Imprimir Republish


Light at the end of the road

A simulator takes into consideration the personality of drivers in order to control the traffic

The strong connection between men and their cars indicates that traffic jams have come to stay. But, there are researchers who believe that it is possible to put order into the chaos of major city traffic though the help of artificial intelligence. At the Federal University of Rio Grande do Sul (UFRGS), a research group, headed by Ana Lúcia Bazzan, a professor at the Information Technology Institute, has developed a traffic simulator making use of free software, which will be made available to cities administrations and the academic community. The technology is being tested in the city of Porto Alegre. Some data concerning the city’s traffic – topography and the counting of vehicles – were used to test the tool.

Firstly data is collected, such as rate of traffic flow, the number of road crossings and the location of traffic lights, as well as their operational plans. The work of the simulator depends, in the first instance, on selecting elements of specific points and to calculate what is the distribution of the vehicles within the transport network. This is nothing new: many cities make use of instruments to make this information available. The next step is to make use of the resources of artificial intelligence, used both for the coordination of the traffic lights according to traffic flow – the well-known “green waves” – and to identify the models of taking a decision that the motorists use and take them into account at the moment of forecasting the traffic. This is the most complicated part since the drivers don’t all behave in the same manner, as many of the available commercial systems suppose.

The simulation and control models used in major cities do not take this into consideration. “In general, they analyze only the vehicles in motion, and not the human beings who are driving them. This is because the existing models are usually too heavy to carry out simulations of this nature in real time”, says Ana Lúcia. Traffic control systems in real time have existed since the decade of the 70’s, but only recently has cost reduction allowed their dissemination. In Brazil, the advances over the last decade were achieved with expensive and imported systems, such as those used in the cities of Sao Paulo and Fortaleza.

Games theory
The research in which professor Ana Lúcia participated in Germany between 1999 and 2002 was the key to understanding this equation. Thanks to a partnership between the UFRGS and the Physics Institute of the University of Duisburg-Essen, in the area of the project Simulation of Social Agents in Traffic (Sociat), professor Ana Lúcia got together with the group of professor M. Schreckenberg, co-author of a micro-simulation traffic model, in order to introduce the variable of motorists’ behavior into this model. Collaboration involved various projects. In Germany the tripartite project named Survive emerged, which had the participation of Reinhard Selten, from Bonn University, the winner of the Nobel Prize in 1994 for his contribution in the area of games theory. The project carried out experiments with human guinea pigs in situations of choosing a route with the object of extracting behavioral standards.

The result, as well as the development of a simulator, contemplated the study of the diverse types of motorist personalities in the scenario of choosing a route. In Brazil the project Integrated Simulation, Control and Optimization of Traffic System (Siscot) was proposed, a project directed towards developing basic simulation infrastructure, anticipating the extension of automaton rules in order to accommodate various types of drivers. Starting from 2003, with the end of the Siscot project, activities continued through the Information and Control System and Urban Mobility Control (SincMobil) project, coordinated by researchers from the Federal University of Santa Catarina. The Sociat, Siscot and SincMobil projects were financially supported by the National Council for Scientific and Technological Development, the CNPq.

The technology is being successfully used in the prevention of traffic jams on German highways, in and around the city of Cologne. The traffic engineers usually classify traffic conditions as fluid, congested and an intermediary situation known as synchronized flow. The model’s novelty was to consider the nuances of personality, that include at one end the aggressive motorists, who constantly change lanes, to those excessively defensive or who brake for any reason – accelerations and decelerations, as is well known, interfere in fluidity. The model included the variable of behavior in its calculations and this is done by way of technologies in the area of agents and multi-agents systems – a field of artificial intelligence that looks, among other things, to coordinate autonomous agents that possess knowledge and limited skill. In the case of the simulation model, the agents are both the motorists and other elements of the scenario – in particular the traffic lights that must be coordinated in order to form green waves or other forms of control.

These models function both in urban areas and on major highways. On highways, since there are no traffic lights, the objective is not so much to control but to forewarn: the system at Cologne, via the internet, informs on the traffic conditions over the next hour with 90% accuracy. The experience has begun to be a victim of its own success. With thousands of people visiting the site every day, changes in the motorists’ standard of behavior have been noted, as a risk towards compromising the forecasts.

This is the main limitation of any system of ordering the traffic. “They stop doing well when people attempt to get around them”, says Ana Lúcia. “It’s that tale about the traffic jams at dawn on the highways at the start of a holiday, when many motorists had the same idea of waiting until the next day to escape from the traffic”, she says. Since the decade of the 70’s, radio stations have begun providing information on the traffic for drivers. As everyone has access to the same information, the capacity to help has been neutralized. The greater and greater number of cars, which is the origin of traffic problems, has neutralized major road reforms in the metropolises carried out between the 1960’s and 1980’s and now conspires against the efficiency of whatever attempt at organizing. The national car fleet is estimated at 29 million vehicles. In 1970 it was 3.1 million.

“In the urban universe or on the highways, an absolute impersonal system of signaling can only operate efficiently if its users follow the norm of equality in the face of the law, and, through this, respect the movements of each other. The essence of traffic control, like all good rules founded on individual equality, is a code that privileges the right of whoever arrives first”, observes the anthropologist Roberto da Matta, in the article “Traffic, equality and hierarchy” published in the newspaper O Estado de São Paulo . For Da Matta, the rates of death on the roads in the country result, to a large degree, from the national custom of challenging these regulations. “In Brazil, to obey the rules in this day to day competition is something bothersome and almost always offensive.”

This aggressive behavior, which is not exclusively Brazilian, is being taken into account. It’s for this reason that the traffic is treated as a multidisciplinary question. Principally in Europe, various measures have been proposed under the form of Intelligent Transportation Systems (ITS) and Advanced Traveler Information System (Atis). “They’re searching to provide to the user of the transport systems dynamic information with respect to routes, parking, accidents, which will help to diminish congestions and aggression”, says Ana Lúcia. “These measures are added to the traditional methodologies such as those that the transport engineer deals with, but do not dismiss a multidisciplinary team such as that of the Survive project, in which physicists, computer scientists and economists, linked into the area of individual behavior, act in conjunction to deal with the question of choosing a route.”

Open code
The exchange between Brazil and Germany has rendered other fruit, such as the visit of researchers from professor Schreckenberg’s group and from the University of Würzburg to Brazil, a doctorate thesis in Germany and two masters’ dissertations by graduates from the UFRGS. These last mentions students are starting their doctorate degrees this semester – one student at UFRGS and the other in Germany. The simulator is being licensed as an open code for distribution to other research groups. A graduate from the Polytechnic School of the University of Sao Paulo, a master in computer science from the UFRGS and a doctor in the area of multi-agent systems from the Universität Karlsruhe in Germany, professor Ana Lúcia laments that there does not exist, outside of engineering, many researchers in the country interested in this field of knowledge. “Few people in the areas of psychology and sociology interest themselves in this question”, said professor Ana Lúcia, who also works with other applications of artificial intelligence and multi-agents systems, such as bioinformatics for genome annotations.