Multi-Agent Control of Traffic Networks: Algorithm and Case Study

As more and more traffic control instruments are installed, the probability increases that conflicts will arise or that coordination opportunities are lost when various traffic control measures are applied in the same area. Traffic control instruments can thus no longer be considered separately, but have to be considered as part of a larger network. By modeling the separate measures as intelligent agents, the actions of the individual instruments can be coordinated. By coordinating neighboring traffic control instruments, it can be prevented that they impede each other and one can realize that they function better as a whole. This paper proposes a procedure through which coordination between traffic control instruments can be achieved. We also illustrate the benefits of this multi-agent coordination approach for a simulation-based case study for the N470 arterial near Delft, The Netherlands, in which the new coordination approach is combined with a recently proposed block-based look-ahead traffic-adaptive control algorithm.