PhD Defense: Learning Motion Skills for a Humanoid Robot

6 September 2023

Abstract: Creating motion skills for humanoid robots is difficult due to the complexity and unstable nature of this robot type. This work investigated two different approaches to create skills for various humanoids. One approach is based on Cartesian quintic splines with optimized parameters. The other one is a reinforcement learning-based approach that utilizes the first approach as a reference motion. A special focus was set on the locomotion skill, but a standing-up and a kick skill are also discussed.

Date and Time: Wed, 06 September 2023, 16:30 CET.

Room F-334, Informatikum Campus, Vogt-Kölln-Str. 30, 22527 Hamburg.