Enhancing agent capability in a large simulation system

Abstract

This paper presents an enhanced agent model to Improve the flexibility for creating different agent capabilities in large simulation systems. The simulation system we are improving is called RoboCup Rescue Simulation System, also known as RCRSS. The RCRSS is an environment simulator which simulates natural disasters such as earthquake. Currently the agent model in the RCRSS is not flexible to support mixed agent behaviors. In order to enhance agent capabilities,' we designed a solution based on the enhanced agent mode which is specialized into different agent models with different. behaviors that can be applied in the rescue simulation system. The solution extends the agent development framework called YabAPI to include the Helper Civilian model into the RCRSS. The aim is to simulate situations where agents with multiple roles who are not helpers can help in rescue. Performance enhancement should result in the population with Helper Civilian (HC) agents, which are configured to have partial or full capabilities, over to the population without any extra capabilities. The experimental results show that the scenarios with a configured HC population consistently outperformed a pure civilian population. Performance increased by over 1.00 percent in terms of the overall score, the number of surviving agents, and the overall agent health condition. This enhanced agent behavior shows significant impact in rescue simulations. The range of agent behaviors increases enabling simulation of more complex scenarios than previously possible with more realism .