Agile Fall Recovery for Quadrotors with Bidirectional Thrust via Reinforcement Learning

2026-06-15

AI Summary

This work addresses the challenge of autonomous recovery for quadrotors after ground impact due to collisions or failures in real-world environments. The authors propose a reinforcement learning approach that relies solely on lightweight onboard sensors, integrating a recurrent policy network, an asymmetric Actor-Critic architecture, and an incremental nonlinear dynamic inversion (INDI) controller. This framework enables zero-shot transfer from arbitrary grounded poses to stable hover without explicit state estimation. Notably, it is the first method to train a robust policy in high-fidelity simulation using only limited and unreliable onboard perception, which can be directly deployed on physical hardware. Experiments demonstrate reliable recovery across diverse initial orientations, under wind disturbances, and with added payload, while ablation studies confirm the efficacy of the key architectural components.

Abstract

Autonomous fall recovery is a critical capability for quadrotors operating in real-world environments, where collisions or failures may leave the vehicle resting on the ground in an arbitrary attitude. This problem is challenging because recovery must be achieved under limited onboard sensing, in constrained free space, with ground contact, and in the presence of unknown disturbances. In this letter, we present an RL-based framework for autonomous fall recovery of a quadrotor from arbitrary ground attitudes to stable hover using only lightweight onboard sensors. To address severe partial observability and intermittent sensor invalidity, we train a recurrent policy within an asymmetric actor--critic architecture, leveraging an Incremental Nonlinear Dynamic Inversion (INDI) controller to track the policy output. Combined with high-fidelity simulations of motor response and optical flow, the overall training framework significantly reduces the sim-to-real gap. Simulation ablation studies validate the importance of the main design choices, while real-world experiments demonstrate zero-shot transfer and robust recovery under different initial attitudes, wind disturbances, and additional payloads. These results demonstrate that agile quadrotor fall recovery can be achieved without explicit state estimation using only limited and unreliable onboard sensing.

Problem

• fall recovery
• quadrotor
• autonomous
• onboard sensing
• arbitrary attitude

Innovation

• reinforcement learning
• fall recovery
• quadrotor
• partial observability
• sim-to-real transfer