DEMO MODESensor readings, alerts, and fallback location are simulated.

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.

Built with GenMB
Built with GenMB