SafeSwarm: Decentralized Safe RL for the Swarm of Drones Landing in Dense Crowds

TL;DR

This paper presents SafeSwarm, a decentralized reinforcement learning-based system enabling a drone swarm to land safely and accurately in crowded environments, avoiding collisions and adapting to different dynamics.

Contribution

The paper introduces SafeSwarm, a novel decentralized safe reinforcement learning approach for drone swarms to perform collision-free landings in complex, dynamic environments.

Findings

Achieved 2.25 cm landing accuracy

Demonstrated collision-free landings in real-world tests

System robustly handles moving landing pads and obstacle avoidance

Abstract

This paper introduces a safe swarm of drones capable of performing landings in crowded environments robustly by relying on Reinforcement Learning techniques combined with Safe Learning. The developed system allows us to teach the swarm of drones with different dynamics to land on moving landing pads in an environment while avoiding collisions with obstacles and between agents. The safe barrier net algorithm was developed and evaluated using a swarm of Crazyflie 2.1 micro quadrotors, which were tested indoors with the Vicon motion capture system to ensure precise localization and control. Experimental results show that our system achieves landing accuracy of 2.25 cm with a mean time of 17 s and collision-free landings, underscoring its effectiveness and robustness in real-world scenarios. This work offers a promising foundation for applications in environments where safety and precision are paramount.