In this talk, Lukas Schmidt and Axel Plinge will present their work "Can You Trust Your Autonomous Car? Interpretable and Verifiably Safe Reinforcement Learning", which was awarded the Best Paper Award at the IEEE Intelligent Vehicles Symposium 2021.
"We propose a two-step framework that allows us to learn performant, interpretable and safe driving strategies with reinforcement learning."
- Lukas Schmidt
Lukas Schmidt has been interested in robotics and autonomous systems for as long as he can remember. He has studied Mechatronics and Computer Science at the FAU Erlangen-Nuremberg, studying design, control, perception, and decision-making for autonomous systems. In his masters, he discovered his interest in explainability and visualization techniques for AI systems. Since 2020, Lukas works as a full-time research staff at Fraunhofer IIS, and researches dependable (i.e., safe and interpretable) Reinforcement Learning in his Ph.D.
- Axel Plinge
Axel Plinge received a diploma and Ph.D. degree in computer science from TU Dortmund University, Germany. He worked in different areas of psychophysical research from hearing to color vision and depth perception at the Leibniz Research Centre for Working Environment and Human Factors. He also worked self-employed, in a startup, and in industry projects spanning topics in multichannel signal processing and machine learning. In 2017, he joined the Fraunhofer IIS where he is currently heading a group working on advanced machine learning topics including dependable reinforcement learning for navigation and radio as well as quantum computing.