2020 年 56 巻 7 号 p. 379-385
Stimulated by strong needs for enhancing efficiency in transportation systems, various traffic simulators have been developed so far. Meanwhile, a number of control algorithms optimizing traffic signals based on sensor data acquired in real-time have been reported in the literature. Due to difficulties in implementing these algorithms on real traffic systems, it is reasonable to conduct demonstrations on a traffic simulator. However, existing simulators do not always fully equip functions for real-time feedback control and optimization. We thus develop a novel simulator configuration interconnecting a traffic simulator, UC-win/Road, a control simulator, Simulink, and an optimization solver, Gurobi. Moreover, we install a mechanism based on parallel computing in order to avoid losing the real-time execution due to a large amount of computational efforts for solving a large-scale optimization over a traffic network. We finally implement an optimal traffic signal control algorithm presented in our antecessor on the simulator and demonstrate its effectiveness.