Reality Check launches first public robot manipulation benchmark with 14,400 real-world rollouts
DJiafei · x · 2026-09-29
Nicolas Keller's team launched Reality Check, a public leaderboard and the first public robot manipulation benchmark. Key facts:
- 14,400 real-world rollouts across four models, multiple tasks and data regimes, including object placement tasks
- Results reported with confidence intervals; tests run on FR3 Duo stations
- Commenter DJiafei notes MolmoAct2 is SOTA on this new benchmark and calls this "the year of RoboEval"
The launch signals robotics is entering a systematic benchmark era akin to LLM evaluations.
More from Embodied
- Tsinghua humanoid robot plays badminton with one policy from just 30 min of human motion data — ChongZzZhang · 2026-09-29
- Bose's $99 USB-C wired ANC earbuds ditch the battery brick by drawing power from your phone — aakashgupta · 2026-09-29
- Unitree's Dex5-1 dexterous hand: 22 DoF, all backdrivable, from $6,500 — lukas_m_ziegler · 2026-09-29
- Niantic opens 100 real-world 3D scenes for robot sim; Vesoma stealth-launches humanoid Kyle — lukas_m_ziegler · 2026-09-29
- Ex-DeepMind director's startup Vesoma unveils humanoid Kyle, robot due 2027 — lukas_m_ziegler · 2026-09-29
- Cognex acquires RealSense for $500M cash to enter robot 3D perception — lukas_m_ziegler · 2026-09-29