Use this URL to cite or link to this record in EThOS:
Title: Goal modelling, recognition and planning for guidance within smart environments
Author: Rafferty, Joseph
ISNI:       0000 0004 5992 8815
Awarding Body: Ulster University
Current Institution: Ulster University
Date of Award: 2016
Availability of Full Text:
Full text unavailable from EThOS. Thesis embargoed until 01 Oct 2018
The global population is aging. This aging is expected to produce an increase in aging related illness which will, subsequently, place an increased demand on health care infrastructures. Smart homes represent a promising approach to providing support to sufferers of aging-related illnesses, particularly when applied to inhabitants suffering cognitive impairments, such as dementia. Smart homes may provide support in a variety of forms such as, on demand assistance with activities of daily living. This support can increase the independence and quality of life of individuals while simultaneously providing care with reduced overheads. Currently smart homes providing such functionality have several deficiencies related to handling variation in activity performance, providing reusable and flexible activity representations, providing dynamic illustrative guidance and having a reliance on dense sensing. Exploring how to address these deficiencies was the focus of this Thesis. In order to address these deficiencies, this Thesis introduced a number of novel elements that extended the domain knowledge; a flexible and reusable goal model, goal recognition algorithms and processes, a process of determining actions presented in videos and a mechanism to nominate videos to assist with such goals. These elements formed components of a novel goal-driven smart home system and were integrated into a prototype assistive smart home system. Evaluation of the system demonstrated promise. The goal model produced was able to flexibly model inhabitant activity, in a reusable manner. Recognition of an inhabitant's intended goal with a reduced set of sensors was performed with a 100%. 83.3% and 64.45% accuracy across three evaluation scenarios. Analysis of video content to identify stepwise activity was performed with 85.04% accuracy. Finally, matching analysed videos to in habitant goals in need of assistance was performed reliably across 3 evaluation scenarios; with 100%,80% and 60% accuracy.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available