Metraipolis › archive › dc_52e210f45c71
Towards Safe Reinforcement Learning with Reduced Conservativeness: A Case Study on Drone Flight Control — arXiv:2608.26852
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.26852
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T22:17:24.636Z
- SHA-256 of the archived text
a8669d0e3a4101fd365495a362c0a580946c5d46ad24b618521d4ce3c8331a8e- Stable URL of this page
https://metraipolis.com/documento/dc_52e210f45c71
How to cite
Towards Safe Reinforcement Learning with Reduced Conservativeness: A Case Study on Drone Flight Control — arXiv:2608.26852 — arXiv (en), https://arxiv.org/abs/2608.26852 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_52e210f45c71 (Metraipolis archive dc_52e210f45c71, sha256 a8669d0e3a4101fd365495a362c0a580946c5d46ad24b618521d4ce3c8331a8e)
BibTeX
@misc{dc52e210f45c71,
title = {Towards Safe Reinforcement Learning with Reduced Conservativeness: A Case Study on Drone Flight Control — arXiv:2608.26852},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.26852},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_52e210f45c71; sha256 a8669d0e3a4101fd365495a362c0a580946c5d46ad24b618521d4ce3c8331a8e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26852. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Incorporating formal methods into reinforcement learning (RL) has the potential to result in the best of both worlds, combining the robustness of formal guarantees with the adaptability and learning capabilities of RL, though careful design is needed to balance safety and exploration. In this work, we propose a framework to mitigate this loss of exploration while still allowing for the safety of the system to be ensured. Specifically, we introduce a less restrictive method that can reduce the conservativeness of formal methods by refining a disturbance model using online collected data and it evaluates the safety of a learning-based controller, using computationally efficient zonotopic reachability analysis for the safety analysis to facilitate a real-time implementation. We validate the framework in a real-world drone flight through a canyon, where the drone is subjected to unknown external disturbances and the framework is tasked with learning those disturbances online and adjusting the safety guarantees accordingly. The results show that the framework enables a less restrictive online training of learning-based controllers without compromising the safety of the system. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-08-28 in eess.SY). Categorie arXiv: Systems and Control (eess.SY). Pagina del lavoro: https://arxiv.org/abs/2608.26852 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T22:17:01.662Z dalla pagina https://arxiv.org/catchup/eess.SY/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).