Advanced traffic monitoring for sustainable traffic management: experiences and results of five years of collaborative research in The Netherlands

ABSTRACTSignificant efforts have been made in modeling a travel time distribution and establishing measures of travel time reliability (TTR). However, the literature on evaluating the factors affecting TTR is not well established. Accordingly, this paper presents an empirical analysis to determine potential factors that are associated with TTR. This study mainly applies the Bayesian Networks model to assess the probabilistic association between road geometry, traffic data, and TTR. The results from this model reveal that land use characteristics, intersection factors, and posted speed limits are directly associated with TTR. Evaluating the strength of the association between TTR and the directly related variables, the log odds ratio analysis indicates that the land use factor has the highest impact (0.83) followed by the intersection factor (0.57). The findings from this study can provide valuable resources to planners and traffic operators in their decision-making to improve TTR with quantitative evidence.

Travel time reliability has attracted increasing attention in recent years and is often listed as a major roadway performance and service quality measure for traffic engineers and travelers. Measuring travel time reliability is the first step toward improving it, ensuring on-time arrivals, and reducing travel costs. Most measures of travel time reliability derive from continuous probability distributions and apply to traffic data directly. However, little previous research shows a consensus for selection of a probability distribution family for travel time reliability. Different probability distribution families could yield different values for the same measure of travel time reliability (e.g., standard deviation). The authors believe that specific selection of probability distribution families has few effects on measuring travel time reliability. Therefore, they proposed two hypotheses for accurately measuring travel time reliability and designed an experiment to prove the two hypotheses. The first hypothesis was proved by ( a) conducting the Kolmogorov–Smirnov test and ( b) checking log likelihoods and the convergences of the corrected Akaike information criterion and of the Bayesian information criterion. The second hypothesis was proved by examining both moment- and percentile-based measures of travel time reliability. The results from testing the two hypotheses suggest that ( a) underfitting may cause disagreement in distribution selection, ( b) travel time can be precisely fitted by using mixture models with a higher value of K (regardless of distribution family), and ( c) measures of travel time reliability are insensitive to the selection of the distribution family. These findings allow researchers and practitioners to avoid testing of various distributions, and travel time reliability can be more accurately measured by using mixture models because of the higher values of log likelihoods.

Travel time is considered to be the key criterion when making travel related decisions. As the travel decisions are made in a dynamic environment, the travel time also changes according to the real-time operations of the transport system. More and more evidence proves that travellers are not only interested in the expected travel time but also in travel time reliability. Especially for trips that are made regularly, reliability is valued more than travel time itself. This dissertation focuses on travel time reliability measures and their eects on travel related decisions as well as network performance. Travel time is studied both at the aggregate level and the disaggregate level. At the aggregate level, personal preferences of travellers towards travel time are explored; while on the disaggregate level, the network performance is evaluated based on travel time reliability. Instead of defining a new travel time reliability index, the travel time distribution is used to address the variation of the travel time and its influences on both travellers at the micro level and the network at the macro level. Both revealed preference (RP) data and stated preference (SP) data are used for the analysis of the travel time reliability. The SP data provided two scenarios based on mode choice and route choice respectively and the data was collected in Switzerland. Proceeding on the SP data, parts of the RP data is also collected in Switzerland, which is later used to reconstruct the actual route choices of the respondents for a route choice model. Tomtom Stats data is obtained to assist the travel time reliability analysis during the procedure. Another part of the RP data, the floating car data (FCD), was collected from Wuhan, China and it is applied to employ network travel time reliability. Route choice models are built with travel time measures using both SP data and RP data. In the SP data based route choice model, as the travel time distribution was applied to generate the route alternatives, it allows us to explore the early and late indierence buers around the preferred arrival time. An exhaustive algorithm searches for the optimum early and late buer combinations and during the procedure, the changes of value of Abstract travel time savings, value of reliability early and late along with the model fit are closely observed. The RP data based route choice model tries to explain the respondents’ route choices reconstructed from the RP data with predicted travel time reliability measures. First regression models are employed to predict parameters of the travel time distribution based on Tomtom Stats data and then these models are applied to predict travel time reliabilities for each route alternative. A Path Size Logit model is used to account for similarities between route alternatives with percentage of early and late buers of travel time to address travel time reliability eects on route choices. FCD data is used to analyse travel time reliability on the network. Three levels of travel time reliability: link level, path level and network level are explored. On the link level, the travel time reliability is closely related to the speed changes along the road so link (un)reliability is defined as the integral of speed changes along the link. On the path level, Monte Carlo simulation is used to obtain the path travel time and the travel time distribution is extracted. The definition of a degradable network is given and a spatial auto regression model is built to calculate expected total network travel time and its variance. Then using method of moments estimation, the total network travel time distribution is reconstructed, from a degradable network as well as an undegradable network.

