Papers

8

Total Citations

477

H-Index

8

About

Felix Berkenkamp is a prominent researcher at the intersection of safe machine learning, reinforcement learning, and robotics, whose work addresses one of the field's most pressing challenges: enabling autonomous systems to learn efficiently while guaranteeing safety. His research has made foundational contributions to safe exploration in reinforcement learning, demonstrating how Gaussian processes and Bayesian optimization can be harnessed to prevent unsafe actions during learning — a critical requirement for real-world robotic applications. His 2016 paper on safe exploration in finite Markov decision processes (68 citations) laid important theoretical groundwork, while his work on Bayesian optimization with safety constraints (80 citations) translated these ideas into practical parameter tuning frameworks for robotics. Berkenkamp has also advanced stability certification for learned controllers, notably through the Lyapunov Neural Network approach (69 citations), which provides formal safety guarantees for adaptive systems. His innovative strategy of intelligently trading off simulations and physical experiments (111 citations) has proven highly influential in making reinforcement learning more sample-efficient and deployable. With over 470 cumulative citations across his key works, Berkenkamp's research offers indispensable tools for researchers seeking to bridge the gap between theoretical safety guarantees and practical autonomous systems.

Research Focus

Key Achievements

8
H-Index
8
Papers
477
Total Citations
60
Avg Citations/Paper
🏆 Most Cited Paper
Virtual vs. real: Trading off simulations and physical experiments in reinforcement learning with Bayesian optimization
111 citations · 2017
📈 Most Prolific Year: 2016 (2 Papers)
🤝 Key Collaborators: 16
🏛 Institutions: Board of the Swiss Federal Institutes of Technology, ETH Zurich, Robert Bosch (Germany)

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 13 days ago