Robust Evacuation for Multi-Drone Failure in Drone Light Shows
Minhyuk Park, Aloysius K. Mok, Tsz-Chiu Au
TL;DR
This work addresses safety and reliability in drone light shows under multi-drone failures that can trigger cascading collisions. It combines a Social LSTM with attention to forecast fallen-drones trajectories within a discretized space-time grid, with an RRT-based parking algorithm to compute near-optimal evacuation paths toward designated parking spaces, and a recovery mode using hidden drones to replace failed units after the incident. Central contributions include a probabilistic occupancy model ${Pr}(\tau)$ and a collision-free probability ${Pr}_{safe}$ to guide planning, a parking formulation that maximizes safety via a path cost $C^*$ where ${Pr}_{safe}(\mathcal{T}_{E^*})=e^{-C^*}$, and a practical recovery strategy to restore the show. The framework demonstrates robustness gains in simulation, enabling safer, continuous swarm performances and informing design of fault-tolerant multi-drone systems for complex show environments.
Abstract
Drone light shows have emerged as a popular form of entertainment in recent years. However, several high-profile incidents involving large-scale drone failures -- where multiple drones simultaneously fall from the sky -- have raised safety and reliability concerns. To ensure robustness, we propose a drone parking algorithm designed specifically for multiple drone failures in drone light shows, aimed at mitigating the risk of cascading collisions by drone evacuation and enabling rapid recovery from failures by leveraging strategically placed hidden drones. Our algorithm integrates a Social LSTM model with attention mechanisms to predict the trajectories of failing drones and compute near-optimal evacuation paths that minimize the likelihood of surviving drones being hit by fallen drones. In the recovery node, our system deploys hidden drones (operating with their LED lights turned off) to replace failed drones so that the drone light show can continue. Our experiments showed that our approach can greatly increase the robustness of a multi-drone system by leveraging deep learning to predict the trajectories of fallen drones.