TacPAC: Tactile Prediction and Real-Time Action Correction in World-Action Models for Contact-Rich Manipulation
Classification
View four quadrants- Major category
- WAMs
- Quadrant
- Not assigned
- Architecture
- Not assigned
- Prediction paradigm
- Not assigned
- Source review status
- Not assigned
Category review. Joint video/tactile/action generation supplies a cached plan, and incoming touch corrects unfinished robot commands against predicted contact. This is a contact-rich WAM controller, not a tactile encoding component. Reading evidence
Contribution
TacPAC predicts future visual and tactile observations while planning an action chunk, then uses incoming touch to revise its unfinished actions against a fixed prediction-and-plan cache. On five physical manipulation tasks, it reports 64% average success versus 48% for T-Rex and 22% for its vision-only ablation. The central evidence is the controlled cache-access ablation, not visual prediction quality alone.
Abstract
An abstract has not been added yet.
Affiliations
School of Data Science, Fudan University; Shanghai Innovation Institute; NeoteAI