Multi-Agent Formations and Sensor Networks

Current interest in cooperative control of groups of mobile autonomous agents has led to the emergence of a broad range of exciting technical problem calling for the application of new and well-established control theoretic concepts together with other concepts devised originally for problems within entirely different fields. For example, the problem of maintaining a formation of autonomous agents by locally controlling the distances between each agent and its nearby neighbors cannot be formulated, let along fully understood, without a 100+ year old concept from the theory of structures, namely the notion of a rigid graph. To understand how a group of agents moving in the plane at the same speed can eventually all end up moving in the same direction by simply adjusting their individual headings to the averages of their neighbors' headings, requires the application of ideas from the theory of non-homogeneous Markov chains, a branch of probability theory which on the surface has nothing to do with the so-called flocking phenomenon just described. Flocking ing is also a good example of a topic originally motivated by a problem from an entirely different field, in this case statistical physics. Some problems can be addressed using ideas from control which one might at first not expect to be relevant.