Towards a Pervasive Prompting System: Improving the Expanding the COACH
This project is a collaboration between Dr. Jesse Hoey (Dundee) and Dr. Alex Mihailidis (Occupational Science and Occupational Therapy, University of Toronto), Prof. Craig Boutilier (Computer Science, University of Toronto), Prof. Geoff Fernie (Medicine, University of Toronto), and Dr. Gary Naglie (Medicine, University of Toronto). The aim of the project is to expand and improve upon the existing COACH system for assisting persons with dementia during handwashing. The project follows on from a previous grant from the same program.
The objective of this next phase is to develop the necessary algorithms needed to further improve and expand the COACH to achieve our overall goal of a pervasive ADL prompting system for use in the home. We will implement improvements to our three existing system modules—sensing, planning, and response. We will continue to develop new markerless tracking and sensing algorithms that will allow us to track not only fine motor movements (i.e., the person's hands), but also overall body movements in order for the system to determine what tasks the person is completing. Another important improvement to the system is the ability to automatically recognize different ADL tasks and plans (beyond handwashing), including the completion of potentially interleaved activities (e.g., two ADLs could be completed at the same time). We will develop new hybrid activity recognition algorithms that will be added to our existing planning techniques.
Finally, we will develop a new response module based on adaptive dialogues. Using speech recognition and more sophisticated artificial intelligence techniques, the system will be able to listen and respond to utterances and questions by the user. For example, we have often observed in our past trials that sometimes the user would ask attempt to have a dialogue with the automated prompts from the system, but then would become frustrated when the system would not respond appropriately.
Once the system has been developed, we will conduct an efficacy study using handwashing and a second representative ADL, such as toothbrushing. We will complete these new trials within the homes of several users.
Funder: American Alzehimer's Assoication Everyday Technologies for Alzheimer's Care
Start Date: 01/09/2008
End Date: 30/09/2010
Research Themes:
For further information about the Towards a Pervasive Prompting System: Improving the Expanding the COACH project please contact Dr Jesse Hoey.