New DMPC Algorithm Doubles Drone Swarm Capacity with Adaptive Safety Zones

A team from Germany (Linda Mümken, Diyar Altinses, Michael Schwung, Stefan Lier, Andreas Schwung) published a paper on a new control algorithm for autonomous drone swarms operating in space-constrained environments like warehouses, inspection corridors, and urban delivery routes. Their key innovation: replacing the traditional fixed-radius safety zone around each drone with an adaptive, speed-dependent safety sphere that scales with braking distance. At low speeds, zones shrink to maximize airspace density; at high speeds, they expand for safety. This addresses the inefficiency of worst-case fixed zones that waste capacity in congested scenarios.

The framework includes both centralized MPC and a distributed DMPC where each drone optimizes locally based on detected neighbors, even accommodating mixed fleets with non-cooperative agents. The authors prove feasibility up to the geometric packing limit at minimum radius, establish Lyapunov stability under sufficient conditions, and extend guarantees to the distributed setting via a contraction condition. Simulations show the adaptive method roughly doubles the admissible drone count, reduces traversal time through constrained passages by about 25%, and enables passage through openings that static safety zones cannot navigate. The centralized variant achieves a larger fraction of theoretical capacity, while the distributed version offers more realistic deployment for mixed-fleet operations.