Ropedia W0 unifies human and robot loco-manipulation world model

Ropedia W0, introduced in this arxiv preprint, is pitched as the first egocentric loco-manipulation world foundation model that spans both humans and robots with diverse embodiments.

The authors curated what they call Xp-Data: 10k hours of egocentric human data plus 3k hours of robot data, run through 'unified multimodal annotations and systematic quality control.' Locomotion is encoded through camera poses, manipulation through projected skeletons. For robots, the paper combines embodiment-specific forward kinematics with 'extensive manual keypoint annotation and correction.' A separate benchmark, Xp-Teleop, is built from 'high-quality dexterous humanoid teleoperation data.'

The team reports 'state-of-the-art performance across five cross-embodiment benchmarks (DreamDojo, Ψ₀, ω-HOME, Xp-Teleop, and AgiBot) and on held-out human egocentric video, from zero-shot inference to few-shot and full-data post-training.' One concrete follow-on: 'After distillation, our model supports real-time, minute-long rollouts, enabling live VR interaction.'

The central empirical claim is simple. 'Scaling human pre-training data improves performance on both human and robot evaluations,' the authors write. That is the paper's bet: feeding more first-person human video keeps lifting the robot side too.

The abstract publishes no per-benchmark numbers, no baseline comparisons, no training compute, no model size. It is a single-team preprint; whether the five benchmark wins hold up under independent reruns is the next step.