The egocentric data boom: Maxinsights has delivered 2M+ hours for robot training
机器之心 · wechat · 2026-09-17
A Jiqizhixin piece on the new scaling axis for robotics: egocentric human experience data. Dyna-2 pushed first-person data to 1M hours with continued gains, and GenesisAI's GENE-26.5 also trains on it, echoing how LLMs scaled on internet data. Behind these efforts sits data vendor Maxinsights: a six-continent collection network, 1.5M+ hours of hand-tracked, language-annotated data, 2M+ hours delivered to clients including DynaRobotics, Google DeepMind, Figure and 1X. Its pipeline features Coverage-aware Collection (agents diagnose data blind spots and dispatch targeted recollection), state-level annotation, in-house video understanding model MaxVLM (near-SOTA at 1/10 the cost), SLAM-based 3D hand-trajectory reconstruction, and a semiconductor-style 'Yield' concept at 98%, targeting 10M hours of training-ready data.
More from Embodied
- ActionPiece rethinks action tokenization for VLA models, hits 94.8% on LIBERO — DeepCybo · 2026-09-17
- Tesla Optimus passes CAPTCHA designed to prove you're not a robot — TansuYegen · 2026-09-17
- Bionic robotic fish flexes body like a real fish at China trade fair — rohanpaul_ai · 2026-09-17
- GPT-6 Astra reportedly pretrained on 100k+ GPUs at Stargate, with big real-to-sim implications — erwincoumans · 2026-09-17
- UBTECH launches hyper-realistic U1 companion humanoid robots, sparking privacy questions — TansuYegen · 2026-09-17
- Ulysses co-founder on Mako modular underwater robots and seabed restoration — Scobleizer · 2026-09-17