Felix Berkenkamp portrait

I am a research scientist at the Bosch Center for AI. Previously, I completed my PhD at ETH Zurich, for which I received the ELLIS PhD award. My PhD advisors were Andreas Krause and Angela Schoellig (University of Toronto.

My goal is to enable dynamic systems to safely and autonomously learn in uncertain real-world environments. This requires new reinforcement learning algorithms that respect the physical limitations and constraints of dynamic systems and provide theoretical safety guarantees.

I held a AI fellowship from the Open Philanthropy Project, was an Associated Fellow at the Max Planck ETH Center for Learning systems and a postgraduate affiliate at the Vector institute. Previously I was the Workflow Co-chair for ICML 2018 and completed research internships at Microsoft Research and Deepmind.

Talks and Lectures

Guest lecture on Safe Bayesian Optimization for CS 159: Data-Driven Algorithm Design at Caltech.

ETH Day 2018 short presentation (in German)

Invited talk at the Workshop on Reliable AI 2017

NIPS/CoRL 2017: "Safe Model-based Reinforcement Learning with Stability Guarantees".

CDC 2016: "Safe learning of regions of attraction for nonlinear, uncertain systems with Gaussian processes".

ICRA 2016: "Safe Controller Optimization for Quadrotors with Gaussian Processes".

Research videos

ICRA paper: "Safe Controller Optimization for Quadrotors with Gaussian Processes".

ECC paper: "Safe and robust learning control with Gaussian processes"

NIPS paper: "Safe exploration in finite Markov Decision Processes with Gaussian processes".

ICRA paper: "Virtual vs. Real: Trading Off Simulations and Physical Experiments in Reinforcement Learning with Bayesian Optimization".

NeurIPS paper: "Safe exploration in finite Markov Decision Processes with Gaussian processes".