T-WLA: Bridging Slow Vision and Fast Touch

Key points

• In contact-rich manipulation, contact conditions can change before a visual policy updates its action plan.1
• The same memory is cached for a lightweight tactile editor, which combines it with the active action chunk and incoming touch to predict residual corrections to unexecuted actions without rerunning the visual branch.1
• Future-image and future-pressure prediction provide training-only supervision, encouraging the visual memory and tactile features to capture interaction dynamics relevant to action generation and correction.1
• We evaluate T-WLA on tactile-conditioned hand-motion prediction using EgoTouch and Tachin, and on four real-world dexterous manipulation tasks.1

Sources (1)

• [1]T-WLA: Bridging Slow Vision and Fast Touch
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 06:36 PM

  We introduce T-WLA, a world-language-action architecture that uses shared visual memory to guide both action generation and tactile correction.

  In contact-rich manipulation, contact conditions can change before a visual policy updates its action plan.

Extractive summary: sentences quoted from the sources.