FallGuard
The simulation was just reset — pick a role and enter your name again to rejoin.
Who are you?
Pick what best describes you — this screen will only show what that person would actually see.
What's your first name?
So the other side of this knows who they're hearing from.
This simulation syncs live over Firebase: open it on two devices or browser tabs, and actions taken on one show up on the other in real time.
Your residents
A mockup of what a caretaker managing several people at once would see — whoever needs attention rises to the top automatically.
Accounts
Whoever has joined so far, live.
A waist-worn accelerometer feeds a Raspberry Pi at roughly 50 readings a second. When the freefall → impact → stillness signature of a real fall is confirmed, a buzzer sounds immediately — and this notification cascade starts in parallel, checking on the care recipient and pulling in their emergency contact if they can't respond.
Demo timing: each notification stage waits 8 seconds for a response instead of the minutes a real device would allow. Detection itself is classical signal processing on the real device — no LLM involved.
Built by Kadair Walker, Dharma Swaroop, Pragnya Eranki, and Nidhi Mendu — SteelHacks 2026.