Research paper
Safety-constrained Bayesian optimization · Tuning robot parameters
The method separates performance objectives from multiple safety constraints and explores from an initially safe parameter set. Gaussian-process assumptions and context variables support constrained tuning, demonstrated in quadrotor experiments.
Reviewed 2026-11-05 · Source and bibliography checked
- Original title
- Bayesian optimization with safety constraints: safe and automatic parameter tuning in robotics
- Publisher
- Springer Science and Business Media LLC
- Authors
- Felix Berkenkamp, Andreas Krause, Angela P. Schoellig
- DOI
- 10.1007/s10994-021-06019-1
- First online publication
- 2021-06-24
- Source language
- en
- Edition
- Machine Learning; 2021; DOI 10.1007/s10994-021-06019-1
- Access
- Publisher abstract · restricted full text
- Publication status
- Published
Related resources and engineering notes
Source overview
The method separates performance objectives from multiple safety constraints and explores from an initially safe parameter set. Gaussian-process assumptions and context variables support constrained tuning, demonstrated in quadrotor experiments.
Use and review
Checked original publisher abstract and bibliography. High-probability constraint satisfaction relies on an initial safe set and regularity assumptions. The date is first online publication, distinct from the 2023 issue.
Sources & editions
Bayesian optimization with safety constraints: safe and automatic parameter tuning in robotics ↗
Springer Science and Business Media LLC · Machine Learning; 2021; DOI 10.1007/s10994-021-06019-1
Machine Learning original publisher abstract and publication metadata, https://link.springer.com/article/10.1007/s10994-021-06019-1; first online publication 2021-06-24, later issue date 2023-10.