ReTac-ACT: A State-Gated Vision-Tactile Fusion Transformer for Precision Assembly

TL;DR

ReTac-ACT is a novel vision-tactile fusion transformer that significantly improves precision in robotic assembly tasks by dynamically integrating visual and tactile feedback, especially under occlusion conditions.

Contribution

The paper introduces ReTac-ACT, a new vision-tactile policy with bidirectional cross-attention, a gating network, and tactile reconstruction, advancing robotic precision assembly capabilities.

Findings

Achieves 90% success rate on peg-in-hole tasks.

Maintains 80% success at 0.1mm clearance.

Outperforms vision-only and baseline methods.

Abstract

Precision assembly requires sub-millimeter corrections in contact-rich "last-millimeter" regions where visual feedback fails due to occlusion from the end-effector and workpiece. We present ReTac-ACT (Reconstruction-enhanced Tactile ACT), a vision-tactile imitation learning policy that addresses this challenge through three synergistic mechanisms: (i) bidirectional cross-attention enabling reciprocal visuo-tactile feature enhancement before fusion, (ii) a proprioception-conditioned gating network that dynamically elevates tactile reliance when visual occlusion occurs, and (iii) a tactile reconstruction objective enforcing learning of manipulation-relevant contact information rather than generic visual textures. Evaluated on the standardized NIST Assembly Task Board M1 benchmark, ReTac-ACT achieves 90% peg-in-hole success, substantially outperforming vision-only and generalist baseline methods, and maintains 80% success at industrial-grade 0.1mm clearance. Ablation studies validate that each architectural component is indispensable. The ReTac-ACT codebase and a vision-tactile demonstration dataset covering various clearance levels with both visual and tactile features will be released to support reproducible research.