Robust Region-Tracking for Multi-Agent Systems Using Sliding Mode Control

Control of autonomous agent swarms is studied for targeted flocking exercises. The desired decentralized control also requires robustness against modeling uncertainties and bounded unknown forces. In this analysis, we consider the task of robustly driving multiple agents to a moving ‘target region’, as repulsive interactions help spread out the agents. An unconventional form of sliding mode control is implemented to provide the robust attraction towards the region’s center. For robustness a ‘boundary layer’ is conceived, which corresponds to the desired target region. The attraction is intentionally softened inside this target region, allowing agents to create a final formation utilizing their repulsion forces. Examples are given for moving circular and elliptical regions which illustrate the effectiveness of the proposed strategy.