Co-Simulation of Human and Machine Generated Drivers on a Single Test Track

Abstract

With autonomous vehicles in the early phases of road testing, we can surely expect them on the road in a few years. But the existence of a mixed tra c with human-driven vehicles will be inevitable for a long time after the introduction of autonomous vehicles and the rami cations of such a tra c are hard to foresee. But by creating a simulation which involves both kinds of vehicles, we can study their behavior and analyze the e ect the autonomous vehicles has on the human-driven vehicular tra c. Using SUMO, which is a microscopic tra c simulator, the tra c of autonomous vehicles can be replicated by using machine generated vehicles whose speed and position in each simulation step is controlled by SUMO. This thesis tries to interface SUMO and Unity, a powerful cross-platform game engine, by using the Tra c Control Interface (TraCI) protocol and in doing so, a car was generated in Unity which can be human controlled and driven in the same track as the machine generated vehicles. This paper reports the details of this co-simulation using a real life test track which facilitates a comprehensive analysis of interactions between the vehicles in the real world.