Distributed Computations and Control in Multi-Agent Systems

For large scale multi-agent systems, the design of optimal or robust distributed control laws has been both a theoretical and computational challenge. Recently, computationally efficient techniques based on special matrix structure preserving algorithms have been developed that allow arbitrarily non-conservative optimal control synthesis for heterogeneous systems arranged on a line. In this paper, we show how such design methods can equivalently be performed on distributed processors with distributed memory, connected in the same structure as the original system; an important step in the development of intelligent and adaptive distributed systems.