Supercomputing and Computation

SHARE

GARFIELD: Graphical Agents Reacting in a Field


Problem Statement:

  • Large scale vehicular mobility simulations are characterized by long-range queuing and behavioral effects.  Control mechanisms and other dynamic phenomena are highly computationally intensive. GARFIELD is a prototype simulation system that incorporates high detail at large scale, executing real-time simulations of US road networks at the scale of a few states (10^6 intersections, 10^8 vehicles).

Technical Approach:

  • Novel field-based formulation is used to exploit  GPU computing power. A canonical regular vector field grid scheme is defined, to which any arbitrary road network can be mapped. Road network control is modeled as a spatially distributed field in which vehicles are immersed. Their movements at each node are influenced by the field.

Benefit:

  • GARFIELD executes simulations in real-time (or faster) on state-sized networks, and enables better insightful analyses and emergency planning through large-scale simulations within short decision time constraints.

 

ASK ORNL

We're always happy to get feedback from our users. Please use the Comments form to send us your comments, questions, and observations.