Robotic Social Interaction with Dementia Patients
The content of this page is provided by Julian Lars Striegl – Freiherr von Hoverbeck genannt von Schoenaich. All rights reserved.
With the mean population age rising, the number of dementia patients also increases strongly every year. The elderly and especially people with dementia often experience social isolation and need assistance while performing activities of daily living. The research on the use of assistance robots to tackle this problem by providing social assistance is fairly new but has already shown some promising results. However, the interaction with assistance robots can be challenging for seniors – who often lack the technical experience – and for cognitive impaired users. Therefore, a natural language interface could lower the hurdle of interacting with a robotic system for the elderly and people with dementia. This thesis investigates the social interaction between people with dementia and assistance robots, focusing on verbal interaction. As a scientiﬁc foundation, the eﬀects of social interaction, communication strategies with dementia patients and the state of research on voice user interfaces for the targeted user group are examined. The outcome of related studies is analyzed and used to create the conceptual design of a social assistance system through a user-centered design process. Based on the design a prototypical system is implemented using the Loomo Robotic Platform and the Alexa Voice Service Software Development Kit. The prototype is evaluated through a pilot study, focusing on the feasibility, acceptance and usability of a social assistance robot with a voice user interface designed for people with dementia. The results of the pilot study seem promising and indicate a high feasibility, usefulness and acceptance of the system. The assessment of the usability of the system shows good results but also suggests some measures to further improve the system’s interface. Those measures are consequently discussed to conclude the thesis.
You can download the full submission as PDF file (3 MB) here.