aiqbench analysis
This research proposes using calibrated relational world models to detect semantic mismatches in robot actions, such as inverted axes or reversed gripper polarity, by treating learned state-action relations as offline diagnostic contracts. Experimental results on LIBERO Spatial tasks demonstrate that joint Gaussian mixture modeling significantly outperforms ridge regression in both recall and localization accuracy for identifying these specific fault types. The study concludes that this approach serves as a retrospective diagnostic tool rather than a guarantee for real-time deployment safety or cross-robot generalization.
Detecting Silent Action-Interface Mismatch with Calibrated Relational World Models
arXiv:2611.06653v1 Announce Type: new Abstract: A robot action can satisfy shape and range constraints yet carry the wrong semantics: axes may be permuted, signs inverted, rotations rescaled, or gripper polarity reversed. We study learned state-action-effect relations as offline contracts for detecting these mismatches and ranking their likely fields. Clean held-out transitions calibrate alarms; a fixed inverse vocabulary supports closed-set localization. On ten LIBERO Spatial tasks comprising 432 source episodes, a capacity-controlled comparison at 2.5% empirical clean false-positive rate gives 46.1% recall for a ridge forward residual, 61.0% for a four-component joint Gaussian mixture, and 65.7% for the exact conditional of that same mixture. The joint score retains a 14.9 percentage-point advantage over ridge, while conditioning improves recall by a further 4.7 points. Joint scoring instead has higher localization: 95.8% versus 94.6%. These matched-rate comparisons are retrospective diagnostics, not deployment calibration guarantees. Earlier Object/Goal results broaden tasks within the LIBERO convention; DROID, all three HYDRA tasks, short temporal faults, small offsets, exact repair, and simulated mitigation fail their respective gates, while xArm contributes one task only. The evidence supports a bounded offline diagnostic and a detection-localization tradeoff, rather than joint-density superiority or cross-robot safety.
Abstract:A robot action can satisfy shape and range constraints yet carry the wrong semantics: axes may be permuted, signs inverted, rotations rescaled, or gripper polarity reversed. We study learned state-action-effect relations as offline contracts for detecting these mismatches and ranking their likely fields. Clean held-out transitions calibrate alarms; a fixed inverse vocabulary supports closed-set localization. On ten LIBERO Spatial tasks comprising 432 source episodes, a capacity-controlled comparison at 2.5% empirical clean false-positive rate gives 46.1% recall for a ridge forward residual, 61.0% for a four-component joint Gaussian mixture, and 65.7% for the exact conditional of that same mixture. The joint score retains a 14.9 percentage-point advantage over ridge, while conditioning improves recall by a further 4.7 points. Joint scoring instead has higher localization: 95.8% versus 94.6%. These matched-rate comparisons are retrospective diagnostics, not deployment calibration guarantees. Earlier Object/Goal results broaden tasks within the LIBERO convention; DROID, all three HYDRA tasks, short temporal faults, small offsets, exact repair, and simulated mitigation fail their respective gates, while xArm contributes one task only. The evidence supports a bounded offline diagnostic and a detection-localization tradeoff, rather than joint-density superiority or cross-robot safety.
Accepted at the CoRL 2026 Workshop on Modeling Uncertainty in Robotic World Models (UncertainWM). 20 pages, including 12 pages of supplementary material
Robotics (cs.RO); Machine Learning (cs.LG)
arXiv:2611.06653 [cs.RO]
(or arXiv:2611.06653v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2611.06653 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 230 words