The multi-agent programming paradigm use for railway applications

Abstract

As railway signalling systems hardware improved from mechanical switches and electrical relays to electronic tubes and semiconductors, the software improved from direct control and event-driven approach to object and agent-oriented programming. This paper aims at the use of agent-oriented programming for railway applications. In order to do that some agent basic concepts such as organizations (reactive and cognitive), models (beliefs/desires/intentions and social) and interactions (independent and collaborative) between agents are presented. The AOSE (Agent Oriented System Engineering), as OOSE (Object Oriented System Engineering) some time ago, is a paradigm on multi-agent programming. The main difference between an agent and an object is that the agent must have autonomy while an object must not. An object property is only activated when a control program asks for that property. An agent, instead, must start and manage its own properties in order to perform its objectives. Therefore, a model proposal for train control systems is presented based on multi-agent programming paradigm. In this proposal, trains and stations are modelled as interacting agents, working together in order to achieve a common goal: an optimized train traffic control. Some profits and drawbacks of this approach are also addressed, detaching the importance of multi-agent approach characteristics for railway and train control applications. The AOSE supplies many tools for multi-agent systems design and programming.