EgoVerse and Newton are emerging as the two global infrastructure layers for embodied AI
量子位 · wechat · 2026-07-23
EgoVerse and Newton are becoming the two global bases for embodied AI
This long post argues that embodied AI is now coalescing around two infrastructure layers:
- EgoVerse: a growing first-person human-operation data ecosystem for robot learning
- Newton: a physics simulation foundation for robot training and evaluation
EgoVerse in numbers
- 1,362 hours of human demonstration data
- About 80,000 episodes
- 1,965 tasks, 240 scenes, and 2,087 contributors
What the ecosystem is trying to solve
- Standardize how human demonstrations are collected and shared
- Make data portable across labs, robots, and task definitions
- Validate whether human-first data actually transfers into robot skills
Who is involved
- Universities: Georgia Tech, Stanford, ETH Zurich, UC San Diego
- Companies: Meta, Scale AI, and several embodied-AI startups
- A Chinese company, Guanglun, is highlighted as the only participant appearing in both EgoVerse and Newton
Why Guanglun matters
The post says Guanglun contributes a full quality loop for human data:
- edge-side agent monitoring during collection
- unified cloud processing across hardware types
- VIO, hand/body annotation, and action semantics
- simulation and benchmark validation to check whether the data is worth training on
The broader point: embodied AI is moving from ad hoc demos to shared infrastructure, and data plus simulation are becoming the new global standards.
More from Embodied
- Johns Hopkins Launches Full-Stack Hands-on Robot Learning Class with SO-101 Arm Kits — _krishna_murthy · 2026-09-11
- SyncWorld: In-Context Robot World Model Simulates Unseen Views and Embodiments Zero-Shot — ChongZzZhang · 2026-09-11
- A 3D Pose Dataset for Dogs Released — ducha_aiki · 2026-09-11
- Swaayatt demos autonomous driving at 52 km/h on mountain roads, self-recovers after skid — sanjeevs_iitr · 2026-09-11
- AUAR's MicroFactory brings a deployable robotic wood-panel factory to the construction site — lukas_m_ziegler · 2026-09-11
- Musk: Cybercab certified at 165 Wh/mi, the most efficient production EV ever — elonmusk · 2026-09-11