M.S. (Master of Science)
Department of Electrical Engineering
The ultimate goal of advanced child-robot interaction is to develop a high-level intelligencecommunication channel in a closed-loop configuration with the child as the focal point of the instructional scenario. This goal can be met by establishing efficient sensing mechanisms on the robot’s side, such as automatic engagement measuring, which will allow the robot to alter its behavior or even the instructional scenario’s execution.Recent researches, regarding engagement recognition systems mostly based on facial or multi-modal features.We would like to investigate this topic using prosodic cues only. We believe speech characteristics are more likely to be related to child speaker engagement level.This thesis proposes a pipeline to account for the child’s interaction with the robot. This pipeline relies on two features: pitch and intensity. These features have been established as significant predictors of child-robot engagement and can be utilized to modify the behavior of a robot.We conduct analysis of speech cues to gain insights tying specific prosodic patterns to the phenomena of engagement. While the performance of our model varies based on task setting and individual interaction, we discover that there are universal engagement-related prosodic patterns.
Oztoprak, Mustafa, "Engagement Detection Using Prosodic Cues: An Approach For Measuring Child Emotional Engagement Level in Child-Robot interaction Scenario" (2022). Graduate Research Theses & Dissertations. 7519.
Northern Illinois University
Rights Statement 2
NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.