Multiple-Agent Task Allocation Algorithm Utilizing Ant Colony Optimization

Abstract

Task allocation in multiple agent system has been widely applied many application fields, such as unmanned aerial vehicle, multi-robot system and manufacturing system et al. Therefore, it becomes one of the hot topics in distributed artificial intelligence research field for several years. Therefore, in this paper, we propose a novel task allocation algorithm in multiple agent systems utilizing ant colony optimization. Firstly, the basic structure of agent organization is described, which include context-aware module, information processing module, the executing module, decision-making and intelligent control module, knowledge base and task table. Based the above agent structure, these module utilize the knowledge in the external environment to process the information in agent communicating. Secondly, we point out that task allocation process in multiple agent systems can be implement by creating the space to the mapping of the multi-agent organization. Thirdly, a modified multiple agent system oriented ant colony optimization algorithm is given, which contain pre-processing steps and the task allocation results are obtained by executing the trust region sqp algorithm in local solver. Finally, performance evaluation is conducted by experiments comparing with Random strategy and Instant optimal strategy, and very positive results are obtained