Policy trained entirely in imagination matches PPO with 7.9× fewer simulator transitions
EpisodeYang · x · 2026-11-06
Matthias Caro's team shows a robot locomotion policy trained entirely "in imagination" (inside a learned world model, no simulator rollouts) achieving 81.8% command-tracking success on the same locomotion test versus 80.7% for a PPO policy trained in simulation — using 7.9× fewer simulator transitions and roughly half the training time. More results teased for soon.
More from Embodied
- Hacker reverse-engineers Apple's unreleased touchscreen MacBook HID APIs — yoimnotkesku · 2026-11-06
- Apple's mystery robot reportedly uses a Mac Mini as its planner brain — micoolcho · 2026-11-06
- Microdock v2 lets robot sit while docked, cutting servo power to zero — natesiggard · 2026-11-06
- Building an Asimov Robot in 22 Hours — and It Worked Out of the Box — whurley · 2026-11-06
- NEO-VLA drops the vision encoder: native VLA pretraining pays off on RoboTwin — liuziwei7 · 2026-11-06
- Gold Tesla Cybercab and Optimus shown at UP Summit — DMaguireARK · 2026-11-06