MaxInsights pushes physical AI data to 2M hours of real-world experience, 450K hours produced monthly
CyberRobooo · x · 2026-09-23
Following Dyna-2's finding that 1M+ hours of egocentric human video can scale robot learning without robot data in pretraining, MaxInsights is pushing to 2M hours of real-world experience.
- Frames physical AI data scaling around two dimensions: Experience Scale (hours) and Experience Density (learnable physical info per hour) — Effective Experience ≈ Hours × Information per Hour; ordinary video and footage capturing hands, contact, force, and temporal changes differ vastly in value
- Dataset combines egocentric video, hand/body motion capture, tactile/force signals, 3D information, and data evaluation for VLA and World Action Models
- Claims 2M hours accumulated, up to 450K hours of video production per month, 50K+ scene types, and 9 categories of physical signals
Related event: MaxInsights Scales Physical AI Dataset to 2M Hours of Egocentric Video(5 posts)→
More from Embodied
- Figure's Helix robot demo slammed as misleading: camera cuts away before tasks finish — RexDouglass · 2026-09-23
- Musk predicts 1 billion humanoid robots within 10 years, 100 billion in 20 — XFreeze · 2026-09-23
- VONDER's $299 AI glasses skip camera and display, offer ChatGPT, Claude or Gemini plus a memory graph — Div_pradeep · 2026-09-23
- Figure Ships Helix 2.5, New Humanoid Robot Demo Goes Viral — adcock_brett · 2026-09-23
- QwenBook vs Googlebook: AI-native PCs start rebuilding the OS around agents — APPSO · 2026-09-23
- Humanoid Robot Pemba Set to Attempt First-Ever Everest Summit in 2026 — DynamicWebPaige · 2026-09-23