Research paper
DADEE: Well-calibrated uncertainty quantification in neural networks for barriers-based robot safety
DADEE combines model-variance and direct-error estimation to improve uncertainty estimates inside and outside the training domain. Simulated robot failures are used to compare CBF controllers. The result concerns uncertainty calibration and simulation behavior, with hardware and distribution-shift validation still needed.
Reviewed 2026-11-05 · Source and bibliography checked
- Original title
- DADEE: Well-calibrated uncertainty quantification in neural networks for barriers-based robot safety
- Publisher
- arXiv
- Authors
- Ataei, Masoud, Dhiman, Vikas
- DOI
- 10.48550/arxiv.2407.00616
- First archive posting
- 2024-06-30
- Source language
- en
- Source country
- US
- Edition
- v1
- Access
- Open archive metadata and manuscript
- Publication status
- Preprint
Related resources and engineering notes
Source overview
DADEE combines model-variance and direct-error estimation to improve uncertainty estimates inside and outside the training domain. Simulated robot failures are used to compare CBF controllers. The result concerns uncertainty calibration and simulation behavior, with hardware and distribution-shift validation still needed.
Use and review
Original summary of the primary archive abstract and bibliographic record. Specialist technical and native-language review remain pending. The reported method does not establish a deployed safety function or achieved PL.
Sources & editions
DADEE: Well-calibrated uncertainty quantification in neural networks for barriers-based robot safety ↗
arXiv · v1
arXiv abstract, authors, submission history and DOI links; first posting 2024-06-30, abstract reviewed in v1. Publisher dates are separate.