In our ageing society the gap between people in need of care and those who are able to provide it is predicted to widen in the decades to come. Employing assistive technologies, such as assistance robots, is one approach among many to cope with this problem. Assistance robots can ease the workload on care staff and help elderly individuals to remain independent in their own homes for longer.
When developing assistance robots for elderly people, many requirements have to be considered. The robot must for instance be easy to use, affordable and useful. In order to provide additional value to its users, it is beneficial or even crucial for a robot to be multifunctional and adaptive. There are different approaches to adaptivity. This thesis considers how a robot can proactively offer suggestions for activities that are adapted to the current context in order to support elderly people with their daily activities (ADL). For instance, the robot might suggest to turn off the lights in the context of the activity ‘Going to Bed’ or offer to document the intake of medication.
In this thesis a concept for a recommender system that derives a suggestion from the current context is proposed. The first step in that concept includes gathering context data from different sources into a context database. The most current data is retrieved on a regular basis and used as input for a rule-based classification module to generate an activity recommendation that suits the current situation. In the last step the robot communicates the suggested activity to the user.
The feasibility of the concept is shown by implementing a prototype in the form of an Android app, while the technology acceptance of the idea of a robot proactively suggesting an activity is evaluated through a pilot study. Most participants found the idea highly useful, however, they were not yet fully convinced by the ease of use of such a robot. Overall, the intention to use was moderate to high and can likely be improved even more with further research and extensive communication.
This thesis shows that it is possible to use context information to create an adaptive assistance robot that supports elderly people during their daily activities. It is a first step in the direction of a context-aware robot and identifies a lot of potential for further research.
You can download the full submission as PDF file (3,4 MB) here.