Uranus: diffusion-based robot simulator hits 24 FPS with open-ended rollout
D-Robotics · hf · 2026-09-24
D-Robotics released Uranus, next-gen simulation infrastructure for embodied AI. It is a data-driven robot simulator built on a joint-trajectory-conditioned autoregressive diffusion model, with three key capabilities:
- Streaming, open-ended rollout: receives future joint-position trajectories online and autoregressively generates one latent frame per step (four RGB frames), with no fixed horizon;
- Low latency: 24 FPS after inference optimization;
- Scalable robot control: a unified interface for synchronized multi-view generation across diverse embodiments and camera setups.
The team evaluates on in- and out-of-distribution data, clearly noting current limitations. Code and model weights are open-sourced.
More from Embodied
- Meta Muse adds voice and live video, coming to Meta glasses for hands-free use — FinanceYF5 · 2026-09-24
- Meta Connect unveils 8 new products and features, from Muse to AI glasses — FinanceYF5 · 2026-09-24
- Creator apologizes after AI agent auto-posted at 2am; revokes unsupervised posting permissions — LinusEkenstam · 2026-09-24
- Microsoft Open-Sources LIBERO Robot Learning Dataset With 282K Rows on Hugging Face — victormustar · 2026-09-24
- BitRobot to host 3 robotics grand challenges at IROS 2026 with 50+ research teams — micoolcho · 2026-09-24
- Meta reportedly made a keychain just for talking to AI — prasenx · 2026-09-24