Self-organization of multiple-agents using information theoretic interactions

The recent sprout in the application of swarm intelligence to problems in varied fields has proved that this visionary technology is becoming practical. However, the organization and coordination of swarms is either heuristic or based on centralized approaches. In this paper, we propose a principled decentralized approach based on an information theoretic interaction principle to self-organize swarms. The goal of our algorithm is to spread the robots uniformly over a circular region using entropy maximization while guiding the collective towards a target. This algorithm minimizes the cost of the robots, since it is requires minimal hardware for its operation. Copyright (C) 2003 IFAC