Techniahqrobot | humanoid robots @techniahqrobot · 02 Sep 2026
We’re cooked if robots can start learning from bodies they don’t even have. NVIDIA Brown Columbia and Harvard researchers released Hydra-0, a world model that represents robot actions as pixel motion. It was trained on about 2,202 hours of multi-embodiment video spanning human https://t.co/0SGYmzl1Ft
8 709Views
43Likes
5Reposts
6Replies
1Quotes
15Bookmarks
Is that a lot?
53.8×vs this author's median162 views is typical
67Percentile for this authorof 9 recent posts
5.32×vs 10K–100K median1 637 views is typical
49.56%Reachviews ÷ followers
0.63%Engagement rateof viewers reacted