Date of Award


Document Type

Open Access Dissertation


Computer Science and Engineering

First Advisor

Jason M O'Kane


This research delivers a new, interactive game-playing robot named CHARLIE and a novel technique for remotely detecting breathing and heart rate using a single-point, thermal infrared sensor (IR). The robot is equipped with a head and two arms, each with two degrees of freedom, and a camera. We trained a human hands classifier and used this classifier along with a standard face classifier to create two autonomous interactive games: single-player ("Imitate Me, Imitate You") and two-player ("Pass the Pose"). Further, we developed and implemented a suite of new interactive games in which the robot is teleoperated by remote control. Each of these features has been tested and validated through a field study including eight children diagnosed with autism and speech delays. Results from that study show that significant improvements in speech and social skills can be obtained when using CHARLIE with the methodology described herein. Moreover, gains in communication and social interaction are observed to generalize from child-to-robot to co-present others through the scaffolding of communication skills with the systematic approach developed for the study. Additionally, we present a new IR system that continuously targets the sub-nasal region of the face and measures subtle temperature changes corresponding to breathing and cardiac pulse. This research makes four novel contributions: (1) A low-cost, field-tested robot for use in autism therapy, (2) a suite of interactive robot games, (3) a hand classifier created for performing hand detection during the interactive games, and (4) an IR sensor system which remotely collects temperatures and computes breathing and heart rate. Interactive robot CHARLIE is physically designed to be aesthetically appealing to young children between three and six years of age. The hard, wood and metal robot body is covered with a bright green, fuzzy material and additional padding so that it appears toylike and soft. Additionally, several structural features were included to ensure safety during interactive play and to enhance the robustness of the robot. Because children with autism spectrum disorder (ASD) often enjoy exploring new or interesting objects with their hands, the robot must be able to withstand a moderate amount of physical manipulation without causing injury to the child or damaging the robot or its components. CHARLIE plays five distinct interactive games that are designed to be entertaining to young children, appeal to children of varying developmental ability and promote increased speech and social skill through imitation and turn-taking.

Remote breathing and heart rate detection Stress is a compounding factor in autism therapy which can inhibit progress toward specific therapeutic goals. The ability to non-invasively detect physical indicators of increasing stress, especially when they can be correlated to specific activities and measured in terms of length and frequency, can relay important metrics about the antecedents that cause stress for a particular child and can be used to help automate the evaluation of a child's progress between sessions. Further, collecting and measuring critical physiological indicators such as breathing and heart rate can enable robots to adjust their behavior based on the perceived emotional, psychological or physical state of their user. The utility and acceptance of robots can be further increased when they are able to learn typical physiological patterns and use these patterns as a baseline for identifying anomalies or possible warning signs of various problems in their human users. We present a new technique for remotely collecting and analyzing breathing and heart rates in real time using an autonomous, low cost infrared (IR) sensor system. This is accomplished by continuously targeting a high precision IR sensor, tracking changes in the sub-nasal skin surface temperature and employing a sinusoidal curve-fitting function, Fast Fourier Transform (FFT), and Discrete Wavelet Transform (DWT) to extract the breathing and heart rate from recorded temperatures.