Microscopic traffic simulation: From correlated database construction to strategic layer formation

Abstract

This document details the development of a Virtual Reality (VR) traffic simulator that models flow dynamics at vehicle-level, or "Microscopic," granularity. Plausible vehicle behavior, coupled with rich graphics and intuitive interaction modes, gives the user an accurate sense of realism and provides a tool with unmatched capabilities, whether used for roadway planning, design, or communications. The simulator may operate in either a console mode (for desktop viewing), or in full-immersion mode in a six-sided VR environment. In a departure from other efforts in this class, i.e. Microscopic simulations, the work attempts to capture a fuller range of driving behavior by adopting a hierarchical model of the human decision-making process, i.e., strategic, tactical and control. Specifically, our effort makes the following contributions: proposed a "fuzzy", or knowledge-based, controller to model the strategic layer (e.g., navigation, route planning); upgraded an existing roadway editor by adding new, and enhancing existing, functionalities; and provided a networked run-time environment to contain the microscopic simulation.