I am motivated every day by leaving behind a world in which humans and machines work together and understand each other. They adapt to each other's needs and perform tasks that they could not do alone.
Currently, I am pursuing my PhD in human-machine interaction in the context of trajectory planning of autonomous mobile robots. My research focuses on machine learning methods and, in particular, on reinforcement learning for motion planning of autonomous vehicle in pedestrian rich environments.
In 2022, I completed my master's degree in electrical engineering and information technology with a focus on control engineering at the Karlsruhe Institute of Technology (KIT) and wrote my master's thesis in a research collaboration at the University of Waterloo in Canada. During my studies, I took part in the Formula SAE international engineering competition and managed the electrical department in my second year. I was then able to use this experience in my internship and my Bachelor's thesis at Mercedes-AMG.
The Pop in Your Job
The great thing about the FZI is that we have the time and flexibility to work on our personal research goals and incorporate them into our projects. We work in a very interdisciplinary and cross-departmental way in a young, committed and highly motivated team, in which you gain new perspectives and inspiration every day, which are invaluable for your own work and personality. Everyone is passionate about their work, pulls together and is a team player. In applied research, we not only create theoretical foundations but also bring these state-of-the-art methods closer to industrial application and make them accessible to the community.
A | Artificial Intelligence & Machine Learning Stream | Case Study
Monday, March 17
02:30 pm - 03:00 pm
Live in Berlin
Less Details
Navigating a vehicle in a pedestrian-rich and varying environment is a big challenge, even for humans, which is not always mastered satisfactorily for everyone involved. This complex navigation and decision-making task requires an understanding of human behavior. It must consider the interaction between the movements of the vehicle and that of people in the vicinity. Accordingly, a technical realization for such a navigation algorithm requires engineering skills and sociological and psychological components to consider the human factor adequately.
In the current landscape, cutting-edge machine learning techniques, such as DRL, are employed to learn a motion planning algorithm among pedestrians. These approaches include social collision avoidance, where people are treated as dynamic objects, and socially aware navigation, where people and their group dynamics are considered. Despite these advances, a critical gap remains: AVs cannot seamlessly adapt their navigation behavior to that of humans, which prevents them from large-scale use and acceptance by humans.
In FZI’s novel approach, they bridge this gap through socially integrated navigation. This innovative method is derived from a sociological definition of social acting. It empowers AVs to adapt their navigation behavior with that of people, leveraging social cues and explicitly considering the dynamics of human-machine interaction. With this approach, AVs act not only socially but also socially acceptable while taking the individual nature of humans into account.