Advanced Sensors for Supportive Environments for the Elderly

Researchers

Principal Investigator: Stephen McKenna
Co-investigators: Alan Newell and Peter Gregor
Post-docs: Hammadi Nait Charif and Fran Marquis-Faulkes

Collaborators

Foxtrot Theatre Co.
Servite Housing Association

Summary of Project

This two year project was funded under the EPSRC EQUAL (Extending Quality Life) programme. It investigated the potential of computer vision-based sensors for supportive home environments for older people. The original objectives in the proposal were achieved and exceeded. Broadly speaking the project consisted of two strands of research: (i) investigation of requirements, constraints, acceptability, social and ethical issues using a novel drama-based methodology, and (ii) development and evaluation of prototype computer vision systems. The research impact is in several areas: (i) the drama-based methodology has a wide potential domain of application in preprototyping, requirements gathering and design; (ii) the results of applying this methodology to the application will inform development of monitored supportive environments; (iii) the algorithms for reliable tracking of human motion have many applications; (iv) high-level activity summarisation and context-dependent inactivity detection are important in other applications: the former provides an efficient coding for storage and retrieval whilst the latter is useful in surveillance.

Publications

This images shows trajectories obtained from extended observation overlaid on a scene. Also shown are automatically learned 2D inactivity zones and 1D entry-exit zones.
This image shows an automatic temporal segmentation and summarisation of activity into: entering through entry zone 1, transitioning to inactivity zone 1, and transitioning to unusual inactivity (a fall).