Reactive Flock Traffic Navigation: The Design of an Emergent Behavior Based on Social Potential Fields-Edición Única

Abstract

Urban traffic management is a recurrent problem in big cities nowadays, due to different issues like the increasing number of cars that enter the city roads, making the demand greater than the roadway’s capacity. Therefore, the need to control urban traffic arises with high priority, in order to decrease side effects like high fuel consumption resulting in high vehicle emissions. Currently, traffic is mainly controlled by traffic lights, static signs and electronic boards, which provide important information about traffic flow, accidents or other related data. Besides, in some places, there have been introduced more advanced schemes like intelligent traffic signal coordination programs; however, this approach brings only 1% of delay reduction [25], which does not represent an efficient solution. Additionally, there are several treatments that are aimed to gain more benefits from the infrastructure and often lead to very high delay reductions, but some of them (the overpass, for instance) are very expensive. In recent years, some other high-level approaches have been proposed, making use of Multiagent Systems (MAS) techniques, which have proven to be very efficient and promissing regarding their evaluation results. One of such innovative mechanisms is the “Flock Traffic Navigation based on Negotiation” method (FTN) [4]. Inspired by nature, this method proposes a mechanism for vehicles to gather up into groups, in order to get a speed bonus that enables them to travel faster, thus reducing congestion levels. The present research work explores a reactive approach to the FTN model, which aimes to overcome specific issues detected on this mechanism, as well as to serve as an alternative solution which will enable an interesting comparative analysis between a deliberative solution against a reactive one. Furthermore, this might lead to the design and development of a hybrid urban traffic management method. The proposed solution, called Reactive Flock Traffic Navigation (rFTN, for short), describes an emergent behavior, which is embedded in vehicles and designed to make them navigate under the FTN paradigm. In order to achieve this, several mechanisms have to be performed by each car, within the rFTN calculations. Each one of this workings are explained. This document also presents experimental data, which is aimed to both explore and validate the performance of the proposed solution model. In this phase, the rFTN was compared against traditional and advanced methods for the traffic management problem.