Reflex enables quadrotors to recover autonomously after collisions
REFLEX: A Learned On-Board Reflex for Robust Autonomous Collision Recovery of Soft Quadrotors
Robotics
Summary
Quadrotor drones can crash into walls, and although some are built to survive impacts, they often get stuck afterward because they lose track of their position and speed. The authors introduce REFLEX, a new control method that helps drones recover their flight immediately after hitting something, without extra sensors or computers. REFLEX uses a special internal policy that stabilizes the drone using only its own movement data and was trained with simulations based on real crash data. Tests showed that REFLEX helped drones recover from crashes much more often and faster than standard autopilots, even at higher speeds than previously demonstrated.
What this means in practice
- •For drone operators: Use REFLEX-enabled drones to maintain stable flight immediately after high-speed collisions without extra sensors or companion computers
- •For robotics engineers: Implement onboard learned reflex policies in quadrotors to improve autonomous recovery from collisions, increasing drone robustness in cluttered environments
- •For aerial surveying companies: Deploy drones with REFLEX for autonomous missions that can continue after unexpected collisions, reducing downtime and mission failure risk$Commercial implications: Provides a practical technology to increase drone reliability and reduce costs for companies offering commercial aerial survey services
Authors
Jakob Hermeler, Luca Girardi, Gabriel Maquignaz, Stefan Leutenegger, Stefano Mintchev
Abstract
Collision-resilient airframes allow quadrotors to survive impacts, but do not ensure autonomous post-collision recovery. Innovation gating in stock onboard estimators often rejects valid external velocity measurements after high-speed impacts, compromising recovery. We present REFLEX, a collision-recovery flight mode that runs entirely on the flight controller, requiring neither contact sensors nor a companion computer. Its proprioceptive recurrent policy stabilizes the vehicle without position, velocity, heading, or altitude estimates. The translational estimate is reinitialized while REFLEX is active, and a trust gate checks estimator consistency before returning control to the stock autopilot. The policy is trained in simulation using proximal policy optimization, with episodes initialized from post-impact states measured in 47 real collisions. In hardware experiments, REFLEX recovered 19 of 22 wall impacts at 5.03 $\pm$ 0.09 m/s, compared with 1 of 14 for the matched stock autopilot. These included 15 consecutive recoveries in a single flight. A separate exploratory flight demonstrated two additional recoveries at 6.02 and 6.05 m/s within the same flight, exceeding the previously reported velocity for autonomous wall-collision recovery by 64%.