robotics
A Humanoid Rewrote the Reward, Not Its Weights
InterEvolve revised staged reward programs at test time and verified candidates before retaining them.
Summary
InterEvolve revised staged reward programs at test time and verified candidates before retaining them.
InterEvolve pairs an object-aware behavioral foundation model with reward programs whose stages, completion tests and constants can be revised from execution feedback. An LLM agent changes program structure while a numerical optimizer tunes constants; candidates are verified across parallel simulations before joining a skill library. The authors report diverse simulated loco-manipulation behaviors and autonomous execution of evolved skills on a physical Unitree G1, but the evidence remains specific to the tested controller, tasks and hardware.
Why it matters
InterEvolve revised staged reward programs at test time and verified candidates before retaining them.
Limits and context
No additional limitation was separately recorded.
Key claims
InterEvolve revised staged reward programs at test time and verified candidates before retaining them.
Evidence: source-2026-10-03-003
Sources
- arXiv preprint 2610.02196arXiv · primary research
Corrections
No corrections have been recorded for this story.