Recent advances in information and communication technology are expanding the possibilities for developing intelligent transportation systems. Many manufacturers have invested in a vehicle design that features an on-board computer, wireless communication devices, cameras, sensors, and navigation systems that enable real-time data collection about such things as fuel consumption and meteorological conditions. Seeing the possibilities for application of these technologies, a group of Brazilian researchers has developed a computer model to help detect, report, and manage vehicle traffic in big cities. The application, dubbed Incident, is based on exchanges among vehicles over a wi-fi network designed specifically for vehicle networks. The model does not need a connection to the Internet and permits real time exchanges of information about traffic conditions in cities and on highways.
Based on data related to the rate of acceleration and average speed, the on-board computer in each vehicle classifies the degree of congestion along a given route and suggests alternate routes from the actual location to the final destination recorded in the GPS, a satellite geo-location system already available from some manufacturers as standard equipment. “Compared to applications like Waze, Beat the Traffic, and Inrix, Incident, in addition to not requiring an Internet connection, does not depend on the driver to open the app or feed it with information about the intensity of congestion,” says computer scientist Jó Ueyama, a researcher at the Institute of Mathematical Sciences and Computation of the University of São Paulo (ICMC-USP) in São Carlos and one of the system’s developers.
The computer model designed by the Brazilian researchers is supported by a relatively new concept known as Vehicular Ad Hoc Network or VANET. The system involves integration and communication between on-board sensors in cars and fixed components installed along streets, avenues, and highways. Incident is only one of the possible applications within a VANET. In it, each car becomes a wireless router, so that vehicles one kilometer apart can connect with each other, creating a broad, dynamic, and mobile network.
In recent years, VANETs have begun to attract more attention from researchers in the field of Intelligent Vehicular Networks because they are a viable solution for relieving congestion in big cities. In 2014, a study by economist Marcos Cintra, a professor at the Getúlio Vargas Foundation in São Paulo, estimated that traffic jams in the city of São Paulo—because of their effect in terms of wear and tear on materials, accidents, and road maintenance, etc.—cost the economy R$7 billion in 2002, a figure that had increased to R$10 billion by 2012, while the cost of the time wasted in traffic jumped from R$10.3 billion to R$30.2 billion in the same period.
Computer scientist Leandro Villas, a researcher at the Unicamp Institute of Computing at the University of Campinas and one of the developers of Incident, believes that use of VANETs could reduce the costs generated by traffic congestion since they furnish updated, dynamic data about traffic conditions, making traffic flow better. He says, however, that it will take some years for the potential offered by VANETs to be more fully exploited. “Even so, Brazil has the technical ability to produce innovation in software for connected vehicles,” Villas says.
An autonomous system
The results of the first simulations performed with Incident were presented in an article published in August 2016 in the journal PLOS ONE. The system architecture is based on computational techniques inspired by neural structures of intelligent organisms, the so-called Artificial Neural Networks (ANN). “Systems based on ANN are able to acquire knowledge through experience. When exposed to new situations on public roads, the system absorbs the information and makes its own adjustments,” explains computer scientist Rodolfo Meneguette, a researcher at the Federal Institute of Education, Science and Technology of São Paulo (IFSP), Catanduva campus, and principal author of the study.
To test the model, researchers ran the system in programs that simulate vehicle traffic flow, carbon dioxide (CO2) emissions, and fuel consumption based on the acceleration and speed of each vehicle. The programs used two maps as basis: one of Manhattan, in New York, and one of the Dom Pedro I highway in inland São Paulo State. According to Meneguette, the system was able to determine, with more than 90% accuracy, the level of congestion on each of those maps. This, he says, suggests a behavior that is fairly stable regardless of the scenario. Incident also maintained a steady and constant transmission of data to vehicles within a radius of up to 30 km.
The idea, according to researchers, is that the system would be a factory-installed component of the on-board computers found in all vehicles manufactured. The problem is that today, vehicles with that type of technology are very expensive in Brazil. Therefore researchers are working to develop versions of the model that can be installed in mobile devices, such as cellphones and tablets, taking advantage of the GPS receivers. A prototype of the system for these devices is expected to be available for free download by June 2017. In order for the system to work properly in areas that have a comprehensive network of roads, however, certain obstacles will have to be overcome, such as increasing the radius within which vehicles can communicate with each other and integrating Incident into other network technologies—not necessarily the Internet—in order to improve its performance.
A framework for vehicular networks to aid big city management (nº 2015/11536-4); Grant Mechanism Regular Research Grant; Principal Investigator Rodolfo Ipolito Meneguette (IFSP); Investment R$ 27,928.00.
MENEGUETTE, R. I. et al. Increasing intelligence in inter-vehicle communications to reduce traffic congestions: Experiments in urban and highway environments. PLOS ONE. August 2016.
MENEGUETTE, R. I. A vehicular cloud-based framework for the intelligent transport management of big cities. International Journal of Distributed Sensor Networks. V. 12, No. 5. pp. 1-9. May 2016.