Study on Traffic Signal Control Based on Q-Learning
Authors:
- Yongquan LiaoYLYongquan Liao
- Xiangjun ChengXCXiangjun Cheng
- 1 August 2009
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3, 581-585
Abstract
In order to reduce the delay of vehicles passing through junction, the signal timing of agent controlled intersection was optimized by Q-Learning approach. On the basis of fuzzy rule set, the effect of signal control was improved through optimizing the combination of control rules with Q-Learning. The result of simulation illustrates that the signal control method based on Q-Learning is better than fixed-time control, actuated control and signal control based on genetic algorithms. The result of this research indicates that the signal control method based on Q-Learning is adapted to the urban traffic control.Keywords
This publication has 4 references indexed in Scilit:
- Reinforcement learning in neurofuzzy traffic signal controlEuropean Journal of Operational Research, 2001
- Using intelligent agents for pro-active, real-time urban intersection controlEuropean Journal of Operational Research, 2001
- Introduction: The challenge of reinforcement learningMachine Learning, 1992
- Technical Note: Q-LearningMachine Learning, 1992