AI Science Tracker
EN

Detecting and Explaining Robot Policy Failures from Sparse Internal Features

Affiliation not stated in the reviewed material · 7 Nov 2026 · computer science

Discovery

Detecting and Explaining Robot Policy Failures from Sparse Internal Features

Source check: Source checked. Not independent scientific confirmation.

Finding: MIND uses sparse-autoencoder features of visuomotor policies to detect and describe failures. The authors report evaluations in simulation and on a physical robot; results are bounded to tested executions and do not establish general deployment reliability.

Technology: MIND sparse-autoencoder feature analysis on visuomotor policies

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

AI contribution
MIND maps policy activations to sparse features associated with robot state and surroundings, then uses feature deviations to detect and describe failures.
Test and result
The study evaluated sparse-autoencoder features for policy-failure detection and diagnosis in simulation and on a physical robot, including held-out executions and ablations.
Limitation / still unknown
Physical-robot and simulation results are limited to the tested tasks and setup; general deployment performance is unknown.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/html/2611.10400v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse computer science discoveries.

Related catalog entries

Links use documented relationships where available, plus shared catalog context. A shared topic or model family does not establish a scientific dependency.