Berkeley Releases DexTacWAM, a Vision-Tactile World Model for Dexterous Manipulation

UC Berkeley and UIUC introduced DexTacWAM, a vision-tactile world-action model that converts pretrained video world models into tactile-aware ones. It outperformed baselines across all dexterous manipulation tasks.

2026-09-30 ~ 2026-09-30 · 2 related posts

• Berkeley's DexTacWAM turns video world models visuo-tactile, averaging 70.6 vs 38.0 on dexterous tasks — berkeley_ai · 2026-09-30

1 near-duplicate retellings: berkeley_ai