Physical World Datasets for Embodied AI
Six production corpora spanning first-person video, synchronised multi-view capture, simulation-ready geometry, human kinematics, aligned multi-modal bundles and dense expert annotation. Every dataset ships with published quality gates and commercial training rights.
MOVAS-Ego 100K
First-Person Video for Embodied AI
100,000 hours of head-mounted, first-person video across commercial, domestic, industrial, food-service and retail environments — engineered for VLA pre-training and world-model learning.
MOVAS-4D Multi-View
Synchronised Multi-Camera 4D Capture
10,000+ hours of time-synchronised multi-camera capture that resolves single-view occlusion and recovers true 3D spatial structure across time.
MOVAS-3D Assets
Simulation-Ready Geometry
Reconstructed object meshes and full environment scans, physically scaled and simulation-ready for Isaac Sim, MuJoCo and Genesis.
MOVAS-MoCap
Millimetre-Precision Human Kinematics
Optical and inertial motion capture of full-body and hand kinematics during real manipulation tasks, aligned to the egocentric and multi-view corpora.
MOVAS-Sync
Aligned Multi-Modal Bundles
Video, audio, text and action streams delivered as hardware-synchronised bundles that drop directly into a training loop without alignment work.
MOVAS-5D Annotation
Expert-Graded Five-Dimension Labels
An annotation layer spanning object detail, camera dynamics, scene summary, visual properties and semantic context — graded by domain experts, not crowd workers.