robotics
More Taps Made the Robot More Confident—and Wrong
An audit traced insertion failures to a 2.1-millimeter boundary error in the observation model.
Summary
An audit traced insertion failures to a 2.1-millimeter boundary error in the observation model.
In a probe-then-commit manipulation pipeline, extra taps sharpened the robot's belief even as the truth left its support in 16.9% of simulated episodes. Conformal calibration restored statistical coverage but did not stop confidently wrong actions from passing a confidence gate. Separate audits of belief coverage and failure prediction pointed to a 2.1-millimeter tap-boundary error; correcting it cut failure from 0.354 to 0.112 on untouched instances and transferred to another engine. Physical insertion trials reproduced the two audit conditions, though the framework remains tied to one-shot decisions with inspectable models.
Why it matters
An audit traced insertion failures to a 2.1-millimeter boundary error in the observation model.
Limits and context
- Conformal calibration restored statistical coverage but did not stop confidently wrong actions from passing a confidence gate.
Key claims
An audit traced insertion failures to a 2.1-millimeter boundary error in the observation model.
Qualification: Conformal calibration restored statistical coverage but did not stop confidently wrong actions from passing a confidence gate.
Evidence: source-2026-09-28-011
Sources
- arXiv preprint 2609.30608arXiv · primary research
Corrections
No corrections have been recorded for this story.