Expressive states with a robot arm using adaptive fuzzy and robust predictive controllers

In order to convey a sense of believability in social contexts, new interactive robots may be able to express dynamic expressive states and to adapt to different situations. This work presents an innovative adaptive control system architecture for generating expressive motions in a robot arm. An adaptive Fuzzy controller to map environmental inputs such as temperature, humidity, luminosity and human proximity to values of the PAD emotional model (Pleasure, Arousal, and Dominance) is proposed. The PAD values are used to change the performance of the robot trajectories designed to express different features. These robot motions are commanded by the Robust Generalized Predictive Controllers (RGPC) using convex optimization by Youla parametrization that involves the robot regulation with the adaptive motion. The proposed approach allows to generate personalized and dynamic behaviors for Human-Robot Interaction (HRI) in non-humanoid robots.