MRTASim: An agent-based multi-robot task allocation simulation

摘要

In hazardous scenarios with life-threatening risks or in costly or time-consuming situations, autonomous systems must first be tested in a computer environment. This paper introduces MRTASim software developed for multi-robot task allocation experiments. MRTASim uses JADE (Java Agent-Based Development Environment) and allows comparing variations of functions such as path planning, bid valuation, and task scheduling. It displays the results on maps prepared in the environment editor, and automated experiments ensure the accuracy of the results. Academic studies produced with the software, which is constantly updated with new methods and parameters, have been presented at conferences and published in journals. • MRTASim is a software for multi-robot task allocation (MRTA) experiments built on JADE (Java Agent Development Environment). • MRTASim supports path planning (e.g. D*-Lite, RRT, RL), task scheduling, bid valuation, and automated accuracy tests. • MRTASim enables testing of hazardous, costly, or complex scenarios in a simulated environment. • Integration of reinforcement learning-based path planning and advanced Q-learning methods are planned to be implemented.