In this article, the authors show their current work on the relationship between local behaviors of agents and global characteristics in a multiagent NBA simulation, which is called RoboNBA. Two sources of different local behaviors are introduced: Decision-making mechanisms and agent strategies. The global characteristics consist of global performance and global patterns. The authors address the problem of how to quantitatively measure the global performance and global patterns in RoboNBA. For global patterns, they focus on the diversity of attack patterns of a team. Through experiments and analysis, they try to examine how agent local behaviors can lead to different global performance and interesting global patterns in RoboNBA.
Scopus Subject Areas
- Modelling and Simulation
- Computer Graphics and Computer-Aided Design
- global patterns
- multiagent systems