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.900-904.