分布式约束满足问题及其在MAS任务分配中的应用

Distributed constraint satisfaction problem and its application to multi-agent system task allocation

针对多智能体系统(MAS)任务分配问题中多个任务与MAS两者的分布式特征,将任务分配问题形式化为分布式约束满足问题(DCSP)进行求解,分别建立了以任务为中心和以agent为中心两种MAS任务分配模型,基于改进的DCSP分布式并行求解算法,提出了基于DCSP的MAS任务分配问题求解框架。该方法适合求解agent间通信有随机延迟以及agent间存在多约束的问题,应用实例的求解表明了其实用性与有效性。

The problem of multi-agent system task allocation was studied considering it to be formed into a distributed constraint satisfaction problem.Both multiple tasks and MAS were distributed,they could build task-centered and agent-centered two kinds of constraint network construction model.Utilized the improved ConcDB algorithm,provided a framework of multi-agent task allocation which based on DCSP.This approach adapts to solution the problem when agents are in uncertain environment or there are a lot of constr...