Nirbhay: Automated Emergency Detection

In moments of danger, calling for help can be impossible — fear, shock, or physical restraint freeze the victim. Elderly people experiencing falls or medical emergencies, people with disabilities, and victims of violence often cannot access their phones exactly when they need them most. Every existing solution we looked at required the victim to act: press a button, open an app, make a call.

Remove the victim from the critical path. The system has three layers: a discreet ESP32 wearable that continuously monitors vital signs and movement, detection that recognizes distress patterns — sudden impacts for falls and assaults, physiological signs of distress, and YAMNet-based real-time audio distress detection — and an automated response that alerts emergency contacts with precise location and health information. A Flutter mobile app with a Firebase backend ties it together for caregivers: real-time alerts, location sharing, and remote health monitoring.

• Wearable (ESP32 firmware): vital-sign and movement sensing in a discreet, comfortable form factor.
• Detection: distress pattern recognition, fall and assault detection from motion signatures, YAMNet audio classification for distress sounds.
• Response pipeline: automatic alert dispatch to emergency contacts with location data and health context — no user action required.
• Caregiver surface (Flutter + Firebase): real-time alerts, location sharing, and vital-sign monitoring for family and responders.
• Design constraint throughout: accessible to all ages and abilities, dignity-preserving, independence-enabling rather than surveillance-feeling.

Safety tech is a trust problem before it is a detection problem. The hardest design questions were not about sensors — they were about false alarms (who gets woken at 3am, and how do you cancel?), privacy (continuous monitoring that does not feel like surveillance), and who the system answers to when the wearer cannot respond. Building for elderly, disabled, and at-risk users forced every interaction to justify itself: if it requires the victim to act during an emergency, it does not belong in the critical path. Honest scope note: this is a working prototype and concept validation, not a clinically validated medical device — real-world testing with vulnerable populations and healthcare integration remain future work.

Hardware demo: the ESP32 wearable pairing with the Flutter app over BLE (71s).