TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

robotics

Touch Tripled a Robot's Success After the Plan Was Already Made

TacPAC compared live tactile images with the contact a plan expected, lifting average success from 22 to 64 percent across five real-robot tasks.

Published Updated Story ID: mp-2026-09-08-004
Read the complete editionStory JSON

Summary

TacPAC compared live tactile images with the contact a plan expected, lifting average success from 22 to 64 percent across five real-robot tasks.

Vision-first world-action models predict before execution, while the decisive touch signal arrives during contact. TacPAC caches the contact prediction and plan representation, then lets a tactile expert compare each new tactile image with that expectation and correct only the actions not yet executed. The authors report that one correction costs 20.7 times less than regenerating the full action chunk. Across five tasks involving insertion, fragile objects, reorientation and long-horizon manipulation, the method led every task and raised average success from 22 percent for the vision-only base model to 64 percent. Those figures are specific to the reported robots, tasks and baseline.

Why it matters

TacPAC compared live tactile images with the contact a plan expected, lifting average success from 22 to 64 percent across five real-robot tasks.

Limits and context

  • TacPAC caches the contact prediction and plan representation, then lets a tactile expert compare each new tactile image with that expectation and correct only the actions not yet executed.
  • Across five tasks involving insertion, fragile objects, reorientation and long-horizon manipulation, the method led every task and raised average success from 22 percent for the vision-only base model to 64 percent.

Key claims

  1. TacPAC compared live tactile images with the contact a plan expected, lifting average success from 22 to 64 percent across five real-robot tasks.

    Qualification: TacPAC caches the contact prediction and plan representation, then lets a tactile expert compare each new tactile image with that expectation and correct only the actions not yet executed.

    Evidence: source-2026-09-08-004

Sources

  1. arXiv preprint 2609.05266arXiv · primary research

Corrections

No corrections have been recorded for this story.