End User Control of Smart Home Automations Through Mobile Augmented Reality and Recommender Systems

In recent years, the so-called smart objects have become increasingly common in our everyday environments, making the vision of an Internet of Things a reality. They promise to improve our lives, optimising comfort, energy management, and in general our daily well-being. To make the most of the possibilities of these ecosystems of connected objects and services, it is necessary to use platforms that allow their coordinated use, enabling the creation of multi-object automations. However, there are several problems with these platforms, both commercial and research ones. Firstly, it isn’t easy to balance expressiveness with ease of use. Indeed, home automation platforms range from allowing only simple "if-then" automations to the possibility of defining actual programs. However, the most expressive platforms tend to become difficult to use and unengaging. An unbalanced expressive capacity can make these platforms of little use to the user. Furthermore, the selection and configuration of functionalities to be used in automations can be a complex operation. It is, therefore, necessary to find solutions to make this task easier for users, using representations that allow them to form an accurate mental model of the functionalities of the tools and consequently use them correctly. This dissertation will present how mobile-enabled Augmented Reality can minimise these problems and empower users to seamlessly create automation in smart environments. It will also discuss how recommender systems can be introduced in this context, further facilitating operations on the platform. Two main cycles of platform development, each culminating in a user study, will be reported, along with the additional preliminary studies and interviews conducted. Together, these activities enabled us to answer the defined research questions and delineate the ‘lessons learned’, which can serve the development of further solutions in this regard.