The impact of work zones on mean travel time and travel time reliability has gained attention as agencies focus on performance measurement. Travel time reliability in work zones often has been difficult for agencies to quantify because of the time and expense required to collect travel time data. Several private-sector companies have begun to sell travel time data; this availability has created an opportunity to examine cost-effectively work zone impacts on travel time reliability on a broader basis. The Virginia Department of Transportation recently acquired probe vehicle-based travel time data for 2010 from a private-sector data provider. These data were used to calculate measures of travel time reliability at 15 work zones and to examine factors that affected travel time reliability. Travel time reliability was quantified with 95th percentile travel time, a buffer index, and a planning time index. The work zones experienced a statistically significant degradation across all measures of travel time reliability as compared with baseline conditions. Work zone mean buffer index, planning time index, and 95th percentile travel time rates were higher by 48%, 18%, and 16%, respectively. Although lane closures occurred during off-peak periods, work zones that involved lane closures experienced increases in their mean buffer index, planning time index, and 95th percentile travel time with rates of 67%, 23%, and 22%, respectively. Annual average daily traffic per lane and the number of access points per mile were found to have the most obvious relationships with declines in reliability at work zones.

The concept of travel time reliability was developed to quantify the variability in travel times. As travel time reliability measures are increasingly used in system planning and performance measurement processes at many transportation agencies, predicting travel time reliability measures has become critical. However, it can be challenging because of the dynamic nature of traffic and the variety of factors contributing to unreliable travel times. This paper developed machine learning models to predict travel time reliability at a planning level. Two random forest algorithms, quantile random forests (QRF) and generalized random forests (GRF), were used to develop prediction models while taking account of a variety of variables from multiple data sources simultaneously. The reliability measures studied are the percentiles of travel times as they are a key component for many commonly used travel time reliability measures. Both QRF and GRF models produced accurate predictions; GRF performed better than QRF at predicting the 50th percentile travel time, and QRF achieved slightly better predictions for the 90th percentile. A case study demonstrated the use of the proposed models for estimating the impact on travel time reliability from an improvement project. The results found both models captured the trend in reliability change, and GRF was preferred over QRF for estimating the level of travel time reliability.

Accurate travel time prediction is very important for real-time traveller information systems. Many existing traveller information systems provide point estimates of forecast travel times. Often the forecast corridor travel time is estimated as a direct summation of the forecast link travel times on the route. This approach neglects the correlation between link travel times and may lead to inaccurate route travel time forecasts. This paper improves upon the simple addition method by accounting for the dependency of link travel times on the arrival time at that specific link which further correlates to its preceding links. In addition, this paper also explores the potential of using the nonlinear autoregressive with exogenous inputs (NARX) model and feedforward neural network model to forecast the corridor travel time mean and reliability metrics. To the authors knowledge this is the first time, short-term travel time reliability is measured by a reliability interval which is based on the forecasts of corridor travel time mean and standard deviation. The prediction methodologies developed in this paper are tested on an urban arterial that has been instrumented with Bluetooth readers so empirical travel times are available. It was found that the proposed NARX model outperforms the other models that were studied with respect to mean corridor travel time prediction. In terms of the reliability interval prediction, the performance of various models is presented as a Pareto Optimal Frontier trading off accuracy and usability. The proposed NARX model and three other tested models are all on the Pareto Optimal Frontier. Highlights Explores the nonlinear autoregressive with exogenous inputs (NARX) model for travel time predictions; Develops models to forecast urban arterial corridor travel time means and reliability metrics; Develops forecasting models with link-based or corridor-based travel time inputs, respectively; Evaluates the performance of various models using a Pareto Optimal Frontier; Compared travel time reliability metrics: reliability interval, 95th percentile, and buffer index.

