System identification is the 'right' sim2real fix, just painful on COTS humanoids
KyleMorgenstein · x · 2026-09-12
Following up on sim2real methods for humanoids, KyleMorgenstein notes that actual system identification is the "right" solution but requires heavy effort and is very hard on COTS robots — much easier if you build the robot yourself and can design test fixtures accordingly. @plafox teases an upcoming arXiv paper with a non-conservative alternative to the domain-randomization-plus-history approach.
More from Embodied
- OpenTac tactile sensing project to be showcased at CoRL — _krishna_murthy · 2026-10-28
- Hybrid LLM agent scores 83.2%, beats nine VLA policies on new SO-101 robot arm benchmark — Truman Hickok · 2026-10-28
- cixliv: robot fighting may be the first realized market where bots beat average humans — cixliv · 2026-10-28
- Can the Hiwonder MentorPi T1 run fully offline as a robot pet? Pi 5 questioned — oulu2006 · 2026-10-28
- Testing Mint Modeler to Build the Menlo Research Robot — freelerobot · 2026-10-28
- Khosla-backed startup builds a 'mind-reading' cap with 100,000 tiny chips to write AI prompts — Distinct-Question-16 · 2026-10-28