Uranus Enables 24 FPS Autoregressive Diffusion for Embodied AI Simulation
September 22, 2026
Uranus uses a joint-trajectory-conditioned autoregressive diffusion model to provide streaming, open-ended robot rollouts. The system achieves 24 FPS inference and supports synchronized multi-view generation across diverse robot embodiments.
HOW THIS AFFECTS YOU
●
builderYou can generate high-fidelity robot training data with low-latency, online trajectory conditioning.
●
researcherYou can scale policy training using a data-driven simulator that supports open-ended rollouts.