Simulation-Based Bidirectional Control between AVs and Adaptive Traffic Signal Control

📌 Project Overview

• Context: Master’s Thesis at TUM Asia & R&D Project with IABG Industrieanlagen-Betriebsgesellschaft mbH
• Location: Mobility Innovation Campus (MIC), Munich, Germany
• Keywords: Autonomous Vehicles (AVs), V2X Communication, SUMO, TraCI, Reinforcement Learning (DQN), Adaptive Signal Control.

🚦 Core Innovation & Methodology

1. Bidirectional Coordination Framework:
   • Integrated connected autonomous vehicle trajectories with roadside unit (RSU) sensors.
   • Enabled real-time bidirectional negotiation between oncoming AVs and traffic signal controllers.
2. SUMO/TraCI Microscopic Simulation:
   • Built high-fidelity microscopic simulation environments mirroring the MIC test site geometry and real-world V2X message flows (SPaT, MAPEM, CAM/BSM).
   • Automated co-simulation pipelines using Python and TraCI APIs.
3. DQN-Based Adaptive Signal Control:
   • Formulated signal phase selection as a Markov Decision Process (MDP).
   • Designed reward functions balancing overall intersection throughput, individual vehicle delay, and queue length.
   • Validated robustness under varying traffic demands (peak/off-peak) and mixed V2X penetration rates (0% to 100%).