Evidential Data Fusion for Characterization of Pavement Surface Conditions during Winter Using a Multi-Sensor Approach

Abstract

The role of a service that is dedicated to road weather analysis is to issue forecasts and warnings to users regarding roadway conditions, thereby making it possible to anticipate dangerous traffic conditions, especially during the winter period. It is important to define pavement conditions at all times. In this paper, a new data acquisition approach is proposed that is based upon the analysis and combination of two sensors in real time by nanocomputer. The first sensor is a camera that records images and videos of the road network. The second sensor is a microphone that records the tire–pavement interaction, to characterize each surface’s condition. The two low-cost sensors were fed to different deep learning architectures that are specialized in surface state analysis; the results were combined using an evidential theory-based data fusion approach. This study is a proof of concept, to test an evidential approach for improving classification with deep learning, applied to only two sensors; however, one could very well add more sensors and make the nanocomputers communicate together, to analyze a larger urban environment.

The Application of Multisensor Information Fusion Technology in Environmental Restoration

“…Due to the complexity and diversity of its application, the research content of information fusion is extremely rich, involving many basic theories, and the commonly used algorithm is Kalman filter, parameter template method, Bayesian inference, adaptive neural network, etc. The development of sensor technology has been relatively mature, and there are suitable sensor types for use in different environments, different types of data, and applications with different functions [20]. Most of the data acquired by the sensor is used to obtain the state of the system or to perceive the surrounding environment [21].…”

Evidential transformer for pavement distress segmentation

“… Probability‐to‐mass fusion (PMF): Diaby et al. (2021) use the feature vector of a pixel location and a softmax layer to compute the Bayesian probabilities reasoning on $$\Omega ^v,\ v=1,\dots ,V$$. The Bayesian probabilities are extended into the Bayesian mass functions on the common frame Ω 0 and aggregated by Dempster's rule. Bayesian‐fusion (BF): Xu et al.…”

“…Another problem is the multimodel fusion (Chen & Jahanshahi, 2017), in which the outputs from different deep networks based on the data from different sources are combined for decision making (Diaby et al., 2021). One challenge in pavement distress segmentation is to utilize the existing networks trained from heterogeneous data sets for obtaining a new one, which can improve the generality and accuracy of distress segmentation.…”

Urban Road Surface Discrimination by Tire-Road Noise Analysis and Data Clustering

“…Conversely, new remote sensing methodologies can address some of these shortcomings; for example, analysing large road sections is possible while handling large amounts of data. These methods include tools such as ground penetration radar, infrared thermography [12], laser scanning [13], image-based [14,15], vibration-based [16][17][18], and acoustic-based [19] methods. As these techniques are not mutually exclusive, more than one technique can be used simultaneously [10,20].…”