Research paper
SafeOpt tuning · Improving controllers above a performance threshold
Starting from an initial controller, SafeOpt explores parameters predicted to remain above a performance threshold with high probability. Quadrotor experiments evaluate automatic tuning under that constraint.
Reviewed 2026-11-05 · Source and bibliography checked
- Original title
- Safe controller optimization for quadrotors with Gaussian processes
- Publisher
- IEEE
- Authors
- Felix Berkenkamp, Angela P. Schoellig, Andreas Krause
- DOI
- 10.1109/icra.2016.7487170
- Publication date
- 2016-05
- Source language
- en
- Edition
- 2016 IEEE International Conference on Robotics and Automation (ICRA); 2016-05; DOI 10.1109/icra.2016.7487170
- Access
- Publisher metadata and open author manuscript
- Publication status
- Published
Related resources and engineering notes
Source overview
Starting from an initial controller, SafeOpt explores parameters predicted to remain above a performance threshold with high probability. Quadrotor experiments evaluate automatic tuning under that constraint.
Use and review
Author abstract and bibliography checked. The performance threshold and GP assumptions need validation; high-probability constraints do not mean zero risk.
Sources & editions
Safe controller optimization for quadrotors with Gaussian processes ↗
IEEE · 2016 IEEE International Conference on Robotics and Automation (ICRA); 2016-05; DOI 10.1109/icra.2016.7487170
ICRA2016 author abstract https://arxiv.org/abs/1509.01066 and https://las.inf.ethz.ch/files/berkenkamp16safe.pdf.