Project innovation
Automatic emergency detection pipeline
The central idea is not GPS or a manual SOS button. It is the AI-inspired system that detects high-risk motion patterns and starts the alert process automatically.
Main innovation
AI-based automatic emergency detection without completely depending on manual SOS activation.
1. Smartphone sensors
Accelerometer, gyroscope, optional device location, and movement timing.
2. Movement detection
Detect walking and identify abrupt impacts, rapid rotation, or irregular motion.
3. Feature extraction
Transform raw readings into acceleration, rotation, variance, and pattern features.
4. AI/ML classification
Classify activity as NORMAL, SUSPICIOUS, or EMERGENCY.
5. Risk score
Combine signals into a 0–100 risk score and compare it to the alert threshold.
6. Emergency confirmation
A high-risk pattern opens a short cancellation window before automatic action.
7. Automatic alert + location
The alert includes the current location snapshot as supporting information.
Prototype boundaries
In DEMO MODE, readings, location fallback, and alert delivery are simulated and labelled. The prototype records alert history to the provided key-value store; it does not send real SMS messages, place calls, or contact emergency services.