Biography

Felix Reinhart is Senior Expert for Machine Learning and Data Science at the Fraunhofer Institute for Mechatronic Systems Design in Paderborn. He received his diploma in computer science at Bielefeld University, Germany, in 2008. In 2011, F. Reinhart received a Ph.D. for his subsequent research at the Bielefeld Institute for Cognition and Robotics (CoR-Lab). F. Reinhart was a visiting researcher at the Advanced Robotic Controls Group at NASA JPL in Pasadena, USA, in 2009, and at Birmingham University, School of Computer Science, in 2012. Felix’ focus of research resides on machine learning, pattern recognition and robotics in industrial applications.

Abstract

Autonomous Exploration of Parameterized Motor Skills

Autonomous exploration of motor skills is a key capability of learning robotic systems. Learning of many motor skills can be formulated as inverse modeling problem, which targets at finding an inverse model that maps desired outcomes in some task space, e.g. via points of a motion, to appropriate actions, e.g. motion control policy parameters. In this talk, I present a set of approaches based on the concepts of intrinsic motivation and goal babbling for the autonomous acquisition of inverse models. Challenges and pitfalls are highlighted when rolling out these techniques to different sensorimotor coordination tasks, e.g. for learning of manipulator kinematics and dynamics, motion primitives, and speech production. The talk closes with an outlook on possible industrial applications.