Time‐aware gated recurrent unit networks for forecasting road surface friction using historical data with missing values

Abstract

An accurate road surface friction prediction algorithm can enable intelligent transportation systems to share timely road surface condition to the public for increasing the safety of the road users. Previously, scholars developed multiple prediction models for forecasting road surface conditions using historical data. However, road surface condition data cannot be perfectly collected at every time stamp, e.g. the data collected by on-vehicle sensors may be influenced when vehicles cannot travel due to economic cost issue or weather issues. Such resulted missing values in the collected data can damage the effectiveness and accuracy of the existing prediction methods since they are assumed to have the input data with a fixed temporal resolution. This study proposed a road surface friction prediction model employing a Gated Recurrent Unit networkbased decay mechanism (GRU-D) to handle the missing values. The evaluation results present that the proposed GRU-D networks outperform all baseline models. The impact of missing rate on predictive accuracy, learning efficiency and learned decay rate are analyzed as well. The findings can help improve the prediction accuracy and efficiency of forecasting road surface friction using historical data sets with missing values, therefore mitigating the impact of wet or icy road condition on traffic safety.

A hybrid neural network for driving behavior risk prediction based on distracted driving behavior data

“…Tang proposed a forecasting framework named the spatiotemporal gated graph attention network to predict the urban traffic flow based on license plate recognition data [ 19 ]. In addition, Pu uses historical data to predict road surface friction [ 20 , 21 ]. Tang used a geographically weighted Poisson quantile regression model to study the spatial heterogeneity and estimated the spatial impact on crash frequency [ 22 ].…”

Urban Safety: An Image-Processing and Deep-Learning-Based Intelligent Traffic Management and Control System

“…Bartlett et al [59] considered the computational cost and network structure optimization and proposed three recurrent neural network models, with the GRU model outperforming the others, achieving an RMSE of 9.26%. To further enhance the accuracy and robustness, Pu et al [60] integrated a decay mechanism as extra gates of the GRU model to handle the missing value problem. Model transferability and reproducibility can be improved by considering both temporal and local features in traffic flow.…”

RoadSurf‐Pedestrian: a sidewalk condition model to predict risk for wintertime slipping injuries

“…Another type is an optical sensor that estimates the surface friction based on measured water/snow/ice layer information using spectroscopic measuring principles (Bridge, 2008; Vaisala, 2017). In addition to measuring devices, there are numerical models that use meteorological information (Juga et al ., 2012) or neural networks and historical friction data (Pu et al ., 2019) to predict the road surface friction. The sensors and devices listed above have been developed to measure or estimate the friction between the road surface and vehicle tyre (Aschan et al ., 2004).…”