ActionPiece rethinks action tokenization for VLA models, hits 94.8% on LIBERO

burny_tech · x · 2026-09-20

The paper ActionPiece: Rethinking Action Tokenization for VLA Models redesigns action tokenization for vision-language-action models, preserving physical relationships between actions via Physical Rank Consistency (PRC) instead of naive discretization.

The key idea is encoding physical constraints directly into the tokenizer to improve robot policy generalization.

Original post →

More from Embodied

Embodied channel →