| title | layout | header | excerpt | classes | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Jakob Thumm |
single |
|
Teaching robots to safely work with humans. |
wide |
{: .align-left} I'm a Ph.D. student of Robotics at the Technical University of Munich (TUM) with the Cyber-Physical Systems Group. My passion for technology and its potential to positively impact people's lives drives me to research and develop solutions for safe human-robot interactions. I am fascinated by the potential of reinforcement learning as a powerful tool for robots to gain experience about the world and learn from it, and especially interested in exploring how we can ensure safety for these reinforcement learning agents in human-robot environments.
I have set out a range of research questions that I am determined to explore in my Ph.D. studies, including:
- We let the robots out of their cages, what now?
- How can we ensure safety for reinforcement learning agents (and other AI-agents) in human-robot environments?
- How do we design our safety methods so that they interfere with the agent as little as possible while still being provably safe?
- Can we teach a robot to dynamically react to human behavior, e.g., evade the human?
- What is needed to create a robot that humans like to work with?
- What could be fun ways to teach reinforcement learning to robotics students?
I believe that by answering these questions, we can create a future where robots and humans can work together seamlessly and efficiently.