To meet the needs of an ever ageing population it has been proposed that assitive technology might provide aid and support for elderly people and allow them to reside for longer in their own homes. Such proposals face many social, ethical and technical challenges. In the research presented in this paper we study how a commercially available robot situated in a fully sensorised, although in all other ways typical home, can meet some of these challenges. We focus especially on the technical integration of the robot, the home and the actions of the house residents especially in relation to the learning and control architectures. We envisage that such a software architecture would not only be useful as a physical and memory prosthetic but would also provides active support for re-ablement and co-learning.