De Montfort University is to use fuzzy logic to monitor the elderly in a project with the University of Missouri’s Center for Eldercare and Rehabilitation Technology.
“The job is to help with technology by putting sensors into sheltered accommodation,” De Montfort’s Professor Bob John told Electronics Weekly.
Sensor data includes ‘anonamised’ silhouettes from cameras, sounds from microphones, and, for example, fridge door sensors.
“We are trying to monitor behaviour - to try to pick out the image of a human falling down, but not a human bending over to pick something up,” said John. “Or to tell the difference between the sound of a dropped book and a person falling down. Or did they open the fridge in the last 24 hours?”
Fuzzy logic is used to make decisions from merging data where the precise relationship between data types is only loosely known.
“What it is good at is dealing with uncertainty and noise,” said John. “When you are not spending much money on sensors because there are a lot of them, you are likely to get a lot of noise.”
De Montfort will be using type two fuzzy logic.
“Image data and noise has not been tackled by type two fuzzy logic before,” said John. “One of the challenges is type two is computationally intensive.”
Two things will cut the computational load: One is that De Montfort’s Dr Simon Coupland has developed an algorithm that helps considerably, and will spend part of this year in Missouri implementing it in prototype sheltered accommodation.
The second is that real-time performance is not required. “We have to get someone there in five minutes,” said John.
The final system may not entirely be fuzzy logic. “These types of artificial intelligence systems are normally mixed,” said John.
The project aim is to have prototype fuzzy logic running in the accommodation by the summer.
The Engineering and Physical Sciences Research Council has contributed £45,000 to the project.