Implementação de um sistema multiagente com mecanismo de negociação descentralizado para um estacionamento inteligente

Abstract

Find a free parking spot is one of the major problems faced every day by the cities population. Usually, in times of great traffic flow to find a parking spot, a driver will waste time or will drive a long way until find a spot. Therefore, wasting fuel and generating traffic jams. In order to shed some light into this problem, the MAPS project (Multi-Agent Parking System) aims to develop a smart parking solution based on Multi Agent techniques. This Multi-Agent system to allocate parking spots was developed using the JaCaMo framework and has a centralized negotiation mechanism, where a figure of a central agent controls the spot allocation, this imposes some limitations to the system because if this central agent fails the whole system is compromised. The work here presented shows the deployment of a Multi-Agent systems based on a decentralized negotiation mechanism, where the agents can freely negotiate the parking spots without the so-called central agent. The MAS described here is named MAPS-OPEN, since it is based on an open MAS, where agents can freely communicate and negotiate. The MAPS-OPEN defines a reasoning model and a negotiation protocol, where the JaCaMo agents can be assigned to different roles (buyer or seller) in order to deal with a parking spot. Moreover, we present some experiments in order to show how our MAS works in a simulated environment. Where it was possible to verify that the number of messages exchanged in the system grows according to the quantity of agents, and that with this growth of the number of agents make they tend to park closer to the point where they wanted to park.