How I got this idea
My co-founder and I realized that proper health and fall trackers are way too expensive (often over ₹10,000) for average Indian families. We wanted to build a simple, reliable health watch under ₹6,000 that normal people can afford for their parents.
Problem
- Vitals monitoring and fall trackers are commercially inaccessible (often costing over ₹10,000) for the middle-class geriatric demographic in India.
- Low-cost devices suffer from high false-alarm rates during daily hand movements like writing or hand waving.
- Lack of real-time remote telemetry sync to emergency contact dashboards.
Solution
- Engineering Approach: Built an integrated hardware-software health watch combining spatial motion sensing and heart rate oximetry with cloud synchronization.
- Architecture: Configured an ESP32 microcontroller reading raw values via I2C, streaming metrics over BLE to a paired client dashboard, which writes updates to a Supabase database.
- What I Personally Built: Programmed the low-level C++ firmware loops under FreeRTOS, configured the MPU6050 sensor calibration offsets, coded the fall-detection algorithm, and designed the real-time Supabase telemetry tables.
- Current Status & Result: Fully functional hardware and software MVP, selected and pitched to mentors at JIC Accelerator 7.0.
ESP32C/C++MPU6050BLEMobile App DevSystem Architecture
Hardest Technical Challenge
Differentiating actual elderly falls from common physical actions to eliminate false alarms. Solved this by structuring a 10-step moving average filter in the ESP32 C++ firmware loop running at 50Hz. The algorithm triggers a fall alert only if the acceleration vector drops below 0.3g (freefall), spike-climbs past 3.0g (impact), and is immediately followed by a 5-second window of immobility.
System Architecture Diagram
What I learned
- Low-level C++ firmware compilation and FreeRTOS multi-threading tasks.
- I2C protocol configuration and hardware sensor calibration.
- Power management and LDO linear regulator circuit integration.