Event-triggered and distributed model predictive control for guaranteed collision avoidance in UAV swarms

Distributed model predictive control (DMPC) is often used to tackle path planning for unmanned aerial vehicle (UAV) swarms. However, it requires considerable computations on-board the UAV, leading increased weight and power consumption. In this work, we propose offload multiple ground-based computation units. As simultaneously communicating recomputing all trajectories not feasible a large swarm with tight timing requirements, develop novel event-triggered DMPC that selects subset of most relevant UAV be replanned. The resulting architecture reduces consumption, while active redundancy provides robustness against unit failures. Moreover, guarantees collision-free UAVs linear dynamics. simulations, demonstrate our method can reliably plan trajectories, saving 60% network traffic required computational power. Hardware-in-the-loop experiments show suitable real quadcopter