A user-guided personalization methodology for new smart homes

Recently, smart homes have become a centre of attention due to their provision of an enhanced quality of life via automation services with-in the homes. Smart homes technology, however, must learn and adapt in accordance with the habits of their residents in order to provide the relevant services.
Human activity recognition is a well-known technique used to under-stand user behaviours and enables the smart home services to run automatically according to the human mind. Observing the pattern of resident’s daily tasks is a useful technique used by the researchers to develop more user-centric personalised services for the occupant. There are several approaches to human activity recognition in case of smart home. Among the most popular ones is the data-driven approach, which provides promising results due to its advancement of machine learning. Regardless of this, several drawbacks such as limited availability of data in the initial phases can be a hurdle in providing smart home services. The purpose of this thesis is to introduce an approach known as useR-guided nEw smart home ADaptation sYstem (READY) for the development of a personalised automation system which provides users with smart home services when they move into their new house. The READY approach integrates several approaches, leverages user feedback, and builds a rich data set that helps the house recognise the user’s daily activities and provide personalised smart home services, accordingly. The system development process was strongly user-centred, where the user was involved in every part of the development to receive fine-grained services from the outlet. Additionally, the research introduced a supplementary method along with READY called User-guided Transfer Learning (UTL) that leverages the existing smart home data set in order to enhance the overall automation functionality and effectiveness.
The approaches presented in this thesis have been tested and validated at Middlesex University Smart Lab by a group of internal and external participants. The results show that 100% of these participants believed that the READY method provides personalized services to the new smart home from outset through user involvement. Moreover, the UTL approach detects new services, and this increases the acceptability of the new smart home. The results of the given approaches prove to be a significant advancement in the domain of smart home technology and become a positive step toward bridging the gap between the new smart home and incoming residents.