Developing a VR Socially Assistive Robot Simulator Employing Game Development Tools

Long-term care facilities (LTCFs) face challenges due to the personalized care required by people with developmental disabilities. The COVID-19 pandemic exacerbated the issues associated with limited staff who are overworked. Such a scenario also disrupted the research and development of socially assistive robots (SARs) as access to care facilities and the elderly was restricted. The restriction sparked creative thinking and innovation aimed at addressing the challenges introduced by the pandemic. Such is the case for developing Aether™, a SAR designed to monitor falls and engage in playful activities with users. This paper presents the use of game technologies to develop a Virtual reality digital twin simulator for overcoming the lack of access to LTCFs and the elderly by creating synthetic data that simulates the robot's behavior, interactions with the environment, and virtual avatars, before its deployment. Our approach additionally allows overcoming the limitation with traditional datasets for training machine learning where depicted people and actions are not representative of the elderly population. Our preliminary results indicate that combining DTs and VR expedites robot development. We tested and compared the robot navigation, person detection, and inspection behavior while observing COVID-19 restrictions.