Data-Driven Architecture to Encode Information in the Kinematics of Robots and Artificial Avatars

Abstract

We present a data-driven control architecture designed to encode specific information, such as the presence or absence of an emotion, in the movements of an avatar or robot driven by a human operator. Our strategy leverages a set of human-recorded examples as the core for generating information-rich kinematic signals. To ensure successful object grasping, we propose a deep reinforcement learning strategy. We validate our approach using an experimental dataset obtained during the reach-to-grasp phase of a pick-and-place task.