Combining adaptation at supply and demand levels in microscopic traffic simulation: a multiagent learning approach

Abstract The effects of traffic congestion can be mitigated with a range of different methods. This paper addresses multiagent reinforcement learning (MARL) as a contribution to this effort. Specifically, while most of the literature on MARL applied to traffic assumes that either only traffic signals learn (while drivers do not change their behaviors) or vice-versa, we consider simultaneous learning (co-learning) by two different classes of agents. In addition, experiments were performed in a microscopic simulator so that the agents’ behavior can be as fine grained as possible. Results show that the co-learning approach outperforms other policies in terms of average travel time.