Traffic Analysis with Generative Intelligence and Crowd Simulation
Simulate • Analyze • Optimize Real-World Evacuation Systems
TRAGIC is a floorplan-aware crowd evacuation simulation engine designed to model realistic human movement, hazard propagation, and egress performance.
Unlike traditional simulators, TRAGIC not only models evacuation dynamics but also analyzes system inefficiencies and generates actionable architectural recommendations.
- 🧍 Agent-Based Simulation — Models individual human behavior with dynamic decision-making
- 🔥 Hazard Modeling — Fire spread, smoke diffusion, and exit failure simulation
- 🧠 Intelligent Routing — Agents select exits based on distance, congestion, and environment
- 📊 Analytics Engine — Generates time-series KPIs and evacuation metrics
- 📍 Bottleneck Detection — Identifies congestion hotspots automatically
- 🧠 AI-Driven Recommendations — Suggests structural improvements to optimize evacuation
- 🚶 200 agents simulated
- ✅ 134 successfully evacuated (67%)
⚠️ Peak congestion observed during mid-phase evacuation- 🔥 High-density hotspots identified at critical junctions
- 🚪 Significant exit utilization imbalance detected
- Increase corridor width at choke points
- Redistribute flow toward underutilized exits
- Improve turn/merge geometry to reduce friction
- Remove local obstructions near high-density zones
This project demonstrates:
- Design of multi-agent systems
- Implementation of real-time simulation engines
- Integration of spatial analytics + visualization
- Application of optimization and decision-making logic
- End-to-end pipeline from simulation → analysis → recommendation
- Hybrid motion modeling (social-force + pathfinding)
- Grid-based spatial environment representation
- Heatmap-based density and behavioral analysis
- Automated report generation with actionable insights
- Scalable architecture for large agent populations
| Output File | Description |
|---|---|
agent_paths.png |
Agent trajectories and flow patterns |
density_heatmap.png |
Crowd congestion visualization |
floorplan_improvement_overlay.png |
Suggested structural improvements |
floorplan_analytics.csv |
Time-series evacuation metrics |
python main.py floorplan.jpg- Building safety and evacuation planning
- Architectural design optimization
- Emergency response simulation
- Research in crowd dynamics and behavior
Sankhyapriyo Dey Computer Science Engineer | Systems & Simulation Developer