Scientists from Kangwon National University in South Korea developed a technology of road bed inspection allowing for improving the traffic situation and cutting down the energy consumption. It is based on the computer vision system installed at the moving vehicle.
Usually road inspections require measures, which make the drivers’ lives more complicated: blocking the lanes, installation of temporary traffic signs and narrowing the carriageway. This causes traffic jams forcing the cars to slowdown, to accelerate and to stand idle. It results in increased fuel consumption, growing CO₂ emissions and losing time, hence, losing the drivers’ calm.
The Korean researchers proposed to refuse from these procedures: the inspection may be performed “on-the-go”. The vehicle is equipped with a line camera fixing the condition of the road paving at the speed lower than 100 km/h. The DETR algorithm based on neural network transformers analyzes the images and automatically identifies cracks and other defects.
To assess the efficiency of this technology, the team performed a computer simulation of road traffic at the section of Gyeongbu highway — on of the busiest roads in the country. They used SUMO and FASTSim simulators to review three scenarios: standard traffic, traditional inspection with blocking the lane and inspection without blocking. Almost 94 thousand cars including fueled with gasoline and diesel were used in the simulation.
The results showed that blocking the lanes reduced the average traffic speed approximately by one quarter and increased the time of travelling more than by 40%. In absolute values, it led to consumption of 5,044 extra liters of gasoline and 3,208 extra liters of diesel fuel spent in waste during 24 hours on just one section of the highway. Respectively, carbon dioxide emissions grew by 11.86 and 8.64 tons. The new inspection method practically eliminated these losses: the fuel consumption grew only by 0.1%, and the traffic remained stable.
Hence, the research demonstrated that the main cause of extra energy consumption during road inspections was associated not with the inspection itself, but with organization of traffic. Refusal from blocking the lanes allows for facilitating the diagnostics, reducing the emissions and decreasing the indirect environmental costs of the road system.
In future the researchers intend to test the system on the roads with different terrain and different traffic density, as well as to improve the artificial intelligence algorithms in order to increase the level of accuracy of identifying the defects in bad lights, in humid weather and in the night time. Simultaneously they plan to expand the assessment model including multilane and long sections. This new technology may become an important element of the energy-awareness concept road maintenance, when not only direct maintenance costs, but indirect losses due to traffic jams, excessive fuel consumption and redundant emissions are taken into account.