Traffic data from private-sector sources is increasingly used to estimate the travel time reliability of major road infrastructure. However, there is as yet no study evaluating the difference in estimating travel time reliability between the private-sector data and automated vehicle identification (AVI) based on radio frequency identification. As ground truth data, the AVI data were collected from an AVI system using toll tags and aggregated into five-minute intervals. As one of the representative traffic information providers, data from HERE was obtained through the Regional Integrated Traffic Information System, calculated in five-minute intervals. For the comparison, four kinds of measures were selected and estimated on the basis of the day of the week, specific time periods, and time of day in five-minute, 15-minute, and one-hour intervals. The statistical difference in travel time reliability was assessed through paired t-tests. According to the results, AVI and HERE data are comparable based on day of the week, specific time periods, and time of day at one-hour intervals, whereas at five-minute and 15-minute intervals, HERE and AVI data are not generally comparable. Thus, when estimating travel time reliability in real time, travel time reliability derived from HERE data may be different from the true travel time reliability. Considering that private-sector traffic data are currently used to estimate travel time reliability measures, the measures should be harmonized on the basis of robust statistics to provide more consistent measures related to the true travel time reliability.

Modeling travel time reliability of freeways using risk assessment techniques

Travel time reliability (TTR) is an important performance indicator for transportation systems. TTR can be generally categorized as either segment based or origin–destination (O-D) based. A primary difference between the two TTR estimations is that route information is implied in segment-based TTR estimations. Segment-based TTR estimations have been widely studied in previous research; however, O-D–based TTR estimations are used infrequently. This paper provides detailed insight into O-D–based TTR estimations and raises three new issues: ( a) How many routes do travelers usually take and what are the TTR values associated with these routes? ( b) Do statistical differences exist between route-specific and non-route-specific (NRS) TTR values? ( c) How can O-D–based TTR information be delivered? Two processes were proposed to address the issues. Three TTR measures—standard deviation, coefficient of variation, and buffer index—were calculated. The bootstrapping technique was used to measure the accuracy of the TTR measures. Approximate confidence intervals were used to investigate statistically the differences between route-specific and NRS TTR measures. A large quantity of taxicab GPS-based data provided data support for estimating O-D–based TTR measures. The results of O-D–based TTR measures showed that no statistically significant differences existed between route-specific and NRS TTR measures for most of the time periods examined. Statistically significant differences could still be found in some time periods. Travelers may take advantage of these differences to choose a more reliable route. Access to both numeric TTR values and route preference, instead of just to TTR information on segments of interest, can be beneficial to travelers in planning an entire trip.

Travel time reliability is a key indicator of system performance and has become increasingly important in today's world as businesses as well as households require on-time transportation for their activities. In order to analyze travel time reliability, an enormous amount of traffic flow data is needed. Recently, the development of advanced traffic flow data collection systems has enabled us to handle that. This article aims to examine the fundamental characteristics of travel time reliability measures using the traffic flow data from the Hanshin Expressway network. Differences and similarities in characteristics (average travel time, 95th percentile travel time, standard deviation, coefficient of variation, buffer time, and buffer time index) are investigated on one radial route (Route 11: Ikeda Line). The result shows that buffer time and buffer time index profiles have a similar tendency as those of the standard deviation and coefficient of variation, respectively. Differences in characteristics among five radial routes in the network are also investigated in order to explore each route's characteristics of time-of-day variation of traffic flow. Additionally, the effect of traffic incidents on travel time reliability measures is analyzed on one radical route (Route 14: Matsubara Line). The results show that traffic incidents are the dominant factor for travel times in off-peak hours on the Hanshin Expressway network.

