DIY Smart Home: Narrowing the gap between users and technology

Over the past decade, we have been building intelligent systems to provide assistance for persons with dementia wishing to age-in-place [1]. These systems are built using partially observable Markov decision process (POMDP) controllers. While the POMDPs provide a framework for long-term model learning and reinforcement learning, they must be engineered to fit each situation and user to ensure adoption. Custombuilding a smart home solution is a time-intensive exercise, in which technologists extract and encode knowledge from end users. As this process is neither easily understood by, nor accessible to, end users, we are exploring this problem from two complementary directions. From the user perspective, our research with older adults and family members has revealed the need for practical, customizable systems that can be set up and managed by “informal caregivers”. From the technological perspective, we have developed a method by which POMDP-based smart-home controllers can be specified at an appropriate level of abstraction using probabilistic relational modelling and a database schema for assistance systems. Our goal is to close the gap between these perspectives by designing interfaces for end-users to express contextual knowledge, needs, and preferences in a natural way to build custom smart home solutions in an organically developing, DIY (“doit-yourself”) fashion. Copyright is held by the author/owner(s). IUI ’13, March 19–22, 2013, Santa Monica, California, USA. ACM xxx-x-xxxx-xxxx-x/xx/xx. Figure 1. Conceptual user interface mock-up for a sensing and control system that would enable end users to build custom, DIY (“do-it-yourself”) smart home solutions. Amy Hwang Graduate Dept. of Rehabilitation Science & Intelligent Assistive Technology and Systems Lab 500 University Avenue Toronto, ON M5G 1V7 Canada [email protected] Michael Liu & Jesse Hoey Faculty of Engineering David R. Cheriton School of