Agent-based Clustering Approach to Transport Logistics

Abstract

Evolution of autonomous cooperation pulls away the traditional centralized approaches towards the decentralized approaches in logistic networks. In this concept, knowledge and decisions are distributed among the participants of the logistic process. Vehicles and goods become intelligent, interactive, and capable of deciding about how to achieve their aims. Software agent technology provides a means of bringing about autonomy by information sharing and decision-making capabilities. This paper presents the approach of integrating agent technology and knowledge management approaches like clustering techniques to ensure robust and efficient planning and scheduling in the transportation domain. Logistic entities are represented as software agents, where the objective is to cluster these entities which have common goals - like packages having the same destination, same type of packages, etc. The approach of autonomy through software agents and clustering techniques is expected to significantly decrease the communication demand imposed upon the logistic network for a set of required tasks to be performed. An enhanced clustering algorithm has been applied on a logistic scenario and compared with the original algorithm in terms of effective cluster formation with less iteration. This approach identifies challenges in the area of communication that arise from the distributed decision process and the interacting components.