“This solution is intended to automate and optimize the vehicle movement tracking system,” said Mikhail Rodionov, Chief Operating Officer at Urbantech Group.
At present, this work cycle is performed manually which, according to the expert, is suboptimal as some of the vehicles may be parked and out of use. It is also a labor-intensive task: it takes up to three hours to check a vehicle.
“Automation and digitization elements allow to process big volumes of data in real time, optimize workflow and make informed decisions on the basis of such data. Taken together, all these steps help enhance control over special vehicle movements and solve traffic safety issues in a comprehensive manner,” noted Tikhov Firsov, Minister of Ecology and Nature Management for the Moscow Region.
This is how it happens. The system processes images from traffic enforcement and “Safe Region” cameras to identify special vehicles (tractors, bulldozers, etc.) in the overall traffic flow, registering their license plate numbers. Thereafter, AI tracks a vehicle along its route, checks whether technical inspection documents are available, and what their validity periods are. Once a potential violation is identified, the violator data file is sent over to the Technical Oversight Authority inspectors.
A logical question to ask is: could AI be wrong? As regards, for example, the safety belt, cameras often fail to recognize it against dark clothes.
“At the present technology advancement level, algorithms cannot assure one hundred percent error-free operation under any set of possible circumstances. Spurious operation is, therefore, evidently possible. However, if the developer continuously improves algorithms, expands its photo and video image base for neural network training purposes, the error rate may be brought down to a minimum. This is a quite feasible task,” Stanislav Shmelyov, Government Relations Director at XOR, an IT company, feels sure.