Jeddah universities offered the researchers from South Ural State University to jointly participate in an international grant on improvement and implementation of the system for monitoring of pollutants emissions from transport at intersections.
The AIMS Eco system, developed by South Ural State University, is not just an additional device for traffic light system. It is a complex for dynamic monitoring of pollutants emissions from vehicles based on big data aggregation using neural network methods.
Street surveillance cameras in cities allow to monitor the area of the street and road network of up to 40 thousand square metres. What can it help trace and recognize using artificial intelligence? First and foremost, these are the road traffic parameters. How many cars pass through an intersection per unit of time, what type of vehicles, their average speed and trajectory, the time loss at the intersection and the level of congestion. There is also a globally accepted parameter called the Level of Service (LOS), which has a range marked with Roman letters from A to F – from free traffic to chaotic traffic jam.
However, what if the task is more complicated: to provide an accurate assessment of the level of emissions, including carbon and nitrogen oxides, hydrocarbons, soot, formaldehyde, sulphur dioxide, benzapyrene, kerosene, petroleum, as well as the finest PM2.5 and PM10 particles, the sizes of which do not exceed 2.5 and 10 nanometres respectively?
From among the above listed, all that the "naked eye" of a camera would be able to see is maybe just the soot trail.
By the way, after a car passes by, the trail "cloud" remains till it disperses. How can we see and assess it?
Of course, we have to resort to mathematical models. Meanwhile, we must remember to take into account multiple parameters, first and foremost, the atmospheric ones: air temperature, humidity, pressure, and wind speed. The urban housing density is also of importance.
Such models are available, for instance, a method of calculation performed using the Magistral software. It was developed in Saint Petersburg, and a relevant method was approved by GOST in 2019. The Western world actively uses the COPERT4 and OSPM methods and models for assessment of the quantities of emissions from vehicles and their dispersion. Those are based on the collected statistical data, and the recirculating part of the trail (the "cloud" that remains) is represented in the shape of a rectangle. Such models do not always yield reliable results.
The team of scientists from the SUSU Department of Automotive Engineering, lead by Vladimir Shepelev, chose a difficult, but reliable path – modelling of the near-the-ground part of the trail using the Navier-Stokes equations. Aleksandr Glushkov, Associate Professor from the Department of Mathematical and Computer Modelling, and chief research fellow of the Digital Industry Research and Education Centre of SUSU Inna Eliukhina also take part in the calculations.
The system of the Navier-Stokes differential equations describes a most complicated, but vital physical process – fluid or gas motion with consideration to turbulence. By the way, Claude-Louis Navier and George Stokes lived in the nineteenth century, but people still have not learned how to solve their equations in generic form. The existence of a solution and the possibility of extracting a derivative from it is one of the seven unsolved math problems of the millennium, for solving of which the Clay Mathematics Institute promised a million-dollar prize in 2000.
Of course, in certain cases and along with using artificial intelligence, it is possible to solve such equations. The SUSU researchers keep "storming" through the variants. The result is used as a base for the calculations performed by the AIMS Eco system.
The AIMS Eco system consists of several modules. Module one calculates the traffic at an intersection (and can work using only one camera). The second one calculates the amounts of emissions, the third one assesses the effectiveness of using the transport infrastructure (LoS, mentioned above). Module four regulates the traffic light phases in an online mode. And finally, module five – the analytical module – studies the traffic flow, evaluates the emissions levels and provides the previous module – the regulating one – with data on how to adjust the traffic lights, so that there are less traffic jams and vehicles do not stay at intersections for too long, polluting the air.
Such AIMS Eco system has already been installed in several cities. On the SUSU website, using this system in an online mode, you can monitor the traffic and emissions from transport at three intersections in the city of Perm and two intersections in Chelyabinsk. This is a demo version. In our city, it monitors the condition of air on the square before the main SUSU building and the crossroads of Voroshilova Street and Komsomolsky Prospekt.
Based on data from 4 cities, SUSU Associate Professor Vladimir Shepelev has managed to obtain interesting facts. For example, the significant part of emissions – up to 40% of the total amount – comes from buses. "One of the causes is that public transport has the lowest speed parameters and the longest delays at regulated intersections as compared to other types of vehicles. The extraction, interpretation and aggregation of big volumes of heterogeneous data with the use of artificial intelligence provides transport engineers with an additional tool in their solving the task of ensuring the "prioritization" of public transport and formation of sustainable intelligent transport systems," explained Vladimir Shepelev.
Finally, when using any digital technology, we always ask ourselves a question: could we do the same with analogue technologies? "Standard" detectors, which measure the emissions levels using the familiar methods of physics and chemistry?
Vladimir Shepelev and his colleagues performed such a comparison with the help of a mobile environmental monitoring station – a vehicle equipped with analogue devices for assessing the meteorological conditions and atmospheric pollution. SUSU ecologists have such a mobile laboratory at hand.
The simultaneous calculation of the emissions concentrations using computer vision (AIMS Eco) and the laboratory-based measurements showed a 75-82% agreement of results. This is a very good indicator.
At the same time, the deployment and maintenance of one AIMS Eco virtual station is 3-5 times cheaper than the use of analogue detectors and exceeds the functional capabilities many-fold!
While the SUSU researchers are publishing their articles in the Mathematics journal and other international scientific editions, this system has been implemented in the city of Magnitogorsk, thanks to the support from Magnitogorsk Iron & Steel Works (MMK). This city-forming enterprise takes care of the residents' health and is implementing an ecological perimeter in the city. At the first stage, 12 stations are already functioning at the main city intersections, and at the next stages it is planned to cover the whole street and road network of the city. The plant is ready to assess the data on the emissions from transport in order to regulate the emissions of the plant itself and prevent the reaching of the maximum permissible concentrations (MPC).
Three digital environmental monitoring stations have been installed in the city of Perm, with the assistance from the local Committee for the Use of Natural Resources. There are also stations in Chelyabinsk. Here, Intersvyaz and Rostelecom telecommunication companies act as partners of the AIMS Eco system's designers.
In the meantime, this system sparked interest in Saint Petersburg, the United Arab Emirates, and Saudi Arabia. There is a chance that international partners might support the implementation of the environmental stations with their government grants. And this means that there is a promising future for the SUSU project on digital emissions monitoring AIMS Eco, and not only in Russia, but probably around the world as well!