TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

robotics

Shared Control Checked Whether the Robot Could Help

Intent confidence alone caused over-helping when autonomous execution was weak.

Published Updated Story ID: mp-2026-09-23-010
Read the complete editionStory JSON

Summary

Intent confidence alone caused over-helping when autonomous execution was weak.

A shared-control system paired a language model's estimate of human intent with an online estimate of a robot policy's own capability. In a 12-participant pick-and-place and stacking study, capability-aware arbitration reached 92% task success, versus 83% for manual teleoperation, 44% for intent-only arbitration and 10% for fixed equal weighting. The capability signal came from variation in sampled action trajectories. The small, specific study supports the design principle rather than a general safety guarantee.

Why it matters

Intent confidence alone caused over-helping when autonomous execution was weak.

Limits and context

  • In a 12-participant pick-and-place and stacking study, capability-aware arbitration reached 92% task success, versus 83% for manual teleoperation, 44% for intent-only arbitration and 10% for fixed equal weighting.

Key claims

  1. Intent confidence alone caused over-helping when autonomous execution was weak.

    Qualification: In a 12-participant pick-and-place and stacking study, capability-aware arbitration reached 92% task success, versus 83% for manual teleoperation, 44% for intent-only arbitration and 10% for fixed equal weighting.

    Evidence: source-2026-09-23-010

Sources

  1. arXiv preprint 2609.25369arXiv · primary research

Corrections

No corrections have been recorded for this story.