No abstract

Road roughness is an important factor in road network maintenance and ride quality. This paper proposes a road-roughness estimation method using the frequency response function (FRF) of a vehicle. First, based on the motion equation of the vehicle and the time shift property of the Fourier transform, the vehicle FRF with respect to the displacements of vehicle–road contact points, which describes the relationship between the measured response and road roughness, is deduced and simplified. The key to road roughness estimation is the vehicle FRF, which can be estimated directly using the measured response and the designed shape of the road based on the least-squares method. To eliminate the singular data in the estimated FRF, the shape function method was employed to improve the local curve of the FRF. Moreover, the road roughness can be estimated online by combining the estimated roughness in the overlapping time periods. Finally, a half-car model was used to numerically validate the proposed methods of road roughness estimation. Driving tests of a vehicle passing over a known-sized hump were designed to estimate the vehicle FRF, and the simulated vehicle accelerations were taken as the measured responses considering a 5% Gaussian white noise. Based on the directly estimated vehicle FRF and updated FRF, the road roughness estimation, which considers the influence of the sensors and quantity of measured data at different vehicle speeds, is discussed and compared. The results show that road roughness can be estimated using the proposed method with acceptable accuracy and robustness.

The evaluation of road roughness plays a critical role in the life-long maintenance of the highway system. This study proposes a Kalman Filter-based scheme to evaluate the road roughness indirectly from the response of a moving adapted monitoring vehicle. Key feature of the scheme is the use of measurements from dynamic tire pressure of unsprung mass components that directly interact with roads. Combination of ideal gas law and elastic contact model results in a nonlinear relationship between the tire pressure and the contact force, in which the parameters are calibrated by the Extended Kalman Filter. Identification of vehicle’s physical parameters adopts the power spectrum method with a known-size bump test. Subsequently, the road roughness is treated as unknowns in the vehicle’s state-space equation and solved by the Discrete Kalman Filter with unknown inputs. The estimated road roughness profiles are then used to calculate the International Roughness Index and compared with that provided by the standardized laser profilometer, an outer-systematic comparison. On the other hand, available measurements are split into groups that measurements of tire pressure are used to predict the accelerations of the car body and wheels and compared with these accelerations directly measured from accelerometers, an inner-systemic comparison. Field tests are carried out on a 900[Formula: see text]m long standardized road under two scenarios of with and without the bump and four different vehicle running speeds from 20 to 50[Formula: see text]km/h. Consistence of comparison from different perspectives proves the reliability of the proposed scheme. In addition, the results unveil that the scenario with a lower running speed can offer a better estimation of road roughness.

No abstract

This study develops and tests an automatic pixel-level image recognition model to reduce the amount of manual labor required to collect data for road maintenance. Firstly, images of six kinds of pavement distresses, namely, transverse cracks, longitudinal cracks, alligator cracks, block cracks, potholes, and patches, are collected from four asphalt highways in three provinces in China to build a labeled pixel-level dataset containing 10,097 images. Secondly, the U-net model, one of the most advanced deep neural networks for image segmentation, is combined with the ResNet neural network as the basic classification network to recognize distressed areas in the images. Data augmentation, batch normalization, momentum, transfer learning, and discriminative learning rates are used to train the model. Thirdly, the trained models are validated on the test dataset, and the results of experiments show the following: if the types of pavement distresses are not distinguished, the pixel accuracy (PA) values of the recognition models using ResNet-34 and ResNet-50 as basic classification networks are 97.336% and 95.772%, respectively, on the validation set. When the types of distresses are distinguished, the PA values of models using the two classification networks are 66.103% and 44.953%, respectively. For the model using ResNet-34, the category pixel accuracy (CPA) and intersection over union (IoU) of the identification of areas with no distress are 99.276% and 99.059%, respectively. For areas featuring distresses in the images, the CPA and IoU of the model are the highest for the identification of patches, at 82.774% and 73.778%, and are the lowest for alligator cracks, at 14.077% and 12.581%, respectively.