Traffic Signal Control

Chung-Ang University Researchers Develop a Meta-Reinforcement Learning Algorithm for Traffic Signal Control

Existing traffic signal controllers are often inadequate in handling traffic congestion and reinforcement learning (RL) has been adopted to solve this problem. However, RL usually works in a stationary environment, while traffic environments are nonstationary. Researchers have now developed a meta-RL model that adjusts its goal based on the traffic environment, maximizing vehicle throughput during peak hours and minimizing delays during relaxed traffic flow. The algorithm has a wide-area coverage and outperforms existing alternative algorithms.