Probe Vehicles Research Articles

Improving Interstate Freeway Travel Time Reliability Analysis by Clustering Travel Time Distributions

Travel time reliability quantifies variability in travel times and has become a critical aspect for evaluating transportation network performance. The empirical travel time cumulative distribution function (CDF) has been used as a tool to preserve inherent information on the variability and distribution of travel times. With advances in data collection technology, probe vehicle data has been frequently used to measure highway system performance. One challenge with using CDFs when handling large amounts of probe vehicle data is deciding how many different CDFs are necessary to fully characterize experienced travel times. This paper explores statistical methods for clustering CDFs of travel times at segment level into an optimal number of homogeneous clusters that retain all relevant distributional information. Two clustering methods were tested, one based on classic hierarchical clustering and the other used model-based functional data clustering, to find out their performance on clustering distributions using travel time data from Interstate 64 in Virginia. Freeway segments and those within interchange areas were clustered separately. To find the proper data format as clustering input, both scaled and original travel times were considered. In addition, a non-data-driven method based on geometric features was included for comparison. The results showed that for freeway segments, clustering using travel times and the Anderson–Darling dissimilarity matrix and Ward’s linkage had the best performance. For interchange segments, model-based clustering provided the best clusters. By clustering segments into homogenous groups, the results of this study could improve the efficiency of further travel time reliability modeling.

Physics-Informed Deep Learning for Traffic State Estimation: A Hybrid Paradigm Informed By Second-Order Traffic Models

Traffic state estimation (TSE) reconstructs the traffic variables (e.g., density or average velocity) on road segments using partially observed data, which is important for traffic managements. Traditional TSE approaches mainly bifurcate into two categories: model-driven and data-driven, and each of them has shortcomings. To mitigate these limitations, hybrid TSE methods, which combine both model-driven and data-driven, are becoming a promising solution. This paper introduces a hybrid framework, physics-informed deep learning (PIDL), to combine second-order traffic flow models and neural networks to solve the TSE problem. PIDL can encode traffic flow models into deep neural networks to regularize the learning process to achieve improved data efficiency and estimation accuracy. We focus on highway TSE with observed data from loop detectors and probe vehicles, using both density and average velocity as the traffic variables. With numerical examples, we show the use of PIDL to solve a popular second-order traffic flow model, i.e., a Greenshields-based Aw-Rascle-Zhang (ARZ) model, and discover the model parameters. We then evaluate the PIDL-based TSE method using the Next Generation SIMulation (NGSIM) dataset. Experimental results demonstrate the proposed PIDL-based approach to outperform advanced baseline methods in terms of data efficiency and estimation accuracy.

A hidden Markov model for the estimation of correlated queues in probe vehicle environments

Queue length estimation is critical for traffic signal control and performance measures. With the development of connected vehicle technologies and the popularization of ride-hailing services, probe vehicle data are now being collected on a large scale. Some studies have shown that queue lengths can be estimated using only probe vehicle data. The relevant literature usually assumes the queue lengths in different traffic signal cycles are independent and identically distributed or treats the queues independently. However, in the real world, the queue lengths in different cycles might be correlated. For instance, when there exists an overflow queue, the queue length in the following cycle is correlated with the queue length in the previous cycle. In fact, the correlation of different cycles can provide additional information and thus improve the queue length estimation accuracy. In this paper, we model such queueing processes in probe vehicle environments using a hidden Markov model (HMM), where the queue length in each cycle is a hidden state, and the observed pattern of probe vehicles is an observation. Based on the HMM, we propose two novel cycle-by-cycle queue length estimation methods. In the case where the parameters of the HMM are unknown, we also provide an algorithm that can estimate the parameters from historical probe vehicle data. Validation results show that the proposed cycle-by-cycle queue length estimation methods outperform the existing methods, and the parameter learning algorithm can estimate the parameters adequately.

Dynamically Collected Local Density using Low-Cost Lidar and its Application to Traffic Models

This article demonstrates the use of traffic density observations collected dynamically in the vicinity of probe vehicles. Fixed position sensors cannot capture the longitudinal evolution of local traffic density in the corridor. In this research, dynamic traffic density observations were collected in a naturalistic driving setting that was free of any controlled experiment biases. Speed from global positioning system and space headway from a light detection and ranging module was collected on one arterial and one freeway segment, 2 and 4 mi long, respectively. The combined data frequency was approximately 3 Hz. Space headway was used to estimate the local density and consequently to identify the density of a specific location in a corridor. Besides, driver behavior was characterized using the relationship between instantaneous speed and local density under different regimes of the Wiedemann car-following model. Macroscopic traffic stream models were used to investigate the relationship between dynamically collected instantaneous speed and local density. Using the longitudinal evolution of density, precise local density across the corridor can be obtained along with the leader and follower trajectories. A method to identify driver behavior across density ranges was developed for different facility types using a microscopic relationship between instantaneous speed and local density. Overall driving behavior on the freeway segment can be represented by translating the instantaneous speed and local density relationship to macroscopic stream models.

Micro-analysis of a single vehicle driving volatility and impacts on emissions for intercity corridors

A deep understanding of driver behavior is an important step to improve road safety and environmental performance. Volatility can be defined as the extent of variations in driving, which can be characterized by accelerations/braking, lane change, and unusual high speed for roadways conditions. There is a lack of knowledge on what concerns the relationship between driver’s volatility and exhaust emissions and how driving volatility can be used as safety eco-indicator. This article explores a driving volatility concept for assessing tailpipe emissions and driving behavior classification. For that purpose, an empirical approach that combined vehicle activity and emission rates for light duty vehicles was used. Field measurements were collected from four probe vehicles in one partly urban/rural, and two highway routes using Portable Emission Measurement Systems, Global Positioning System receivers, and On-board Diagnostic scan tool, to measure real-world tailpipe emissions, position and engine activity data. Acceleration-based parameters, including relative positive acceleration and mean of positive acceleration, acceleration, vehicular jerk, and power demand thresholds were used to detect differences in emissions for different driving styles. Results indicated that vehicular jerk impacted carbon dioxide and nitrogen oxides per unit distance regardless of driving style and route type, especially from negative to null jerk values and during positive accelerations. There is potential to incorporate the analyzed thresholds into a driver decision support algorithm by considering safety and environmental aspects through warning messages.

Traffic state estimation by backward moving observers: An application and validation under an incident

This study analyzes measurements by backward moving observers that could be probe vehicles running backward on the opposite lane observing forward moving traffic to be investigated. These probe vehicles are called as backward probe vehicles (BP) and they are proven to measure the traffic flow and density. Using some advanced technology, a BP is assumed to estimate the flow of vehicles running forward from their passing time measurements along the BP trajectory. Then, as a useful application for the flow measurement by a BP, we propose a data assimilation method that estimates traffic states under an incident on an expressway section utilizing BP measurements in addition to conventional probe vehicles moving forward (forward probe vehicles) and detector data. Ample literature exists on traffic state estimation using several sensing data. However, they have difficulty in estimating traffic states during an incident, since the observations of the incident period and the declined flow rate due to the incident may not be sufficiently accurate. Therefore, this study proposes a state space model (SSM) that estimates traffic states under an incident on an expressway utilizing BP measurements. The model validation using a hypothetical network with an incident confirms the promising potential of the proposed model; that is, the reproducibility of traffic states using BP measurements is superior to one using forward probe measurements.

Maximum Likelihood Estimation of Probe Vehicle Penetration Rates and Queue Length Distributions From Probe Vehicle Data

Queue length estimation plays an important role in traffic signal control and performance measures of signalized intersections. Traditionally, queue lengths are estimated by applying the shockwave theory to loop detector data. In recent years, the tremendous amount of vehicle trajectory data collected from probe vehicles such as ride-hailing vehicles and connected vehicles provides an alternative approach to queue length estimation. To estimate queue lengths cycle by cycle, many existing methods require the knowledge of the probe vehicle penetration rate and queue length distribution. However, the estimation of the two parameters has not been well studied. This paper proposes a maximum likelihood estimation method that can estimate the parameters from historical probe vehicle data. The maximum likelihood estimation problem is solved by the expectation-maximization (EM) algorithm iteratively. Validation results show that the proposed method could estimate the parameters accurately and thus enable the existing methods to estimate queue lengths cycle by cycle.

Vehicular crowd-sensing: a parametric routing algorithm to increase spatio-temporal road network coverage

ABSTRACT Current vehicles are equipped with a number of environmental sensors to improve safety and quality of life for passengers. Many researchers have shown that these sensors can also be exploited for opportunistic crowd-sensing. Useful new services can be developed on top of these data, like urban surveillance of Smart Cities. The spatio-temporal sensing coverage achievable with Vehicular Crowd-Sensing (VCS), however, is an open issue, since vehicles are not uniformly distributed over the road network, undermining the quality of potential services based on VCS data. In this paper, we present an evolution of the standard A routing algorithm, meant to increase VCS coverage by selecting a route in a random way among all those satisfying a parametric constraint on the total cost of the path. The proposed solution is based on an edge-computing paradigm, not requiring a central coordination but rather leveraging the computational resources available on-board, significantly reducing the back-end infrastructure costs. The proposed solution has been empirically evaluated on two public datasets of 450,000 real taxi trajectories from two cities, San Francisco and Porto, characterized by a very different road network topology. Results show sensible improvements in terms of achievable spatio-temporal sensing coverage of probe vehicles.

Improved grey system models for predicting traffic parameters

In transportation applications such as real-time route guidance, ramp metering, congestion pricing and special events traffic management, accurate short-term traffic flow prediction is needed. For this purpose, this paper proposes several novel online Grey system models (GM): GM(1,1|cos(ωt)), GM(1,1|sin(ωt),cos(ωt)), and GM(1,1|e-at,sin(ωt),cos(ωt)). To evaluate the performance of the proposed models, they are compared against a set of benchmark models: GM(1,1) model, Grey Verhulst models with and without Fourier error corrections, linear time series model, and nonlinear time series model. The evaluation is performed using loop detector and probe vehicle data from California, Virginia, and Oregon. Among the benchmark models, the error corrected Grey Verhulst model with Fourier outperformed the GM(1,1) model, linear time series, and non-linear time series models. In turn, the three proposed models, GM(1,1|cos(ωt)), GM(1,1|sin(ωt),cos(ωt)), and GM(1,1|e-at,sin(ωt),cos(ωt)), outperformed the Grey Verhulst model in prediction by at least 65%, 16% and 11%, in terms of Root Mean Squared Error, and by 82%, 58% and 42%, in terms of Mean Absolute Percentage Error, respectively. It is observed that the proposed Grey system models are more adaptive to location (e.g., perform well for all roadway types) and traffic parameters (e.g., speed, travel time, occupancy, and volume), and they do not require as many data points for training (4 observations are found to be sufficient).

Online Estimation and Prediction of Large-Scale Network Traffic From Sparse Probe Vehicle Data

Network traffic prediction based on probe vehicle data is important for traffic management and route recommendation and has been intensively studied. Previous traffic prediction methods mainly focused on recurring traffic congestion. Predicting non-recurring traffic congestion, caused by events and accidents, is significantly more important; however, it has not been intensively studied. To predict non-recurring traffic congestion using probe data, we need to estimate the current traffic conditions based on <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sparse</i> observations for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">large</i> traffic networks to track traffic changes <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">online</i> . Conventional traffic forecasting methods have not been able to solve all of these problems. To address these problems, we propose a data assimilation method using a state space neural network (SSNN) with an incorporated topology of road networks. The SSNN model can easily model network traffic and can easily estimate its states and parameters by data assimilation using Bayesian filtering. In this study, we adopted a decoupled extended Kalman filter (DEKF) based data assimilation, which is scalable and applicable to large-scale network traffic, to estimate the states and parameters online. We evaluate the proposed method using an open dataset that includes a road network comprising over 30000 road segments. The results show that our method achieves higher prediction accuracy for predicting unknown traffic congestion and is more robust against data sparsity than conventional state estimation methods.

Travel Time Comparison Using Google Map and OSM Tracker

In India, people’s socio-economic characteristics in society have changed over the decades which have resulted in increase of vehicle ownership at a huge rate which poses a lot of problem in both traffic control and management system. In the area where continuation is impossible, congestion can be overcome by managing and controlling the traffic. There are major developments in ITS (Intelligent Transportation System) which needs travel time as a major input. Measure the quality of any transportation network travel time is a very popular parameter. The estimation of travel time in urban network became more difficult because of the rapid change in the system and traffic. Emerging technologies such as GPS which become widely available within the last decade, can offer promising improvements in traffic monitoring technologies. Probe vehicles equipped with relatively cheap devices provide researchers with vast amounts of disaggregate traffic data that can be easily used for determining link travel speeds. This study is done in order to estimate the impact of different travel modes on travel time with different day. Data has been collected in a study stretch of 8.1 km length in Khardaha to Dunlop more, KOLKATA using mobile GPS application (OSM tracker in android Operating System) . Different private modes of transportation such as motor bike, bus and car have been used as test vehicle for the collection of data in different traffic flow scenarios in different times and also days. Latitude, longitude, speed and the travelled path at one second interval has been extracted from the GPS application. After completing a data collection, data extracted from the Osm Tracker that is GPX file to convert the value into Excel format. The travel time and average speed of different modes has been compared and the effect of vehicle composition on travel time and average speed has been analyzed.

Trajectory data based freeway high-risk events prediction and its influencing factors analyses

The frequent crash occurrences have caused massive loss of lives and properties all over the world. In order to improve traffic safety, it is vital to understand the relationships between traffic operation conditions and crash risk, and further implement safety countermeasures. Emerging studies have conducted the crash risk analyses using discrete and aggregated traffic data (e.g., loop detector data, probe vehicle data), where crash events were selected as the prediction target. However, traditional traffic sensing data obtained at segment level cannot describe the detailed operation conditions for the vehicle platoons near crash locations. Thus, more microscopic and high-resolution traffic sensing data are needed. In addition, considering the random occurrence feature of crashes, high-risk events should be paid more attentions given their higher occurrence probability and consistent causations with crashes, which could proactively reduce crash likelihood. In this study, HighD Dataset from German highways was utilized for the empirical analyses. First, high-risk events were obtained using safety surrogate measures with Modified Time to Collision (MTTC) less than 2 s. Traffic operation characteristics within 5 s prior to event occurrence were extracted based on vehicle trajectory data. Then, a total of three different logistic regression models were established, which are standard logistic regression model, random-effects logistic regression (RELR) model, and random-parameter logistic regression (RPLR) model. Among which, the RPLR model was showed to have the best fitness and prediction accuracy. The results showed that the disturbed traffic flows in both longitudinal and lateral directions have positive impacts on high-risk events occurrence. Besides, too close following distance between vehicles would lead to high-risk events. Moreover, RPLR models could provide a high prediction accuracy of 97 % for 2 s ahead of the high-risk events. Finally, potential safety improvement countermeasures and future application scenarios were also discussed.

Approach to Quantify the Impact of Disruptions on Traffic Conditions using Dynamic Weighted Resilience Metrics of Transport Networks

Transport networks are essential for societies. Their proper operation has to be preserved to face any perturbation or disruption. It is therefore of paramount importance that the modeling and quantification of the resilience of such networks are addressed to ensure an acceptable level of service even in the presence of disruptions. The paper aims at characterizing network resilience through weighted degree centrality. To do so, a real dataset issued from probe vehicle data is used to weight the graph by the traffic load. In particular, a set of disrupted situations retrieved from the study dataset is analyzed to quantify the impact on network operations. Results demonstrate the ability of the proposed metrics to capture traffic dynamics as well as their utility for quantifying the resilience of the network. The proposed methodology combines different metrics from the complex networks theory (i.e., heterogeneity, density, and symmetry) computed on temporal and weighted graphs. Time-varying traffic conditions and disruptions are analyzed by providing relevant insights on the network states via three-dimensional maps.

