A Hybrid Approach to Recognising Activities of Daily Living from Patterns of Objects Use

Over the years the cost of providing assistance and support to the ever-increasing
population of the elderly and the cognitively impaired has become an economic
epidemic. Therefore, the emergence of Ambient Assisted Living (AAL)
has become imperative, as it encourages independent and autonomous living
by providing assistance to the end user by conducting activity and behaviour
recognition. Accurate recognition of Activities of Daily Living (ADL) play
an important role in providing assistance and support to the elderly and cognitively
impaired. Current knowledge-driven and ontology-based techniques
model object concepts from assumptions and everyday common knowledge
of object used for routine activities. Modelling activities from such information
can lead to incorrect recognition of particular routine activities resulting in
possible failure to detect abnormal activity trends. In cases, where such prior
knowledge are not available, such techniques become virtually unemployable.
A significant step in the recognition of activities is the accurate discovery of
the object usage for specific routine activities. This thesis presents a hybrid approach
for automatic consumption of sensor data and associating object usage
to routine activities using Latent Dirichlet Allocation (LDA) topic modelling.
This process enables the recognition of simple activities of daily living from
object usage and interactions in the home environment. In relation to this, the
work in this thesis addresses the problem of discovering object usage as events
and contexts describing specific routine activities, especially where they have
not been predefined. The main contribution is the development of a hybrid
knowledge-driven activity recognition approach which acquires the knowledge
of object usage through activity-object use discovery for the accurate specification
of activities and object concepts. The evaluation of the proposed approach
on the Kasteren and Ordonez datasets show that it yields better results compared
to existing techniques.