Preprint
Oct 2026
Test-Time Compute for Humanoid Behavior Foundation Models
Humanoid Behavior Foundation Models (BFMs) can execute impressive motions, but downstream applications such as loco-manipulation demand high whole-body positional accuracy. Existing BFMs suffer from a systematic command-execution gap, where executed motions deviate from desired ones. We show that additional test-time c...
Wei-Ji Xie, Jun-Song Wu, Zi-Bo Zhou et al.
· 1 citation