eadewusic/Traffic-Optimization-Capstone-Project_

This project deploys a trained PPO reinforcement learning agent on Raspberry Pi hardware with push-button inputs and LED traffic lights, demonstrating adaptive traffic control for African intersections to reduce congestion. # Traffic-Optimization-Capstone-Project_Initial Software Product/Solution Demonstration ## **Project Description** This project presents the initial demonstration of an **IoT-Based Adaptive Traffic Control System** that combines **machine learning (PPO reinforcement learning)** with **priority queue optimization** for a four-way intersection. The solution simulates real-time traffic management using **push buttons as traffic inputs** and **LEDs as traffic signals**, enabling a simple yet effective demonstration of adaptive control logic. The PPO model component remains integrated in the backend for inference-based decision-making, while the hardware showcases the system’s physical responsiveness. **Core objectives:** * Demonstrate functional **hardware setup** for adaptive signal switching. * Integrate a trained **PPO model** for lane selection based on queue inputs. * Collect initial **performance metrics** from controlled hardware tests. * Compare PPO-based adaptive switching to a simple fixed-time baseline. ## **GitHub Repository Link** GitHub Repository – IoT Adaptive Traffic Control System (Demo) ## **Environment Setup** ### **Hardware Requirements** | Component | Quantity | Description | | --------------------------------- | -------- | ----------------------------------------- | | Raspberry Pi 4 Model B 2GB | 1 | Main controller running PPO inference | | Push Buttons | 4 | Simulate vehicle arrivals per lane | | LEDs (Red, Yellow, Green) | 12 | Represent traffic lights for 4 directions | | 220Ω or 330Ω Resistors | 12 | For LED current limiting | | Jumper Wires | 1 set | M-M, M-F for connections | | Breadboard | 2 | Circuit assembly | | 5V Power Supply | 1 | Raspberry Pi power sou …