LEARN: Learning End-to-End Aerial Resource-Constrained Multi-Robot Navigation

TL;DR

LEARN is a lightweight, end-to-end reinforcement learning framework enabling resource-constrained nano-UAVs to navigate cluttered environments safely and efficiently, outperforming existing planners in simulation and real-world tests.

Contribution

The paper introduces LEARN, a novel two-stage RL approach that combines low-res sensors and simple planning with attention-based policies for multi-UAV navigation.

Findings

LEARN outperforms state-of-the-art planners by 10% in simulation.

LEARN enables onboard flight of nano-UAVs in complex environments.

LEARN achieves real-world navigation at speeds up to 2.0 m/s.

Abstract

Nano-UAV teams offer great agility yet face severe navigation challenges due to constrained onboard sensing, communication, and computation. Existing approaches rely on high-resolution vision or compute-intensive planners, rendering them infeasible for these platforms. We introduce LEARN, a lightweight, two-stage safety-guided reinforcement learning (RL) framework for multi-UAV navigation in cluttered spaces. Our system combines low-resolution Time-of-Flight (ToF) sensors and a simple motion planner with a compact, attention-based RL policy. In simulation, LEARN outperforms two state-of-the-art planners by $10\%$ while using substantially fewer resources. We demonstrate LEARN's viability on six Crazyflie quadrotors, achieving fully onboard flight in diverse indoor and outdoor environments at speeds up to $2.0 m/s$ and traversing $0.2 m$ gaps.