Self-supervised monocular distance learning on an 18-gram DelFly @microuav
Self-supervised monocular distance learning on an 18-gram DelFly  @microuav
Uploaded June 2018 | Updated November 2026, 3 hours ago
Video of an 18-gram DelFly that quickly learns how to estimate distance with a single camera. The DelFly has a tiny infrared sensor allowing it to avoid obstacles. When it avoids an obstacle, it thinks back of what it saw during the seconds just before the avoidance maneuver. In this way, it can learn to see distances and avoid risky, abrupt avoidance maneuvers.

Published at:
Lamers, K., Tijmons, S., De Wagter, C., & de Croon, G. (2016, October). Self-supervised monocular distance learning on a lightweight micro air vehicle. In Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International Conference on (pp. 1779-1784). IEEE.
Self-supervised monocular distance learning on an 18-gram DelFlySwarm exploration by tiny flying robots - short versionARdrone RC controledNanoFlowNet: Real-Time Dense Optical Flow on a Nano QuadcopterIMAV 2014 Presentations - Fabrizio SchianoIMAV2015 MAVlab TUDelftTU Delft Quadshot - Proof of FlightEnd-to-end Neural Network Based Optimal Quadcopter ControlIMAV 2022 - Greenhouse ChallengeNovel versatile hydrogen drone developed by TUDelft - MAVLab [EN]Stability-based Scale Estimation for Monocular SLAMLisaS on quad and Heli
MAVLab TU Delft |

Self-supervised monocular distance learning on an 18-gram DelFly

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER