Laurence R. Rilett

Expected shortest paths in dynamic and stochastic traffic networks

The dynamic and stochastic shortest path problem (DSSPP) is defined as finding the expected shortest path in a traffic network where the link travel times are modeled as a continuous-time stochastic process. The objective of this paper is to examine the properties of the problem and to identify a technique that can be used to solve the DSSPP given information that will be available in networks with Intelligent Transportation System (ITS) capabilities. The paper first identifies a set of relationships between the mean and variance of the travel time of a given path and the mean and variance of the dynamic and stochastic link travel times on these networks. Based on these relationships it is shown that the DSSPP is computationally intractable and traditional shortest path algorithms cannot guarantee an optimal solution. A heuristic algorithm based on the k-shortest path algorithm is subsequently proposed to solve the problem. Lastly, the trade-off between solution quality and computational efficiency of the proposed algorithm is demonstrated on a realistic network from Edmonton, Alberta.

FORECASTING FREEWAY LINK TRAVEL TIMES WITH A MULTILAYER FEEDFORWARD NEURAL NETWORK

One of the major requirements of advanced traveler information systems (ATISs) is a mechanism to estimate link travel times. This article examines the use of an artificial neural network (ANN) for predicting freeway link travel times for one through five time periods into the future. Actual freeway link travel times from Houston, Texas, that were collected as part of the automatic vehicle identification (AVI) system were used as a test bed. It was found that when predicting one or two time periods into the future, the ANN model that only considered previous travel times from the target link gave the best results. However, when predicting three to five time periods into the future, the ANN model that employed travel times from upstream and downstream links in addition to the target link gave superior results. The ANN model also gave the best overall results compared with existing link travel time forecasting techniques.

Heuristic shortest path algorithms for transportation applications: state of the art

There are a number of transportation applications that require the use of a heuristic shortest path algorithm rather than one of the standard, optimal algorithms. This is primarily due to the requirements of some transportation applications where shortest paths need to be quickly identified either because an immediate response is required (e.g., in-vehicle route guidance systems) or because the shortest paths need to be recalculated repeatedly (e.g., vehicle routing and scheduling). For this reason a number of heuristic approaches have been advocated for decreasing the computation time of the shortest path algorithm. This paper presents a survey review of various heuristic shortest path algorithms that have been developed in the past. The goal is to identify the main features of different heuristic strategies, develop a unifying classification framework, and summarize relevant computational experience.

Forecasting multiple-period freeway link travel times using modular neural networks

With the advent of route guidance systems (RGS), the prediction of short-term link travel times has become increasingly important. For RGS to be successful, the calculated routes should be based on not only historical and real-time link travel time information but also anticipatory link travel time information. An examination is conducted on how realtime information gathered as part of intelligent transportation systems can be used to predict link travel times for one through five time periods (of 5 minutes’ duration). The methodology developed consists of two steps. First, the historical link travel times are classified based on an unsupervised clustering technique. Second, an individual or modular artificial neural network (ANN) is calibrated for each class, and each modular ANN is then used to predict link travel times. Actual link travel times from Houston, Texas, collected as part of the automatic vehicle identification system of the Houston Transtar system were used as a test bed. It was found that the modular ANN outperformed a conventional singular ANN. The results of the best modular ANN were compared with existing link travel time techniques, including a Kalman filtering model, an exponential smoothing model, a historical profile, and a real-time profile, and it was found that the modular ANN gave the best overall results.

Direct Forecasting of Freeway Corridor Travel Times Using Spectral Basis Neural Networks

With the advent of advanced traveler information systems, the prediction of short-term link and corridor travel times has become increasingly important. The standard method for forecasting corridor travel times is a two-step process in which the link travel times are first forecast and then combined into a corridor travel time. If link travel times are not independent, however, there is the potential for erroneous corridor or route travel time estimates. As an alternative to the two-step approach, a direct or one-step approach for freeway corridor travel time forecasting is proposed that automatically takes into account interrelationships between link travel times. The use of spectral basis neural networks to directly forecast multiple-period freeway corridor travel times is examined first. The model is tested on observed travel times collected as part of the automatic vehicle identification component of the Houston Transtar system. The direct forecasting model is also compared with the two-step model, which uses forecast link travel times as input. It was found that the direct forecasting approach gave better results than any of the other models examined and that link travel time forecasting errors are not additive.

A comparison of the performance of artificial neural networks and support vector machines for the prediction of traffic speed

The ability to predict traffic variables such as speed, travel time or flow, based on real time data and historic data, collected by various systems in transportation networks, is vital to the intelligent transportation systems (ITS) components such as in-vehicle route guidance systems (RGS), advanced traveler information systems (ATIS), and advanced traffic management systems (ATMS). In the contest of prediction methodologies, different time series, and artificial neural networks (ANN) models have been developed in addition to the historic and real time approach. The present paper proposes the application of a recently developed pattern classification and regression technique called support vector machines (SVM) for the short-term prediction of traffic speed. An ANN model is also developed and a comparison of the performance of both these techniques is carried out, along with real time and historic approach results. Data from the freeways of San Antonio, Texas were used for the analysis.

DYNAMIC AND STOCHASTIC SHORTEST PATH IN TRANSPORTATION NETWORKS WITH TWO COMPONENTS OF TRAVEL TIME UNCERTAINTY

Abstract The existing dynamic and stochastic shortest path problem (DSSPP) algorithms assume that the mean and variance of link travel time (or other specific random variable such as cost) are available. When they are used with observed data from previous time periods, this assumption is reasonable. However, when they are applied using forecast data for future time periods, which happens in the context of ATIS, the travel time uncertainty needs to be taken into account. There are two components of travel time uncertainty and these are the individual travel time variance and the mean travel time forecasting error. The objectives of this study are to examine the characteristics of two components of travel time uncertainty, to develop mathematical models for determining the mean and variance of the forecast individual travel time in future time periods in the context of ATIS, and to validate the proposed models. First, this study examines the characteristics of the two components of uncertainty of the individual travel time forecasts for future time periods and then develops mathematical models for estimating the mean and variance of individual route travel time forecasts for future time periods. The proposed models are then implemented and the results are evaluated using the travel time data from a test bed located in Houston, Texas. The results show that the proposed DSSPP algorithms can be applied for both travel time estimation and travel time forecasting.

REAL-TIME OD ESTIMATION USING AUTOMATIC VEHICLE IDENTIFICATION AND TRAFFIC COUNT DATA

Origin-destination (OD) matrices are a key input to many advanced traffic management operations. In this study, two constrained OD estimators, based on generalized least squares and Kalman filtering, were developed and tested in order to examine the possibility of estimating OD matrices in real-time. A one-at-a-time processing method was introduced to provide an efficient organized framework for incorporating observations from multiple data sources in real-time. The estimators were tested under different conditions based on the type of prior OD information available, the type of assignment available, and the type of link volume model used. The performance of the Kalman filter estimator was also compared to that of the generalized least squares estimator to provide insight regarding their performance characteristics relative to one another for given scenarios. Automatic vehicle identification (AVI) tag counts were used so that observed and estimated OD parameters could be compared. AVI data was incorporated primarily in three ways: as prior observed OD information; the inclusion of a deterministic link volume component that makes use of OD data extracted from the latest time interval from which all trips have been completed; and through the use of link choice proportions estimated based on link travel time data. Results show that using prior observed OD data along with link counts improves estimator accuracy relative to OD estimation based exclusively on link counts. The findings also show that the incorporation of constraints creates estimators that are less sensitive to limitations such as deterministic modeling errors, unreliable OD data, and assignment error. Under these limitations, the constrained Kalman filter is more robust than constrained generalized least squares when incorporating prior OD information.

Support Vector Machine Technique for the Short Term Prediction of Travel Time

A vast majority of urban transportation systems in North America are equipped with traffic surveillance systems that provide real time traffic information to traffic management centers. The information from these are processed and provided back to the travelers in real time. However, the travelers are interested to know not only the current traffic information, but also the future traffic conditions predicted based on the real time data. These predicted values inform the drivers on what they can expect when they make the trip. Travel time is one of the most popular variables which the users are interested to know. Travelers make decisions to bypass congested segments of the network, to change departure time or destination etc., based on this information. Hence it is important that the predicted values be as accurate as possible. A number of different forecasting methods have been proposed for travel time forecasting including historic method, real-time method, time series analysis, and artificial neural networks (ANN). This paper examines the use of a machine learning technique, namely support vector machines (SVM), for the short-term prediction of travel time. While other machine learning techniques, such as ANN, have been extensively studied, the reported applications of SVM in the field of transportation engineering are very few. A comparison of the performance of SVM with ANN, real time, and historic approach is carried out. Data from the TransGuide Traffic Management center in San Antonio, Texas, USA is used for the analysis. From the results it was found that SVM is a viable alternative for short-term prediction problems when the amount of data is less or noisy in nature.

Improved Transit Signal Priority System for Networks with Nearside Bus Stops

Transit signal priority (TSP), which has been deployed in many cities in North America and Europe, is a traffic signal enhancement strategy that facilitates efficient movement of transit vehicles through signalized intersections. Most TSP systems, however, do not work well in transit networks with nearside bus stops because of the uncertainty in bus dwell time. Unfortunately, most bus stops on U.S. arterial roadways are nearside ones. In this research, weighted-least-squares regression modeling was used to estimate bus stop dwell time and, more important, the associated prediction interval. An improved TSP algorithm that explicitly considers the prediction interval was developed to reduce the negative impacts of nearside bus stops. The proposed TSP algorithm was tested on a VISSIM model of an urban arterial section of Bellaire Boulevard in Houston, Texas. In general, it was found that the proposed TSP algorithm was more effective than other algorithms because it improved bus operations without statistically significant impacts on signal operations.