Tether: Alleviating Loneliness with Assistive Robotic Social Media Dialog

点击翻译按钮获取中文摘要

Loneliness affects 40% of Canadians age 65 or older, particularly under-represented older adults. By 2025, that represents 16 million Canadians. Voice assistants and assistive robots are an emerging healthcare home technology (expected to reach $26 billion USD by 2030). Our interactions with these assistants are currently limited to short conversations about functional tasks. "Tether" is a research effort to build assistive robotic social media dialog technologies to reduce loneliness. To do this, Tether develops cloud computing, voice and robot software and hardware to alleviate loneliness and social isolation by connecting older adults to social media content. My main objective is to create low-cost, speaker-size robots that speak the way we do, rather than making us speak like robots, to alleviate loneliness. Instead of robots using repetitive scripts that don't support long-term interaction, I propose using social media to enable personally relevant, flexible, ongoing, multi-response conversations without repetitive question-and-answers. Social media can provide the backbone of data about relationships and conversation topics so that these systems can talk directly about people in our lives - in contrast to the recent trend of conversational agents that talk about functional tasks like playing music. Therefore, Tether's objective is an integrated system architecture and protocol for building social media assistive robots that employ perception, reasoning and artificial intelligence, while addressing loneliness. I shall use microphone arrays and human-robot interaction data to classify user activity and detect two types of loneliness. I propose to develop a dialog system for naturalistic casual conversations using a hybrid recommendation engine and expert model to pick what social media messages to show lonely people, then use machine learning of human casual conversation with a mutual information model that rewards novelty to determine how to say it. Furthermore, whereas there are social concerns about older adults becoming "friends" with a robot, my research program balances this by building intelligent robots that have social mediation roles. The societal goal of the technology is to learn how to naturally mediate social data to connect people with their communities. Today, it's important for people to be able to take control of their social lives and how they use technology. Thus, this program directly involves citizens with loneliness from under-represented groups in the robot development process and integrates social media with robotics to also fit the habits of today's younger generations. This program culminates in delivering fully autonomous, low-cost proof-of-concept household robots to three gender-based-analysis (SGBA+) populations: older adults with loneliness, minority 60+ers with loneliness and lonely young adults, as part of a longitudinal controlled experiment.