Multi-agent programming contest 2017: BusyBeaver team description

作     者：Pieper, Jonathan 

作者机构：Tech Univ Clausthal Clausthal Zellerfeld Germany 

出 版 物：《ANNALS OF MATHEMATICS AND ARTIFICIAL INTELLIGENCE》 (Ann. Math. Artif. Intell.)

年 卷 期：2018年第84卷第1-2期

页      面：17-33页

核心收录：

学科分类：08[工学] 0701[理学-数学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Multi agent programming contest 

摘      要：This paper describes the team BusyBeaver, that participated in and won the Multi-Agent Programming Contest 2017. Its strategy is based on dividing agents into three static groups modeling the work chain of buying, assembling and delivering items. The team is coordinated by a centralized agent doing most of the high-level planning, usually using greedy algorithms and specialized heuristics. There is a heavy focus on proactively buying and assembling some items, in order to quickly complete upcoming jobs.