Multi-Modal Traffic Simulation Calibration and Integration with Real-Time Hardware in Loop Simulator

Abstract

The ongoing research in intelligent transport systems and connected and automated vehicles, enabled by advancements in artificial intelligence, integrating traffic simulations has become an essential part of product/software development for the automotive industry. Nowadays, traffic simulations are used to mimic real-world environment scenarios for virtual testing of advanced transportation technologies. With the increase in data collection methods for traffic flow, the calibration of the microscopic traffic simulations has emerged as an important research area. The underlying question in traffic modeling is how accurately simulations can mimic the real environment traffic flow conditions? This paper attempts to create a framework for microscopic traffic simulation calibration procedure which can be scaled for large networks. This paper makes the following major contributions. First, a calibration framework is proposed which harnesses the existing data set collected from The Ohio State University (OSU) campus bus service (CABS) busses using Global Positioning System (GPS) sensors to determine the traffic state in the real environment and create a microscopic traffic simulation. The traffic simulation is implemented for a section of the OSU campus (“Woody Hayes Drive") in an open-source traffic simulator – Simulation of Urban MObility (SUMO). The traffic flow generation is probabilistic to introduce variability between scenarios. The second contribution is the development of a communication interface between real-time dSpace ASM Hardware in Loop setup with SUMO to create a complete real-time simulation of urban environments for advanced driver assist systems (ADAS) virtual testing. Ademonstration scenario is the Ohio State University campus network with traffic demand generated using the calibrated model from the first part of the work.