Energy- and context-aware approaches to aiding human memory

Abstract

Our memory defines us, but it is also fragile and fallible. When memory fails us, a range of supportive technologies can be used, from calendars and diaries to electronic devices for reminding. As problems with memory are strongly associated with age related diseases such as dementia, and even with healthy ageing memory, the rapidly ageing Western population has driven particular interest to the field of electronic memory aids.

In this thesis, the state of the art in these devices is explored, and the contributions from the engineering and psychology literature are compared. This leads to the proposal of a structured design approach for the design of future aids. This approach is then used in the design of a memory aid for names, which is evaluated in a practical user study. This system is then extended to take into account the limited power budget: a rule system is introduced to control the energy expenditure, and this is then trained via a machine learning approach, leading to significant improvements in battery life. The remainder of the thesis then considers how another memory aid focussed on the recognition of context can be developed, and an energy-efficient approach to context awareness is proposed and demonstrated.

More information

Identifiers

ORCID for Geoff Merrett: orcid.org/0000-0003-4980-3894

Catalogue record

Export record