Droplet

Reward page
AI page
Hardware
live sensor testing
insights

Inspiration

Water waste is one of the most overlooked everyday problems. Simple actions like leaving the tap running while brushing teeth or taking slightly longer showers can waste significant amounts of water.

The key issue isn’t intention — it’s awareness.

We realized that unlike steps or electricity, people don’t get real-time feedback on water usage. That lack of visibility makes it hard to change behavior.

Droplet was inspired by a simple idea:

If people could see their water usage in real time and understand its impact, they would naturally use less.

What it does

Droplet is a smart water monitoring system that combines hardware, software, and AI to track and reduce water waste.

It provides:

Real-time tracking of water usage using sensors
Live session monitoring (time, flow intensity, estimated usage)
Room-based insights (bathroom, kitchen, sink)
Usage analytics through charts and breakdowns
AI-powered suggestions and daily summaries
Smart alerts for long usage or anomalies
Gamification system with points, streaks, and rewards

The system helps users understand:

how long they use water
where they waste the most
how they can improve

How we built it

Droplet is a full-stack system with three main components:

Hardware (Arduino)

Built using an Arduino UNO R3
Uses 3 probe sensors to detect water flow via conductivity
Flow intensity is determined by probe activation:
- 1 probe → Low flow
- 2 probes → Medium flow
- 3 probes → High flow

Backend (Python + FastAPI)

Processes real-time sensor data via a serial bridge
Tracks:
- active sessions
- completed sessions
- usage metrics (duration, gallons, cost)
Detects:
- long usage
- anomalies
- possible leaks
Generates AI-based advice and summaries

Frontend (Flutter)

Multi-page mobile/web app with:
- Home dashboard (live session + rewards)
- Rooms page (usage by location)
- Insights page (charts and analytics)
- AI chat assistant
- History view
Displays real-time updates and visual feedback

We also used demo seeded data alongside real sensor input to simulate a full multi-room environment.

Challenges we ran into

1. Real-time data pipeline

Arduino → Serial → Python → Backend → Flutter

Connecting was challenging. Initially, the data wasn’t reaching the app. We solved this by building a serial bridge that actively forwards events to the backend.

2. Hardware limitations

We didn’t use expensive flow meters, so we had to estimate usage using probe-based detection and timing, balancing simplicity and accuracy.

3. Data consistency issues

We faced problems with:

timestamp mismatches (naive vs UTC)
inconsistent JSON structures between backend and frontend

We solved this by standardizing formats and making models flexible.

4. UI felt like a demo initially

The app initially looked like a basic dashboard. We improved it by adding:

navigation across multiple pages
charts and visual insights
room-based breakdown
reward system
AI summaries

Accomplishments that we're proud of

Built a complete IoT system from hardware to mobile app
Achieved real-time water tracking using low-cost sensors
Created a clean, multi-page UI with charts and insights
Successfully integrated AI-based feedback and summaries
Designed a system that works even with limited hardware using demo + live hybrid data
Made the app feel like a real product, not just a prototype