SIGNATURE WORK
CONFERENCE & EXHIBITION 2022

Solving highway traffic light control issues will be an important part of the operation of ensuring traffic order, safe, and fast. However, existing traffic signal light control systems are single fixed timing control, and cannot be adjusted according to the actual traffic conditions. In this project, we study and implemented this paper: “A Deep Reinforcement Learning Network for Traffic Light Cycle Control”. This paper adopts Deep Q learning method to deal with traffic light control at an intersection, which exceeds the traditional method in performance. We replicated the conclusions of the paper on the data simulated by SUMO, but for the dueling DQN technique, we got a result that could not converge, which is different from the original paper and we will continue to study it.