Classification of Road Surfaces Based on CNN Architecture and Tire Acoustical Signals

Abstract

This paper presents a novel work for classification of road surfaces using deep learning method-based convolutional neural network (CNN) architecture. With the development of advanced driver assistance system (ADAS) and autonomous driving technologies, the need for research on vehicle state recognition has increased. However, research on road surface classification has not yet been conducted. If road surface classification and recognition are possible, the control system can make a more robust decision by validating the information from other sensors. Therefore, road surface classification is essential. To achieve this, tire-pavement interaction noise (TPIN) is adopted as a data source for road surface classification. Accelerometers and vision sensors have been used in conventional approaches. The disadvantage of acceleration signals is that they can only represent the surface profile properties and are masked by the resonance characteristics of the car structure. An image signal can be easily contaminated by factors such as illumination, obstacles, and blurring while driving. However, the TPIN signal reflects the surface profile properties of the road and its texture properties. The TPIN signal is also robust compared to those in which the image signal is affected. The measured TPIN signal is converted into a 2-dimensional image through time–frequency analysis. Converted images were used together with a CNN architecture to examine the feasibility of the road surface classification system.

Deep learning-based wind noise prediction study for automotive clay model

“…However, machine learning methods have gained extensive application in automotive noise prediction. Commonly used machine learning methods include backpropagation neural networks (BPNNs) [39], radial basis neural networks [40], Elman neural networks [41], support vector machines [42,43], convolutional neural networks (CNNs) [44][45][46][47], and others. In recent years, machine learning methods have become a developmental trend in fluid dynamics prediction.…”

Automatic pavement texture recognition using lightweight few-shot learning

“…Dong et al [13] proposed an effective encoder-decoder CNN architecture based on the residual network for the reconstruction of pavement texture, which proved to be effective for three-dimensional feature analysis. Yoo et al [14] introduced a novel method of pavement classification based on deep learning by using the tire-pavement interaction noise, which reflects the surface contour and texture properties of the road, as the data source. Chhay et al [15] proposed a faster region-based CNN based on deep learning to count the exposed aggregate number representing texture wavelengths on digital images of exposed aggregate concrete pavement.…”

“…Yoo et al . [ 14 ] introduced a novel method of pavement classification based on deep learning by using the tire-pavement interaction noise, which reflects the surface contour and texture properties of the road, as the data source. Chhay et al .…”

Application of Convolutional Neural Networks for Road Type Classification

“…On the other hand, the paper [7] focuses on discovering the road friction coefficient. More precisely, the authors use the information acquired from tires as they are in direct contact with the driving surface and they can give an understanding of the road friction which depends on both the road type and the current weather conditions.…”