Smart Environments Research Themes

The smart environments research group operates across a number of research theme relating to Smart Environments, Smart Communities and Smart Cities.

Activity Recognition within Smart Environments

One of the key areas of research within the group is to develop computational models for the purposes of activity recognition. This has involved the development of techniques to segment data streaming from sensor data gleaned from smart environments in addition to modelling behaviours of both normal and abnormal behaviours.

Our work has introduced knowledge-driven approaches to real-time, continuous activity recognition based on multisensor data streams in smart homes.

Our approaches go beyond the traditional data-centric methods for activity recognition in three ways.

  1. Make extensive use of domain knowledge in the life cycle of activity recognition.
  2. Use ontologies for explicit context and activity modeling and representation.
  3. Exploit semantic reasoning and classification for activity inferencing, thus enabling both coarse-grained and fine-grained activity recognition.

Behaviour change and lifestyle monitoring

As the demographics of many countries shift towards an ageing population it is predicted that the prevalence of chronic conditions such as Chronic Obstructive Pulmonary Disease (COPD) and Dementia will continually increase. One approach to reducing the socio-economic burdens imposed by such illnesses is within a self-management paradigm whereby a person is empowered to manage the day-to-day implications of living with their condition. A key aspect of self-management is empowering the patient.

Our work is currently developing a range of mobile based self-management solutions to support persons with dementia, COPD and stroke. One strand of this empowerment is through education whereby the individual can navigate to, and consume clinically validated educational resources.

A further facet is collaborative goal setting whereby a patient chooses life goals with guidance from a healthcare professional. In order to measure the impact of education upon lifestyle behaviour and to facilitate the choosing of goals that are measurable it necessary that health metrics can be captured, quantified and appropriately displayed. We have integrated within our mobile apps a number of commercially based vital sign monitoring devices and activity monitors.

Crowd Labelling of Physical activity

The use of smart phones can be viewed as one possible manner in which large amounts of data may be captured unobtrusively. Many handsets now have a range of in-built sensors, large memory storage, fast processing and low power communications.

Unfortunately, using mobile devices to gather data on a large scale can prove difficult. In particular the integrity of the user annotation can be questionable. Nevertheless, a large scale fully annotated data set is recognised as being the key step to improving and increasing the widespread adoption of activity recognition applications.

Our current work recognises class transitions from a set of specific activities (for example walking and running) to the standing still activity. Once this transition is detected the system prompts the user to provide a label for their previous activity. This label, along with the raw sensor data, is then stored locally prior to being uploaded to cloud storage where it can be further analyzed. This approach has established the infrastructure which now supports the large scale collection and annotation of data.

Online data repositories

An underpinning theme within the Group over the last 10 years has be the creation of online data repositories to support the development of data driven approaches to activity recognition.

In this work issues surrounding the collection, annotation, management and sharing of data gathered from pervasive health systems have been addressed.

The overarching motivation for this work has been to provide an approach whereby annotated data sets can be made readily accessible to the research community in an effort to assist the advancement of the state-of-the-art in activity recognition and behavioural analysis using pervasive health systems.

The definition of common formats for data storage and a common set of tools for data processing and visualisation have been produced to support this work.

Technology adoption modelling

Reminders are an intrinsic feature of many assistive technology solutions. Nevertheless, many of solutions fail to appreciate the complexities of effectively creating and delivering reminders.

Our work has sought to develop and evaluate reminding solutions designed specifically at older adults with dementia.

This has led to the realisation that there is no one size fit all solution and that the carer has a large role in whether or not the technology is successfully adopted.