Abstract
The RoboCup Soccer environment is one of the most difficult scenarios for autonomous agents. With the potential for so many things to go wrong, debugging and analyzing agents’ behaviors becomes a significant task. We propose RoboViz, an open-source program for integrating agent-driven visualizations into a real-time, 3D rendered environment; the scene becomes a shared, interactive whiteboard for all agents, and the user can moderate by filtering drawings they are interested in. Visualization is an effective tool for tracking down errant behaviors and explaining algorithms. RoboViz is embraced by the RoboCup Soccer Simulation 3D sub-league as the de facto monitor application, and the latest revision makes it useful for other leagues as well. We are currently testing RoboViz in the Standard Platform League (SPL).