The classification of pavement crack heavily relies on the engineers’ experience or the hand-crafted algorithms. Convolutional Neural Network (CNN) has demonstrated to be useful for image classification, which provides an alternative to traditional imaging classification algorithms. This paper proposes a novel method using deep CNN to automatically classify image patches cropped from 3D pavement images. In all, four supervised CNNs with different sizes of receptive field are successfully trained. The experimental results demonstrate that all the proposed CNNs can perform the classification with a high accuracy. Overall classification accuracy of each proposed CNN is above 94%. Upon the evaluation of these neural networks with respect to accuracy and training time, we find that the size of receptive field has a slight effect on the classification accuracy. However, the CNNs with smaller size of receptive field require more training times than others.

Maintaining desired pavement friction is an effective way to reduce the crash rate. Pavement texture plays an important role in tire-pavement friction performance. However, the contribution of texture at different depths or wavelengths to friction measurement has not been fully understood. This study conducts a multiresolution analysis using three-dimensional (3D) asphalt pavement texture data to correlate pavement texture to friction measurements. Between 2014 and 2018, multiple rounds of pavement texture and friction data were acquired on two field sites. The top-surface topography of 3D texture data is sliced with eight depths to explore the critical contact depth between pavement texture and tire rubber for friction testing. At each depth, 3D texture data is decomposed into sub-images via discrete wavelet transform to extract texture features at 11 wavelengths ranging from 0.1 mm to 102.4 mm. Subsequently, the wavelet energy of each wavelength is calculated as the texture indicator. Finally, the texture energy matrix at different wavelengths and depths along with the ambient temperature during friction testing are correlated with the measured friction data using cross-validation and step-wise multivariate linear regression. The pavement texture of wavelengths up to 3.2 mm at the top 2.5 mm of top-surface topography is identified as the critical contact zone for pavement-tire interaction.

Traditional contact and water-based testing devices exhibit some limitations in evaluating pavement skid resistance. This study demonstrates the viability of deep learning (DL) based non-contact pavement skid resistance evaluation using 3D laser imaging technology. Pavement 3D images and friction numbers were collected from 28 field sites in 10 states in the U.S. to cover pavement sections with various texture and friction characteristics. A total number of 28,000 pairs of 3D images and friction numbers were prepared as the database to train and validate the friction prediction model, named as FrictionNet-V, using the convolutional neural network. The input of FrictionNet-V is the 3D pavement image from the left wheel path, and the output is the corresponding friction number ranging from 0.2 to 0.9. Based on the experimental results on 5,600 testing images, FrictionNet-V achieved better performance than other DL models in terms of speed and accuracy. The result demonstrates the potential of highway speed non-contact pavement skid resistance evaluation using high resolution 3D texture images and DL methods at the network level in the future.

This study presents an automatic method for horizontal curve identification and measurement using a mobile mapping system (MMS) with consideration of the impacts of vehicle wandering on the measurement. The hierarchical cluster model combined with the linear fit analysis achieved the automatic point of curvature (PC) and point of tangent (PT) station detection. The arc length method and chord offset method combined with the vehicle trajectory data were utilized for curve radius calculation. Finally, according to the adjacency patterns of the single curves, the types of curve were detected. A case study on a 9.756-km road section is provided, and on the test section, 21 single curves were detected and measured. Among them, 12 single curved segments were the simple curves, and nine single curved segments formed four reverse curves. A validation test based on the field measurement was conducted. The test results showed that the average errors for automatic PC/PT station detection, curve length measurement, and curve radius measurement were 2.15, 3.07, and 4.93%, respectively, which demonstrated that the presented method had a high accuracy for horizontal curve measurement. The significance of the proposed method was threefold. First, it solved the common problems of vehicle wandering on mobile survey. Second, it detected the curve types and the PC station of two adjacent single curves. Third, it had high accuracy, effectiveness, and automaticity, and can be used for roadway survey at network level.

Grooves are frequently used on airport runways to address safety concerns by improving frictional properties and drainage capacity during wet weather. Groove performance deteriorates over time mostly due to traffic loading, accordingly periodic measurement and evaluation of groove dimensions are important for runway to maintain adequate frictional properties and avoid hydroplaning risks. However, such evaluation is difficult to perform due to lack of sufficient technologies to identify shallow and abrasive or worn grooves. This paper proposes new methodologies for automatic airport groove identification, measurement, and evaluation using high resolution texture data collected by point laser based profiling equipment developed by Federal Aviation Administration (FAA). Euler-Bernoulli beam filter and geometry contour based algorithms are developed to locate dips within a profile for potential groove identification. Subsequently improved moving average filter, forward and backward traversal methods are implemented to determine the starting and ending positions of each dip. Finally groove dimensions are computed per the FAA AC No. 150/5320-12C. The algorithm performance of the proposed method is evaluated based on Precision-Recall (PR) graphs with that of the FAA ProGroove software. Results indicate that the proposed methodology is more robust and accurate in runway groove identification. With the proposed method, operators of airfield runway pavements have a robust tool set to access groove performance and provide corrective maintenance actions.

The warm mix asphalt (WMA) technology is commonly applied for highway pavements in high-altitude plateau regions in China, where extreme weather conditions are common. The study is to evaluate the performance of WMA with different additives in the areas. The tests show that all the additives could reduce the viscosity of asphalt binder to reduce the mixing and compacting temperature of asphalt mixture. The solid additives could also decrease the penetration and ductility value, and increase the softening point and the PG high temperature of asphalt binder; therefore, are suitable for improving the high-temperature performance of WMA in the high-temperature regions. However, the surface-active liquid additives are suitable in the plateau areas in low-temperature conditions. The lab data for the Ya-Kang highway pavement to Tibet Plateau show it is suitable to control the mixing temperature at about 20°C lower than the normal hot mix asphalt during the construction period.