Learning flexible, multi-modal human-robot interaction by observing human-human-interaction

Abstract

This paper presents a technique to learn flexible action selection in autonomous, multi-modal human-robot interaction (HRI) from observing multi-modal human-human interaction (HHI). A model is generated using the proposed technique with symbolic states and actions, representing the scope of the observed mission. Variations in human behavior can be learned as stochastic action effects while execution time perception noise is taken into account, using likelihood models. During execution, the model is used for dynamic action selection in HRI situations. The model as well as the evaluation system integrate the interaction elements of spoken dialog, human body configuration and exchanged objects. The technique is evaluated on a multi-modal service robot which is both able to observe the demonstration of two humans as well as execute the generated mission autonomously.

Partially Observable Markov Decision Processes in Robotics: A Survey

“…Other tasks where POMDP based HRI has been studied include human-robot social interaction [130], [26], medical diagnosis [113], and assistive tutoring of students [123]. In human-robot social interaction, [26] uses a POMDP to optimize robot social interaction in the task of yielding or going first in a driving simulator.…”

“…The dynamics model is learned from simulations. In other work, [130] investigates human-robot social interaction by observing human-human-interaction. Contrary to typical POMDP based HRI, in [113] the POMDP decides on the next task in autism spectrum disorder (ASD) diagnosis instead of on high or low level actions.…”