Reinforcement learning for traffic signal control in hybrid action space

Haoqing Luo, sheng jin

The prevailing reinforcement-learning-based traffic signal control methods are typically staging-optimizable or duration-optimizable, depending on the action spaces. In this paper,…