PennApps 2025: ShadeNav

Climate change has created a devastating feedback loop: rising temperatures make walking unbearable, forcing people to drive more, pumping more greenhouse gases into the atmosphere, making cities even hotter.

ShadeNav breaks this cycle by making walking comfortable again through intelligent shade-optimized routing. Using satellite analysis and real-time sun positioning, we find paths that maximize shade coverage - because shaded areas feel up to 20°F cooler than direct sunlight.

A comprehensive route-planning system for downtown Philadelphia that optimizes paths based on shade coverage and essential services using advanced satellite imagery analysis and OpenStreetMap data.

See: https://devpost.com/software/shadenav

Project Structure

• frontend/: Vite + React + TypeScript with Leaflet map component for interactive visualization
• backend/: Advanced tree detection and waypoint pathfinding system with satellite data analysis

Features

Interactive Map of Downtown Philadelphia

• High-Resolution Urban Mapping: Street-level visualization with building footprints and infrastructure
• Real-time Interface: Interactive Leaflet map with dynamic route rendering and gradient shade analysis
• OpenStreetMap Integration: Comprehensive road networks and urban data foundation

Advanced Shadow Simulation System

• Building Height Analysis: Precise height extraction from OpenStreetMap and satellite data
• Solar Ray Tracing: Computational geometry for sun ray simulation at any time/date
• Dynamic Shadow Mapping: Real-time calculations using sun position, building geometry, and seasonal variations
• Multi-Source Shade Detection: Combines building shadows with tree canopy coverage
• Temporal Prediction: Forecasts shade patterns throughout the day

Intelligent Pathfinding Algorithms

• Dijkstra's & A Search*: Shortest path computation with shade-weighted edge costs and heuristic optimization
• Custom Shade Metrics: Edge weights incorporating shade coverage, temperature differentials, and comfort indices
• Multi-Objective Optimization: Balances route efficiency with shade maximization
• Dynamic Re-routing: Real-time path adaptation based on sun position changes

Tree Detection & Environmental Analysis

• Satellite Analysis: Google Earth Engine vegetation detection (NDVI, NDWI, GNDVI indices)
• Multi-Strategy Fusion: Weighted combination of satellite and OpenStreetMap data
• Ultra-Fine Resolution: 5m grid analysis with density scoring for precise tree detection

Smart Waypoint Integration

• Essential Services: Water fountains, convenience stores, and climate-controlled spaces
• Adaptive Selection: Dynamic waypoint adjustment based on route length and weather conditions

Weather and UV Integration

• Hourly Weather & UV Updates: Real-time weather and UV index integrated directly into routing, refreshing every hour.
• Heat-Aware Routing: Helps pedestrians track heat stress conditions, avoid unsafe UV exposure, and choose cooler, safer walking paths.

Terrain Heuristic for Pathfinding

• Grade-Aware Costing: Incorporates slope from DEM tiles to penalize steep segments.
• Surface & Roughness Scoring: Weights edges by OSM surface (e.g. cobblestone, unpaved) when respective toggles are checked
• Accessibility Constraints: Applies hard costs to stairs/steps and soft penalties for camber/irregularity, enabling wheelchair-friendly routing and safer sidewalk selection

Quickstart

Backend Setup

1. Navigate to backend directory:

   cd backend

2. Install dependencies:

   python -m venv .venv
   .\.venv\Scripts\Activate.ps1  # Windows
   # or
   source .venv/bin/activate      # macOS/Linux
   pip install -r requirements.txt

3. Set up Google Earth Engine (for satellite data):

   # Install Earth Engine API
   pip install earthengine-api
   
   # Authenticate (this will open a browser window)
   python -c "import ee; ee.Authenticate()"
   
   # Test authentication
   python -c "import ee; ee.Initialize(); print('Earth Engine working!')"

4. Optional: Configure project settings:

   # Copy the example environment file
   cp .example.env .env
   
   # Edit .env and add your Google Cloud Project ID (optional)
   # This is only needed if you want to use a specific project

5. Run tree detection:

   python tree_detection.py

6. Run waypoint pathfinding:

   python waypoint_pathfinding.py

7. Start the API server:

   uvicorn app:app --reload --port 8000

Frontend Setup

1. Navigate to frontend directory:

   cd frontend

2. Install dependencies:

   npm install

3. Start development server:

   npm run dev

Pre-generated Data

The backend includes pre-generated JSON files from running the detection systems:

• tree_positions.json: 703 detected trees with satellite analysis data
• waypoints_data.json: Essential services waypoint data

These files can be used immediately without running the detection algorithms.

