DawgWalk — Safer Night Walks for UW Students

Inspiration

Walking home at night around UW can feel uncertain, and traditional map apps don't factor in safety. Students still rely on "text me when you get home," which is inconsistent and easy to forget. We wanted a tool built specifically for Huskies — something that prioritizes safety, adds real context, and provides genuine peace of mind. That’s how DawgWalk started.

What It Does

DawgWalk recommends the safest walking route, not just the fastest.

Our system analyzes:

Multiple Google walking routes
Weather visibility and conditions from WeatherAPI
Basic risk factors like route length, lighting, and environmental conditions

The app returns:

A Safety Score (0–100)
Safety tags with short explanations
ETA and distance
A visualized safest route on the map

How We Built It

Frontend (React Native + TypeScript)

Map view with polylines for each route
Clean, intuitive UI optimized for quick decision-making
Real-time updates from backend APIs

Backend (Python + Flask)

Google Directions API for route candidates
WeatherAPI for visibility and weather data
Custom safety scoring engine that:
- Decodes and samples route polylines
- Computes safety metrics from conditions
- Ranks and returns the safest option
/safe-route endpoint provides full route, score, and tags

Challenges

Finding reliable, free visibility data and converting it into actionable scores
Debugging inconsistent Google Maps route outputs and API key restrictions
React Native map inconsistencies across devices
Limited time for integrating UW Alerts and campus callbox datasets

Accomplishments

Built a fully functional MVP with real routing, real weather checks, and a custom scoring algorithm
Designed a modular backend that’s easy to extend with future datasets
Delivered a clean, smooth mobile UI
Created something meaningful for the UW community — an app we would actually use

What We Learned

How to unify multiple APIs into a single scoring model
How to design modular, fast endpoints for mobile clients
How to prioritize core features under tight time constraints
That even simple context (like visibility) can meaningfully improve perceived safety