Quick Start

AI Traffic Control API - 5-Minute Setup Guide

1. Select Your Trained Model

python select_model.py

This will show you all available trained models and help you copy one to the API.

Available sweeps: pressure, queue, diff-waiting-time
Locations: C:\Users\gemer\Sumo\my-network\Results\sweeps*\

2. Start the Services

# Windows
start.bat

# Linux/Mac
./start.sh

3. Test the API

# Get a traffic action
curl http://localhost:8080/api/traffic/action

# Or run the comprehensive test
python test_api.py

---

API Endpoints

Python FastAPI Service (Port 8000)

• GET /health - Health check
• POST /predict_action - Action prediction
• GET /model_info - Model information
• GET /docs - Interactive API documentation (Swagger UI)

Java API Gateway (Port 8080)

• GET /api/traffic/health - Health check
• GET /api/traffic/action - Get traffic action (auto-generated observations)
• POST /api/traffic/action - Predict action with custom observations

---

Example Requests

Get Traffic Action

curl http://localhost:8080/api/traffic/action

Response:

{
  "predictedAction": 2,
  "signalState": "GREEN",
  "timestamp": 1710521600000,
  "status": "success"
}

Custom Observations

curl -X POST http://localhost:8080/api/traffic/action \
  -H "Content-Type: application/json" \
  -d '{
    "observations": [0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]
  }'

Health Check

curl http://localhost:8080/api/traffic/health

---

Architecture

Flow: HTTP Request → Java Gateway (8080) → Python FastAPI (8000) → RL Model

Components: 1. Java Spring Boot Gateway - REST API entry point 2. Python FastAPI Service - RL model inference 3. Trained PPO Model - Action prediction

Docker Commands

# Start services
docker-compose up --build

# View logs
docker-compose logs -f

# Stop services
docker-compose down

# Check service status
docker-compose ps

---

Troubleshooting

Services won't start

1. Ensure Docker is installed: docker --version
2. Check if ports 8000 and 8080 are available
3. Check logs: docker-compose logs

Model loading fails

1. Verify model file exists: rl-inference-service/app/trained_models/model.zip
2. Rerun model selector: python select_model.py
3. Rebuild: docker-compose up --build

API returns errors

1. Check Python service logs: docker-compose logs rl-inference
2. Verify model is loaded: curl http://localhost:8000/model_info
3. Test health: curl http://localhost:8080/api/traffic/health

---

Next Steps

1. Integrate with SUMO: Modify SUMO simulation to call the API
2. Custom Observations: Replace dummy observations with real traffic data
3. Production Deployment: Use Kubernetes or cloud platforms
4. Monitoring: Add logging and metrics collection

---

Files Overview

File	Purpose
select_model.py	Interactive model selector
test_api.py	API test client
start.bat/.sh	Quick startup script
docker-compose.yml	Service orchestration
rl-inference-service/	Python FastAPI service
java-api-gateway/	Java Spring Boot gateway
README.md	Full documentation

---

For detailed documentation, see INDEX.md