Robust Distributed MPC Design for Large-Scale Multi-Robot Systems

For large-scale multi-robot systems $(\mathcal{L}\mathcal{S}\mathcal{M}\mathcal{R}\mathcal{S}{\text{)}}$ with polytopic uncertainty, a novel resilient distributed model predictive control $(\mathcal{R}\mathcal{D} - \mathcal{M}\mathcal{P}\mathcal{C})$ is proposed in this study. The time-varying system is first divided into a number of related subsystems. A distributed Kalman filter, which estimates unknown system characteristics using local data and measurements of nearby subsystems that are accessible through a network, is used to obtain interactions between subsystems. In order to ensure the closed-loop system’s stability, quadratic boundedness is utilized. The suggested approach for a massive multi-robot system is then put to the test in an unpredictable environment. Its effectiveness is contrasted with that of both centralized and decentralized forecasting techniques.