Article information
2023 , Volume 28, ¹ 2, p.19-26
Isaeva A.S., Denisenko M.A., Kovalev A.V.
Neural network application to road surface type identification
Road condition monitoring is an essential goal for transport infrastructure. It is important for the fast and safe evolution of autonomous vehicles, useful for advanced driver assistance systems and efficient road repair. In this paper we propose a solution to the problem of identifying the type of pavement using machine learning methods. Asphalt road, gravel road and cobbled road were the types of pavement quality, which were identified. The research community uses various types of sensors and data to solve this classification problem. This paper evaluates pavement type identification using data received from the inertial measurement unit installed in a vehicle and, in particular, data generated by the accelerometers. One car was used. The traffic route was chosen so that all three types of road surface were located on a small section of the road. The obtained data was used in training the long short-term memory recurrent neural network. The achieved accuracy of identification the type of road surface was 88.2 %.
Keywords: LSTM recurrent neural networks, inertial measurement unit, identification of the type of road surface
doi: 10.25743/ICT.2023.282.003
Author(s): Isaeva Alina Sergeevna PhD. Position: Senior Research Scientist Office: Southern Federal University Address: 344006, Russia, Rostov, 105/42 BolshayaSadovaya Str.,Rostov-on-Don
Denisenko Mark Anatolievich PhD. Position: Leading research officer Office: Southern Federal University Address: 344006, Russia, Rostov, 105/42 BolshayaSadovaya Str.,Rostov-on-Don
Kovalev Andrey Vladimirovich Dr. , Associate Professor Position: Head of department Office: Southern Federal University Address: 344006, Russia, Rostov, 105/42 BolshayaSadovaya Str.,Rostov-on-Don
Bibliography link: Isaeva A.S., Denisenko M.A., Kovalev A.V. Neural network application to road surface type identification // Computational technologies. 2023. V. 28. ¹ 2. P. 19-26
|