I am finished PhD in the field of machine learning and robotics at a joint program between the Max Planck Institute and TU Darmstadt. My research interests include time series modeling and forecasting, probabilistic methods for machine learning, robotics and real time systems. During my PhD, I worked on a robot table tennis setup. My research projects include imitation learning (Learning from human demonstrations), developing the real time vision system of the robot and time series modeling of the ball and robot trajectories.
After finishing my PhD, I started to explore the field of TinyML: Machine learning on resource constrained devices such as microcontrollers or small processors. I would like to help machine learning algorithms to impact positively the life of people, with devices that are low cost but still capable of producing useful information to its users from sensors like accelerometers.
Machine learning Time Series modeling Robotics
Robot table tennis
This video depicts and example application of the robot learning method published in "Adaptation and robust learning of Probabilistic Movement Primitives". The idea is to learn a robot movement policy from a human teacher. There is no notion of goal, therefore, the robot will only send the ball to the opponent's court if the teacher does it.
Robot coffee preparation
In this video, we show an alternative application of our approach to learn robot movements from a human teacher. Namely, we use our approach to prepare some good quality Colombian coffee. We show that we can change the position of the grinder and coffee machine and our approach still works.
Our goal is to understand the principles of Perception, Action and Learning in autonomous systems that successfully interact with complex environments and to use this understanding to design future systems