Technology News and Headlines

Nvidia Releases DreamDojo, a Robot ‘World Model’ Trained on 44,000 Hours of Human Video

Nvidia unveiled DreamDojo, a robot world model trained on a 44,000‑hour human egocentric video dataset, enabling robots to acquire physical intuition by observation before hardware‑specific fine‑tuning. The DreamDojo‑HV dataset is 15× longer, contains 96× more skills and spans 2,000× more scenes than any prior world‑model collection. A two‑phase training pipeline first learns general physics from the video corpus, then post‑trains on target robots, delivering real‑time 10 FPS rollouts for over a minute on multiple humanoid platforms. The launch dovetails with Nvidia’s broader push into robotics amid record AI infrastructure spending.