Project
ElderCare Watch
- Role
- Author
- Context
- Undergraduate thesis · UFF
- Team
- Solo
- Stack
A Mamdani fuzzy inference engine written from scratch, in Swift.
The problem
Monitoring risk in elderly people involves signals with no sharp boundary: a heart rate that is "a bit high", movement that is "reduced", time still that is "too long". A rigid threshold rule turns those gradations into an alarm that fires at 101 and stays silent at 100.
Fuzzy logic handles gradation directly, but using an off the shelf library hides exactly the part the thesis needed to demonstrate.
The solution
A Mamdani fuzzy inference engine written from scratch in Swift, with fuzzification, rule base, aggregation and centroid defuzzification.