A GPU-based Framework for Large-scale Multi-Agent Traffic Simulations

In order to improve the reproducibility of real situations, agents should respond to dynamic environmental changes, as well as considering efficient computation of them since the simulation often becomes huge scale. In this paper, an approach and the basic architecture for GPGPU-based efficient and scalable framework is presented, by applying OpenCL-based multi-platform agent code conversion engine. We present a prototype implementation of the framework to easily test and try the implemented path planning codes in various settings.