Collaboration of an assistive robot and older adults with dementia

Objective Assistive robots that help older adults with dementia (OAwDs) to carry out daily activities or socially interact with them need to be designed to account for their unique interaction patterns and cognitive limitations. We report on some of the challenges and recommendations involved in the design of a collaborative human-robot interaction (HRI) framework for assistive robots that help OAwDs in daily activities.  Method An observational study was conducted with 10 OAwDs and their familial caregivers in a simulated home environment. OAwDs completed a tea-making task in collaboration with a tele-operated assistive robot. The robot guided each OAwD to the kitchen, provided assistance (through prompting and answering questions), engaged in social conversation when appropriate, and guided the OAwD back to the caregiver. Participants were interviewed to solicit their opinions about the interaction. All study activities were video recorded.  Results The recorded robot-OAwD interactions and interviews were analyzed using behavioral coding techniques to identify specific interaction behaviors of OAwDs that may present challenges to the design of a collaborative HRI framework. Reported are five different behaviors (interactiveness, natural dialogue, team behavior, emotion, cognition and memory) that relate to four key human factors (trust, grounding, communication, and situational awareness). Preliminary recommendations for future development include conducting a large study to better understand the needs of the target population and designing an intelligent robot control interface capable of working collaboratively with OAwDs.  Conclusion Design of a collaborative HRI framework may require intelligent modifications to existing machine learning, speech recognition, and natural language processing algorithms.