SafeSt

The app scope covers onboarding-based personalization, route comparison, safety alerts, peer connectivity, and emergency support actions in one flow.

Algorithm Design: Safety Score Calculation

The core route-ranking engine uses a weighted safety score combining street lighting, pedestrian density, open businesses, incident reports, surveillance presence, and environmental conditions. Weights adapt to user preference signals from onboarding. This creates a personalized decision function where the same source-destination pair can produce different route rankings based on user comfort preferences.

Route Recommendation Engine

SafeSt generates candidate routes and attaches interpretable route-level labels.Rather than a single static best path, the app presents a ranked set to support informed user choice while preserving transparency on why each route was surfaced.

Evaluation Logic and Decision Flow

We modeled user-choice correlation as a progressive elimination/check mechanism inspired by round-robin comparison to align route traits with selected safety factors.As onboarding responses accumulate, factor weights and route compatibility are re-evaluated, producing updated recommendations that remain explainable. This evaluation framing helped us validate that route suggestions were both preference-aware and consistent with observed street characteristics.

Demo

Map Route Logic

Personalization Quiz

Run

cd "/Project/safe-st"
npm install
npm run start

Then:

• Press i for iOS simulator
• Press a for Android emulator
• Or scan the QR code with Expo Go

Implemented

• Real map rendering using react-native-maps
• Current location support via expo-location
• Typed origin/destination geocoding via Nominatim
• Multi-route walking alternatives via OSRM (Route A/B/C)
• Safety scoring per route:
  • street-lamp density (OpenStreetMap Overpass)
  • crowd unsafe reports near route
  • route duration and time-of-day adjustment
• User unsafe-point reporting:
  • report from current location during walk
  • red markers shown on map
• Multi-stage navigation flow (search, route options, walking, arrived)