Simulating a system of agents that navigate in a physical space is a challenging problem when the environment is sophisticated and the number of agents involved is large. We describe experiments in establishing a “virtual Prolog engine” in each agent in a simulation and the scalability and performance issues with such an approach. We report on experiments with a simple predator-prey animat model and discuss how this approach allows us to impart a degree of reasoning and intelligence to what are otherwise very simple animat agents.
Kenneth A. Hawick, Heath A. James, C. J. Scogings