Social cognition for assistive robots
Keywords: Human-robot interaction, user adaptation, emotion understanding, pair-walking model, learning from demonstration.
Overview of Research
Older adults may be unfamiliar with advanced technologies such as assistive robots. In addition, older adults with Alzheimer's disease and other dementias often have cognitive and behavioural changes that may make it more difficult to learn new skills or adapt to new technologies. This triggers the need to incorporate some social cognitive abilities into assistive robots designed for older adults with dementia so that the use of the robot is simple and intuitive. This research aims to design a human-robot interaction (HRI) architecture while exploring the social cognitive abilities commonly used in human-human interaction. Specific research focuses include the development of a user adaptation model (that will enable a robot to gradually adapt to the specific behavioural patterns, preferences, liking-disliking, etc. of an older adult with dementia and generate actions accordingly), a pair-walking model that will enable a robot to generate human-like walking patterns, e.g., walking side-by-side, leading or following the older adult, etc. depending on the context), and a natural conversation system (that will enable an older adult with dementia to convey his or her needs to the robot using natural speech).
Research Team
Momotaz Begum (Toronto Rehabilitation Institute)
Rajibul Huq (University of Toronto)
Frank Rudzicz (Toronto Rehabilitation Institute)
Rosalie Wang (Toronto Rehabilitation Institute, University of Toronto)
Alex Mihailidis (University of Toronto)
