SafePath: The Fail-Safe Solo Travel Guardian

The screen that the emergency contact will receive

Inspiration

"How many times have you texted someone 'I'm on my way', not because you wanted to chat, but because you were a little scared and just wanted someone to know where you were?" That message is often the only safety net we have.

We are a group of four girls. Two of us have experienced sexual harassment simply while going out alone, walking home, riding a motorbike, or even going for a run. We were followed, stared at, and subjected to inappropriate comments in situations that should have felt normal.

And what we realized is this: It's not rare. It's routine. Almost every woman we know has a similar story. Over time, these moments don't just pass — they accumulate. They quietly change how we move through the world. We hesitate, avoid certain streets, and pretend to be on calls. Not because we want to, but because we don't feel safe.

And that led us to a deeper realization: the world is asking individuals to adapt to unsafe environments, instead of building systems that actively protect them.

What it does

SafePath is built around a simple idea: you shouldn't have to ask for help to be protected.

You open SafePath, enter your destination, and begin your journey. Then you put your phone away. From that moment, the app becomes a silent guardian that watches your route in real time and understands how you're moving. It only intervenes when something feels wrong.

If you suddenly stop for too long, go off route, or your trip takes significantly longer than expected, your phone vibrates:

"Are you safe?"

You tap "I'm safe" and continue.

But if you don't respond — because you can't — SafePath doesn't wait. It escalates automatically. Your emergency contacts receive a real-time alert with your identity and live location.

Not because you asked for help, but because the system knew something wasn't right.

How we built it

SafePath operates through four core layers:

1. Tracking Layer

Continuous GPS monitoring
Route matching against planned path

2. Detection Layer

We detect anomalies using spatial and temporal signals:

Deviation \(> 50\text{m}\) from route
Stationary within \(20\text{m}\) for 5 minutes
Trip duration significantly exceeds ETA

3. Intelligence Layer (AI Agent)

Instead of blindly triggering alerts, the system understands context:

Heavy traffic? Probably normal.
Taking a parallel route? Acceptable.
Stopped in an unfamiliar area with no clear reason? Investigate.

4. Intervention & Escalation Layer

When an anomaly is detected, the phone vibrates every 20 seconds and a 3-minute countdown begins:

IF user confirms → reset IF user requests help → immediate alert IF no response → automatic escalation

5. Emergency System

We use Firebase Cloud Messaging (FCM) instead of SMS to deliver:

Instant push alerts
Live location links
Real-time delivery at scale

This ensures low latency, high reliability, and cost efficiency.

Challenges we ran into

Designing anomaly detection without excessive false positives
Synchronizing real-time mobile and backend state
Handling edge cases: app restart, stale sessions, duplicate triggers
Ensuring vibration + alert timing works reliably across devices
Integrating AI reasoning without slowing down the system

Accomplishments that we're proud of

Designing a system that is silent enough to not attract attention, yet strong enough to signal urgency
Building from lived experiences — turning emotional pain into system logic
Representing a widespread issue responsibly
Delivered a solution that is: technically robust, emotionally meaningful