| This research intends to design and develop a traffic simulation platform for the purposes of research and education. After briefly reviewing traffic simulation and existing simulation approaches, the general procedure of conducting traffic simulation from the perspective of a transportation researcher is analyzed. With the specific needs in data requirements, modeling, etc. addressed, state-of-the-art software engineering design patterns such as Model-View-Controller, inversion of control, etc. are introduced to support such a simulation platform. Data structure and architecture are thereby proposed and implemented.;A highly open and flexible data structure is presented, along with a compact GIS model for traffic simulation that separates geometry and vehicle movements. A plug-in framework that supports such flexibility is devised with the intention to allow user to inject and customize simulation models efficiently.;Six algorithmic models are extracted out of generic traffic simulation procedure. They are simulation model, lane-changing model, car-following model, vehicle propagation model, random generation model, and routing model. Each of these models are replaceable plug-in in the platform to execute traffic simulation. Together with two non-parameter attributes functions, dynamic vehicle and driver properties, they consists the core part in running simulation on the platform.;TripSim (https://www.github.com/tripsim/trafficsim) are created as reference implementation. |
