Traffic Signal Control Agent Interaction Model Based on Game Theory and Reinforcement Learning

Game theory is the best mathematical tool to study human society interaction. Learning approach has an important influence on interaction. The traffic signal control agent (TSCA) interaction model is the basis of research on coordinated control of urban area traffic signal. As for the interactive intersection, this research constructed structure models of two TSCAs, such as intersection agent and management agent. Based on this, the TSCA interactive frame model was established. This research constructed TSCA interaction mathematical model via game theory. In the interaction mathematical model, the renewed Q-values in the distributed reinforcement Q-learning was used to build the payoff values. Therefore, interaction has taken on from the action selection between TSCAs. Next, an algorithm of distributed Q-learning based on distributed weight function is brought forward. The interaction model paves the way for traffic control simulation in the future.