Efficient autonomous driving freight trains in bidirectional crossing loop avoiding stops

Abstract

This paper presents an intelligent approach based on software agents able to conduct and coordinate trains in simple railroad stretches, with a main objective: to develop an autonomous driving and coordination of trains system; and two secondary objectives: optimizing the use of the railway and reducing environmental impacts. In the brazilian railways, due to the limited duplication of railways, trains that travel in single sections should make mandatory stops to wait for other trains to use a loop crossing safely. Technological developments resulted in the emergence of new rail traffic control systems. However, rail drive systems that make use of software agents are still poorly explored. In this study we designed a Multi-Agent System capable of simulating the railway environment using conducting agents on trains and also agents at a higher level of control with the process of management skills. The behaviour of the agents was based on specialized driving rules and coordination between them through the exchange of messages, always with the objective of avoiding stops during the trip. Results showed an average reduction of 22.5% in travel time and 25.5% in fuel consumption compared to trips using the traditional method of driving. The reduction in travel and time consumption also results in reducing CO 2 emissions from burning fossil fuels.