Andalusia rehearse with a new traffic monitoring system

Monitoring systems of traffic which have been developed in recent years, are based mostly on images from Visual cameras or centralized sensors. For this reason, have limitations due to the variability of the lighting conditions during the day and night, the presence of phenomena that limit the visibility as fog, rain or smoke, and the complexity of the stage with forks, tunnels or changes of direction.

For this reason, these experts proposes a system of perception based on multiple clustered devices, like cameras with different multispectral bands (i.e., to detect different parts of the spectrum) totaling nodes that contain different sensors distributed across the road. This integrated mechanism will make it possible to obtain accurate 3D information of mobile objects, even before environmental conditions and adverse lighting. "The increase in the number of sensors increase reliability, and makes a weighting between them, depending on the confidence that is given to each." "In addition, is covered across the road and it can be monitored the State, i.e., if it is affected by rain or snow or if you have poor pavement", explains the Coordinator of the study José Ramiro Martinez of God, of the University of Seville.

The researchers, belonging to the Group of robotic vision and Control of the University of Seville with the engineering of systems and automation at the University of Malaga tested your device on a test track at the end of the year. They will also address the detection of risk situations such as blockades of roads and accidents, the location of aggressive behaviour and traffic monitoring. In this way, the data collected by the techniques of perception proposed could be integrated in models of traffic management, enabling a dynamic perception. "For example, devices will allow us to detect an accident or unforeseen increase of traffic on a road and management model be adapted dynamically replanificando traffic lights or indicating alternative routes." "This approach represents an advantage over traditional methods of traffic management that are based on static statistical data and that, in general, not they can be considered unforeseen".

The system, developed in the framework of a research project of excellence that has encouraged the Ministry of economy, innovation and science with 103.800 euros, combines cameras and sensors. With regard to the first, researchers have experienced multi-spectral vision systems through the integration of cameras in the visible and infrared cameras, allowing for night vision. These vision systems make it possible to obtain images of the same scene in different spectral ranges, i.e., allowing objects that could not be by any of the spectral bands of separately. "For example, using multispectral camera in the near infrared and far infrared bands you can see through the fog, which cannot be achieved by any of the two spectral bands of independently", says Martínez.

The perception obtained from the Chambers will be integrated with data from different sensor networks with the aim of spatial information of the objects identified in the multispectral images.

A set of nodes communicated among themselves will be displayed in the project. These points include sensors of luminosity, of rank and presence and even small low-resolution cameras, but allowing to distinguish colours and detect movement. All these devices joins a wireless system that allows communication between the nodes. "They tend to be low-cost, low energy, small size and protection to work in adverse conditions," qualifies the expert.

This network of points of detection allows individualized follow-up of each car, as the information is transmitted from sensor to sensor along all the way. Thus, if a node located a car at a point and recognizes parameters such as its size and speed, the following receives the data from its closest counterpart and infers that it is the vehicle that detected the first.