ECAI 2004 Conference Paper

[PDF] [full paper] [prev] [tofc] [next]

Algorithms for Distributed Exploration

Thomas Walker, Daniel Kudenko, Malcolm Strens

In this paper we propose algorithms for a set of problems where a distributed team of agents tries to compile a global map of the environment from local observations. We focus on two approaches: one based on behavioral agent technology where agents are pulled (or repelled) by various forces, and another where agents follow a approximate planning approach that is based on dynamic programming. We study these approaches under different conditions, such as different types of environments, varying sensor and communication ranges, and the availability of prior knowledge of the map. The results show that the simpler behavioral agent teams perform at least as well, if not better, than the teams based on approximate planning and dynamic programming. We also compare our approaches to theoretical lower bounds on optimal performance. The research has not only practical implications for distributed exploration tasks, but also for any distributed search problem that is analogous to spatial exploration problems.

Keywords: Multi-Agent Systems

Citation: Thomas Walker, Daniel Kudenko, Malcolm Strens: Algorithms for Distributed Exploration. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.84-88.


[prev] [tofc] [next]


ECAI-2004 is organised by the European Coordinating Committee for Artificial Intelligence (ECCAI) and hosted by the Universitat Politècnica de València on behalf of Asociación Española de Inteligencia Artificial (AEPIA) and Associació Catalana d'Intel-ligència Artificial (ACIA).