Travel time reliability (TTR) is an indicator of the level of service, especially the level of congestion experienced on it. Measures of TTR attempt to quantify the variability of the travel time of travelers’ experiences along the same route at different times. Traffic guidance is an alternative method to improve the TTR of road networks. However, the traditional traffic guidance based on variable message signs (VMS) encounters the delay effect of guidance information, which degrades the effectiveness of route guidance. The traffic guidance based on the connected vehicle (CV) is an effective method to enhance the TTR. Impact of real-time guidance information on TTR measures in a CV environment is analyzed. A traffic guidance algorithm based on CV technologies is proposed, and is validated by an integrated simulator for connected vehicle technologies. The simulation results show that traffic guidance has a significant influence on TTR in a CV environment. The increase of incident duration will produce the more serious traffic congestion, which results in the decline of TTR. A high penetration rate will improve the TTR of road networks. The proportion of non-CVs guided indirectly has also an influence on TTR, namely, a high following rate can help alleviate the impact of the incident on TTR.

The study reported here aimed to quantify the travel time reliability of urban midblock under heterogeneous traffic condition. Travel time index is taken into account for the measure of reliability for urban midblock. Geometric data, traffic volume count, and travel time data of seven urban midblock in Kerala, India forms the database for this study. Statistical evaluation of travel time index of each midblock is calculated for reliability estimation. A model was developed and validated for travel time index with traffic flow rate to predict the travel time reliability. The exponential model gave an accurate prediction on travel time reliability with traffic flow rate. This work also concentrates on proposing level of service criteria using travel time reliability measure. Hence, these new criteria can be used for travel time reliability prediction and also act as a basis to assess the level of service criteria for urban midblock under heterogeneous traffic condition.

Chapter 4 - Data-Driven Approaches for Estimating Travel Time Reliability

ABSTRACTAs with travel time collection, the accuracy of observed travel time and the optimal travel time data quantity should be determined before using travel time reliability (TTR) data. The statistical accuracy of TTR measures should be evaluated so that the statistical behavior and belief can be fully understood. More specifically, this issue can be formulated as a question: using a certain amount of travel time data, how accurate is the TTR for a specific freeway corridor, time of day, and day of week? A framework for answering this question has not been proposed previously. Our study proposes a framework based on bootstrapping to evaluate the accuracy of TTR measures and answer the question. Bootstrapping is a computer-based method for assigning measures of accuracy to multiple types of statistical estimators without requiring a specific probability distribution. Three scenarios representing three traffic flow conditions (free-flow, congestion, and transition) were used to evaluate the accuracy of TTR measures under different traffic conditions and quantities of data. The results of the accuracy measurements demonstrated that: (1) the proposed framework supports assessment of TTR accuracy and (2) stabilization of the TTR measures did not necessarily correspond to statistical accuracy. The findings in our study also suggested that moment-based TTR measures may not be statistically sufficient for measuring freeway TTR. Additionally, our study suggested that 4 or 5 weeks of travel time data is sufficient for measuring freeway TTR under free-flow conditions, 40 weeks for congested conditions, and 35 weeks for transition conditions.

Estimation of travel time is of increasing importance to travelers and transportation professionals alike as congestion worsens in major urban areas. In fact, the reliability of travel time estimates on a given corridor may be more important for travelers, shippers, and transport managers than the travel time itself. This paper examines the uses of measured travel time reliability indices for improving real-time transportation management and traveler information with the use of archived intelligent transportation system data. A literature review of travel time reliability and its value as a congestion measure is followed by a description of a content analysis of 20 regional transportation plans from across the nation. Results from the content analysis indicate that travel time reliability is not currently used as a congestion measure and that the most common measures of congestion are the volume-to-capacity ratio, vehicle hours of delay, and mean speed. As a case study using data from Portland, Oregon, several reliability measures are then tested including travel time, 95th percentile travel time, travel time index, buffer index, planning time index, and congestion frequency. The buffer index is used to prioritize freeway corridors according to travel time reliability. Metropolitan planning organizations should use travel time reliability in the following ways: (a) incorporate it as a systemwide goal, (b) evaluate roadway segments according to travel time reliability measures, and (c) prioritize roadway segments using those measures.