Distributed model predictive control for constrained multi-agent systems: a swarm aggregation approach

In this paper, a distributed receding horizon control scheme is developed for teams of autonomous agents customized as swarms within platoon configurations. Coordination and collision avoidance specs for multi-agent systems prescribe the use of high memory requirements for local computations and the exploitation of a growing number of sensors as the involved agents increase. In order to mitigate such a drawback, two key ingredients are exploited: 1) the swarm formation modelling that allows to consider in some sense several agents acting as a singleton; 2) an ad hoc model predictive scheme capable to adequately exploit swarm kinematics properties to ameliorate energy consumption savings.