Self-learning of the containers service coordinator agent in multi-agent automation environment of transit cargo terminal

Abstract

The article deals with some problems of transport logistics, concerning the improvement of the organization and automation of basic processes of transit cargo terminal. The terminal operates with three-dimensional blocks which contain packaged goods: it receives arriving blocks, temporarily stores and sends them to the customers. Blocks are transported on trucks in a receptacle of limited size, conventionally called a container. When the resources of loading and unloading of containers, transport and storage are limited, and there are not enough some ordered blocks, queues of containers waiting for loading and unloading may occur. The authors applied multi-agent approach to the terminal management: the work is distributed among the four agents: containers unloading agent, warehouse agent, containers loading agent and the main coordinator agent. A new function of coordinator agent - self-learning based on the results of its previous work - is presented and described in the article. Self learning is an important property of intelligent agent. This property can contribute to increasing the effectiveness of using the agents for the organization and automation of transit cargo terminal.