Learning Safe-by-Design Neural Network Controllers
Yang Zhao, Jungeun Lee, Jeong hwan Jeon, Sze Zheng Yong · 2026
Read Learning Safe-by-Design Neural Network Controllers by Yang Zhao, Jungeun Lee, Jeong hwan Jeon, Sze Zheng Yong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-26534.txt.
Description
Yang Zhao and three colleagues learn neural network controllers that are safe by design, without bolting on a hand-tuned safety filter afterward. Given a valid control barrier function, they jointly train the controller and barrier parameters set by a neural network. Affine safety constraints hold by construction, with no online quadratic-program filter at run time. A lightweight projection design enforces constraints without full constraint enumeration, improving efficiency and scale. Extensive simulations show reliable, scalable safety at reduced computational cost.
Catalog record
| CREATOR | Yang Zhao, Jungeun Lee, Jeong hwan Jeon, Sze Zheng Yong |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35285 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.26534