Berkeley's DexTacWAM turns video world models visuo-tactile, averaging 70.6 vs 38.0 on dexterous tasks
berkeley_ai · x · 2026-09-30
UC Berkeley's Humanoid Intelligence Center presents DexTacWAM, a visuo-tactile world-action model for dexterous manipulation.
- Adapts a pretrained video world model into a visuo-tactile one via continual vision-to-touch learning, for predictive multi-finger contact modeling and action generation
- Achieves the highest score on all six contact-rich tasks, averaging 70.6 vs 38.0 for the strongest baseline
- Key ablation: swapping the tactile world-model latent for direct tactile features drops the 4-task mean from 74.7 to 26.6
- Takeaway: world-model-style tactile latents beat injecting tactile features directly into the policy
- Motivation: multi-finger tactile data are scarce and expensive; vision transfer offers a scalable path
More from Embodied
- Raspberry Pi's $30 Smart Display Module Ships With Optional Hailo AI Acceleration — JeremyCMorgan · 2026-09-30
- Dyna Robotics' Taku masters new whole-body tasks from just 30 minutes of data — charlieholtz · 2026-09-30
- Dyna Robotics: Taku learns new whole-body tasks from just 30 min of robot data — JasonMa2020 · 2026-09-30
- YC boosts "agentic manufacturing" infra: describe a robot part, get it delivered — ycombinator · 2026-09-30
- Humanoid robot resumes laundry after being interrupted mid-task in impressive agent demo — JasonMa2020 · 2026-09-30
- Sentdex asks followers to name his robot, leaning toward "chonk" — Sentdex · 2026-09-30