CI-EN: Enhancement of a Large-scale Multiagent Simulation Tool

George Mason University, Fairfax VA

Investigators

Abstract

An agent-based simulation is a software simulation of many independent actors (people, robots, animals, companies, etc.) interacting in complex ways. For example, one might build a simulation of a swarm of ants, a school of fish, a large group of robots collectively building a house, a city of people under siege by a large medieval army, or the spread of urban legends over a social network. These kinds of simulations are used for everything from building software for swarms of delivery robots, to understanding the spread of disease in third-world slums, to predicting the impact of climate change on human migration patterns. Similarly, these simulations help researchers and policy makers in engineering and robotics, in artificial intelligence, and in the biological and social sciences. Such simulations can get very large, with large numbers of actors. This project involves building a software tool to assist in the development of large-scale agent-based simulations spread over potentially many separate computers, and to make it easy for them to place their agents in simulated locations on the Earth. At the same time the tool will be easy to use for high school and undergraduate students. The project will develop a distributed agent-based modeling tool for constructing large simulations of swarms and groups of agents. The project enhances an existing, successful open-source Java multiagent simulation library called MASON. MASON is designed to run on a single machine; but the enhanced version will also allow distribution over many machines. The enhanced tool will also include facilities for embedding agents in geographical information systems (GIS), model optimization and automation, interfaces for alternative programming languages targeting the Java Virtual Machine, statistics facilities, internal testing and verification, and integration with software tools. The enhancement will not only provide a distributed simulation facility for researchers in the social sciences, biology, and engineering, but will also improve on MASON's graphical interface and language facilities to make it easier to use as a teaching tool.