Apple robotics paper: 91% success in unfamiliar homes using synthetic sim data
FuSheng_0306 · x · 2026-11-17
The author, long skeptical of near-term humanoid commercialization, highlights Apple's new paper: an off-the-shelf robot carrying a Mac mini as its compute, powered by open-source Qwen as its brain, completed tasks like crawling under tables, pushing boxes, and sitting on chairs across 6 unfamiliar homes with a 91% success rate.
- No real human demonstration data was used: training relied entirely on 1,000 virtual households and 300 hours of synthetic, compute-generated data
- Takeaway: real home deployment is still far off, but bodies can be bought, data can be manufactured, and the brain keeps upgrading with LLMs — the humanoid breakthrough may ultimately be driven by large models
More from Embodied
- NVIDIA open-sources Alpamayo2-Edge, a VLA model for autonomous driving — nvidia · 2026-11-17
- AtomDog: a 3D-printed open-source quadruped robot you control from the browser — Stefania_druga · 2026-11-17
- Open-source LiDAR SLAM hits PyPI, no ROS 2 required: pip install lidarslam — rsasaki0109 · 2026-11-17
- Robotics week in review: Path raises $300M, Runway's Praxis-1 tops open leaderboard — lukas_m_ziegler · 2026-11-17
- 'Not in SF, Shenzhen, Bengaluru or Zurich? You're NGMI in robotics' sparks debate — retr0jirachi · 2026-11-17
- dynamic-3d-object-removal: NumPy-only LiDAR map cleaning hits 100 GitHub stars — rsasaki0109 · 2026-11-17