Robust distributed MPC for cooperation of linear multi-agent systems

In multi-agent systems, achieving coordinated behavior is a fundamental challenge, especially in the presence of disturbances and decentralized decisionmaking. This thesis presents a sequential robust distributed model predictive control scheme for cooperative task of multiagent linear systems subject to individual constraints. We extend a distributed MPC algorithm that uses tracking MPC with artificial references to enable cooperation among agents without external guidance. By integrating Tube-based MPC, we ensure robustness against additive disturbances. Each agent solves an individual optimization problem for the artificial reference and compute an input to track it, sharing only the reference with its neighbours. We prove the recursive feasibility and asymptotic stability of the closed-loop system. Finally, we validate the proposed control scheme through numerical simulations by applying it to consensus and formation tasks.