API Integration

The generated JSON files can be easily integrated with frontend applications:

import json

# Load tree data
with open('tree_positions.json', 'r') as f:
    tree_data = json.load(f)

# Access trees
trees = tree_data['trees']
for tree in trees:
    lat, lon = tree['latitude'], tree['longitude']
    density = tree['density']
    source = tree['source']

Development Process

Developing the graph

Initial street network extraction from OpenStreetMap showing downtown Philadelphia's road infrastructure. This visualization demonstrates the comprehensive coverage of walkable paths that form the foundation of our routing graph.

Processing edge shadow data

Shadow analysis pipeline in action, processing individual street segments (edges) to calculate shade coverage. The visualization shows the computational process of analyzing building heights and sun angles to determine shadow patterns for each walkable path.

Profiling edges at different times of day

Temporal shadow analysis across different hours of the day, demonstrating how shade patterns shift as the sun moves. Each visualization captures shadow distribution at key times (morning, midday, afternoon, evening), showing the dynamic nature of urban shade coverage that drives our intelligent routing decisions.

Comprehensive Shade Analysis

Aggregated shade coverage analysis that combines temporal shadow data across all time periods to create comprehensive shade profiles for each street segment. This visualization represents the culmination of our shadow analysis pipeline, showing average shade coverage that enables optimal route planning throughout the day.

Development Highlights

These are completed components that demonstrate core capabilities but are not yet fully integrated into the main ShadeNav system:

• Tree Detection System: Used the Google Earth Engine API with multi-spectral vegetation indices (NDVI, NDWI, GNDVI) for precise tree location detection and size estimation, successfully identifying 700+ urban trees in downtown Philadelhia with geometric shadow computation and pathway intersection analysis.

Visualization of the 5m grid tree detection system showing vegetation density scoring across downtown Philadelphia.

The tree detection system uses Google Earth Engine API to process Sentinel-2 satellite imagery with NDVI, NDWI, and GNDVI vegetation indices at 5m resolution, creating a dense grid where each cell's vegetation density score serves as a proxy for tree size estimation. Once trees are detected and sized, the system performs geometric shadow computation using solar ray tracing algorithms that calculate shadow patterns based on sun position, tree height, and canopy geometry throughout the day. These computed tree shadows are then intersected with urban pathway networks using polygon-polyline intersection algorithms to determine shade coverage percentages for each route segment, enabling the routing system to weight pathways based on their shade coverage for optimal pedestrian comfort.

Interactive map interface showing tree shadows, pathway intersection analysis, and shade-aware routing.

• Water Stop Waypoint Routing: Dynamic re-routing through convenience stores and water fountains using POI snapping and path-projection heuristics.

Future Work

Core Enhancements

• Crime & Safety Integration: Balance shade optimization with real-time safety data
• Accessibility Features: ADA-compliant routing with wheelchair accessibility
• Emergency Shelter Network: Integration with cooling centers for extreme heat events

Urban Planning Analytics

• Optimal Shelter Placement: AI-driven analysis to identify strategic locations for new shade structures that maximize citywide walking comfort
• Heat Island Mitigation: Data-driven recommendations for tree planting and infrastructure modifications to reduce urban heat effects
• Pedestrian Flow Analysis: Integration with foot traffic data to prioritize shade improvements in high-usage corridors
• Cost-Benefit Optimization: ROI analysis for shade infrastructure investments based on pedestrian comfort gains

Advanced Technologies

• IoT Sensor Integration: Real-time temperature and air quality data from city sensors
• Machine Learning Personalization: User-specific heat tolerance and comfort preferences
• Municipal Planning Tools: City dashboard for pedestrian comfort analytics

Dependencies

Core Dependencies

• osmnx: OpenStreetMap data processing
• geopandas: Geospatial data manipulation
• pandas: Data analysis
• folium: Interactive mapping
• numpy: Numerical computing

Optional Dependencies

• earthengine-api: Google Earth Engine integration for satellite data
• python-dotenv: Environment variable management
• scipy: Advanced interpolation for satellite data processing

External Services

• Open-Meteo: Provides real-time weather and UV index data for routing decisions. Attribution