Abstract

A highly relevant task in health care is to support the independency of persons with dementia, under consideration of the individual mental state and capabilities. Inappropriate care strategies could cause fast increase of care dependency. We understand that social assistive robots will support caregivers and persons with dementia at the same time, having the potential to support self-care, independency and the well-being of the elderly.

The objective of the Austrian research project AMIGO is to develop an entertaining and motivating multisensory robot coach on the basis of the social robot Pepper and the playful training app amicasa. The goal is to empower persons with dementia to perform daily stimulating activities in a concept of integrated multimodal intervention. The robot will additionally be capable to ask about the health status, remind important events or tasks, involve persons with dementia into a dialogue, and provide personalized entertainment like reading news of interest from all over the world. A research objective is to adjust Pepper’s dialogue and motivation style based on emotional feedback sensed in interaction. The system will motivate to perform personalized exercises, to stimulate cognitive processes and physical activities, and to maintain and extend social bonds.

Pepper’s head camera is used for motion analysis. Pepper with its motion and interaction repertoire offers affordances for entertainment and state-of-the-art video analysis will offer an interpretation of locomotion specific behavioral parameters. Eye tracking sensors within the Tablet PC enable eye movement measurements for the analysis of executive function parameters, for early prevention and early neuropsychological warning procedures. Easily usable interfaces enable planning and autonomous daily practice to formal as well as to the informal caregiver in a weekly rhythm so that people with dementia can stay longer at home and the progress of dementia is slowed down.

The overall project framework with system architecture and health care logistics are presented. First results gained with the robot camera and deep learning analysis are very promising since we expect to robustly extract motion features that will enable to determine energy and impulse of the user during physical training. Furthermore, a qualitative study with content analysis was carried out to investigate the feelings, knowledge and the utility of social robots. Individual interviews and focus groups with potential users - persons with dementia, relatives, dementia trainers and caregivers - were performed . The major benefit for a personal use of a robot, like Pepper, is seen for the categories Communication/Contacts with others, Recreational Activities, Avoiding Dangers and Daily Activities. The results will support the further development of Pepper for the nursing care practice.

AMIGO is a novel motivational framework for the empowerment of persons with dementia for multimodal intervention. First results demonstrate that users are inclined to accept and welcome the social robot methodology as a social and motivational modality.