IBM Research and Kyoto University Create System to Simulate Urban Transportation

10 June 2008

Kyoto University and IBM’s Tokyo Research Laboratory jointly have developed a system that can simulate a broad range of urban transport situations—including millions of individual vehicles involved in complex traffic interactions—to predict what will happen if a new office building, sports arena or other major facility is built and improve planning of roads and public transportation.

Led by Kyoto University and IBM’s Tokyo Research Laboratory, with support from the Ministry of Internal Affairs and Communications’ Strategic Information and Communications R&D Promotion Program, the joint research is intended to simulate transport on a large scale using drivers with a variety of driving characteristics and human intentions.

Although similar agent-based simulators have been developed for other applications, this is the first traffic simulation system of its kind. It will address the growing congestion problem by driving transport measures, such as changing mass-transit to provide more trains or buses, optimizing traffic lights route planning to reduce jams, and other long-term solutions, for the entire Kyoto metropolitan area.

The system runs large-scale traffic situations involving millions of vehicles and shows the impact of changes. For example, the simulator can validate what kind of effect a new shopping mall opening or a traffic regulation will have on wide-area traffic.

The system provides the current status of traffic and the alignment of roads to drivers, including current speed and positions of vehicles, the distance between cars, the curvature and gradient of road on which the specific vehicle is running. City planners can use this data to model how each driver will react. In addition, by adding a variety of attributes to the model, the system can simulate traffic conditions with an eye to reducing carbon dioxide emissions and potential accidents.

The Ishida & Matsubara Research Lab at the Kyoto University Graduate School of Informatics created a system which models a variety of drivers, including seniors and young drivers. In order to perform large-scale, high-speed simulations of traffic based on these varied driver models, IBM’s Tokyo Research Laboratory developed the IBM Zonal Agent-based Simulation Environment platform as well as the IBM Mega Traffic Simulator which runs on top of the IBM Zonal Agent-based Simulation Environment platform.

A comparison of simulation results to the actual results of traffic volume made in Kyoto in October 2007 during an experimental program called “Enjoy Walking in the City” conducted by Kyoto City’s Traffic Policy City Planning Bureau showed good reproducibility.

The IBM Zonal Agent-based Simulation Environment provides a large-scale, multi-agent simulation environment that can increase the number of agents (the unit used on a server to indicate individual drivers) from hundreds of thousands to several million under a single processor on a PC server, or to further increase the number under a parallel computing environment. In addition to traffic simulations, the system platform can also be used to perform evacuation guidance simulations, emissions trading market simulations, auction simulations, and other simulations.