Floor-Vibration Fall Detection for Older People Living Alone: A Detection Budget for a System with No Wearables and No Cameras
Year: 2026
Type: preprint
Abstract: SIMULATED, NOT MEASURED. A design study computing a detection budget for floor-vibration fall detection for older people living alone, with no wearables and no cameras. From stated physical assumptions, the vibration signature of a hard fall is modelled against footsteps, dropped objects, door slams, chair moves and a cat jumping down, as a function of trigger threshold, sensor count and floor type (joisted timber vs concrete slab). With a classifier rejecting 90% of confusers, no configuration meets both a 0.80 hard-fall detection floor and a 0.01 false-alarms-per-day keep-on ceiling; with 99% rejection exactly one does (timber, four sensors). Soft falls are essentially undetectable at every usable operating point.
Author: Aldred, William Alexander
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225744666
Cites:
Cited by:
Related to: