Research paper
Safe learning in robotics · Control and reinforcement learning survey
A review connects model uncertainty, learning-based control, reinforcement learning and methods for checking learned policies. It helps organize the assumptions and evidence needed when a learned controller acts on a robot.
Reviewed 2026-11-04 · Source and bibliography checked
- Original title
- Safe Learning in Robotics: From Learning-Based Control to Safe Reinforcement Learning
- Publisher
- arXiv
- Authors
- Lukas Brunke, Melissa Greeff, Adam W. Hall, Zhaocong Yuan, Siqi Zhou, Jacopo Panerati, Angela P. Schoellig
- DOI
- 10.48550/arXiv.2108.06266
- Publication date
- 2021-08-13
- Source language
- en
- Source country
- US
- Edition
- arXiv:2108.06266v2
- Access
- open_archive_metadata_and_manuscript
- Publication status
- Preprint
Source overview
A review connects model uncertainty, learning-based control, reinforcement learning and methods for checking learned policies. It helps organize the assumptions and evidence needed when a learned controller acts on a robot.
Use and review
This record describes the arXiv manuscript. The displayed date is its first archive posting. A survey and its reviewed methods require application-specific evidence before use in a robot safety function.
Sources & editions
Safe Learning in Robotics: From Learning-Based Control to Safe Reinforcement Learning ↗
arXiv · arXiv:2108.06266v2
arXiv abstract, author metadata and submission history. First posted 2021-08-13; v2 dated 2021-12-06.