Parameter Selection for “PCMA*” Using Surrogate-Assisted Black-Box Optimization and Microscopic Traffic Simulation

Routing algorithms have become significantly more sophisticated in recent years as increasingly more data on road networks, vehicles and drivers becomes available. The correct parametrization of such algorithms is a difficult task since the underlying traffic model needs to reflect the complexity of the algorithm and should resemble reality closely. A major drawback of such models is that evaluating a certain parameter setting may become computationally expensive, preventing the use of conventional optimization algorithms. In this work a combination of surrogate assisted black box optimization and microscopic traffic simulation is used to optimize the parameters of a recently published routing algorithm.