End-to-End RL Drone Policy Passes Sim2real on Multiple Hardware

yacineMTB · x · 2026-08-27

A generalist drone policy trained end-to-end via reinforcement learning (RL) from pixels is presented. It has passed sim2real validation on multiple drone types and operated both remotely and on a Teensy 4.1 microcontroller. The training process took only 2 minutes.

Original post →

More from Embodied

Embodied channel →