Research Statement
New areas such as humanoid and service robotics require that modern robots need to act in unstructured environments where the robot system needs to be capable of adapting to its surroundings. As a result, methods from machine learning will increasingly be needed in order to create more autonomous robots. My research centers around robot learning approaches for the physical interactions with objects. Here, grasping and manipulation of unmodeled, novel objects remains one of the key problems in order to bring robots into human-inhabited environments. In order to find suitable solutions to such manipulation problems, we develop domain-oriented machine learning approaches. These include active learning of vision-based grasping, reinforcement learning for preshaping of grasping movements as well as learning long-term plans for object manipulation.

Current Position
Ph.D. student at the Technische Universitaet Darmstadt
Topic: Learning Grasping and Manipulation for Robotics
Supervisor: Dr. Jan Peters
Dept.: Intelligent Autonomous Systems