On February 7 and 8, 2012, Dr. Freek Stulp visited PAL robotics as part of an ongoing academic collaboration. Dr. Stulp is an assistant professor at ENSTA ParisTech, and his research focuses on using motion primitives and direct reinforcement learning for robust and flexible manipulation in human environments.
The current collaboration aims at comparing two approaches for generating motion plans for household task execution: Dynamic Movement Primitives (DMPs) and single-query probabilistic motion planners. These approaches complement each other very well, and it is our aim to take a first step in determining the conditions under which they should be best used.
Project participants are shown in the accompanying picture (from left to right: PAL’s Adolfo Rodríguez, interns Hilario Tomé and Carmen Lopera, and Freek Stulp).