Detecting the occurrence times and locations of multiple traffic crashes simultaneously with probe vehicle data

When we detect the occurrence times and locations of traffic crashes in real life, we will face the situation where multiple crashes occur along a long road during a specific time period. Although there has been a proliferation of approaches that attempt to detect the occurrence times and locations of traffic crashes, most, if not all, of them typically manipulate the crashes singly when there are multiple crashes. In this research, we propose a new approach that can detect the occurrence times and locations of multiple crashes simultaneously. We first construct the speed contour plot using probe vehicle data on candidate links. We then formulate the detection process as an integer programming model and develop a set of novel constraints to estimate the spatiotemporal impact regions associated with multiple crashes by leveraging the spatiotemporal propagation of congestion. We subsequently take the time and location when the travel speed begins to drop in each impact region as the occurrence time and location of each corresponding crash. Finally, we validate our model using real data in Beijing and find that our model can handle the situation where multiple crashes are incorporated in the speed contour plot. Besides, our model can reduce the average time and location bias by 79.52% and 81.29%, respectively.

Grid Mapping for Road Network Abstraction and Traffic Congestion Identification Based on Probe Vehicle Data

AbstractTraffic congestion monitoring is a long-term concern in urban areas. However, due to the complex structure of urban road networks and large amounts of traffic data, it is necessary to find ...

Variable speed limit strategy with anticipatory lane changing decisions

This paper develops a novel anticipatory Variable Speed Limit (VSL) control strategy that incorporates driver behavior based on trajectory data from probe vehicles. In particular, the research examines how the speed limit controls can be coordinated and optimized to reduce lane changing and overall braking, thus to achieve greater traffic throughput, safety, and sustainability. Driver behavior, such as acceleration/deceleration and lane changing, as reflected in individual trajectory data from GPS enabled probe vehicles are crucial for early detection of shockwave formation and proactive selection of speed limits; and consequently delay, or even eliminate breakdown formation. The core of this approach is the incorporation of a lane changing model which provides a more robust integrated speed limit selection and shockwave detection framework than the one we have developed in an earlier paper. The control principle is formulated in a generic fashion that finds the optimal speed limit control variables for either separate or simultaneous reduction of travel time, crash rates, and fuel consumption over a prediction horizon. The findings from the paper suggested that the VSL strategy was able to result in fewer lane changing rate (LCR) compared to the No-VSL case. Also, the developed algorithm was able to produce higher frequency of lower acceleration and deceleration rates than the No-VSL case, which indicated a smoother acceleration and deceleration pattern that corresponded to lower emission and fuel consumption. The performance of the algorithm was also examined under different probe penetration rates and congestion levels to identify the % of probe needed to achieve simultaneous mobility, safety, and environmental objectives.

Risk evaluation of traffic standstills on winter roads using a state space model

A method that evaluates the risk of traffic standstills on winter roads in real time using a state space model is proposed herein. In Japan, large-scale anomaly events such as traffic standstills that cause serious road disturbances occur frequently every year because of heavy snowfall. However, if the risk of anomaly events is known in advance, appropriate preparation and management can be undertaken to prevent such events and/or alleviate their impacts on road traffic. Therefore, this study attempts to evaluate the risk of standstills based on the degraded road performance estimated from probe vehicle speeds using sequential Bayesian filtering in a state space model (SSM). The SSM comprises a system model constructed by learning historical data and a measurement model using several exogenous variables such as snowfall amounts and temperature. The risk of anomaly events is then determined as the deviation of the filtered vehicle speed by the SSM from the statistically feasible speed distribution. The validation is performed by applying the proposed model to 58 traffic-standstill cases in northern Japan, and we confirm that the model successfully evaluates risks at a reasonable level that permits the practical use.

Influence of Weather Features in Determining Sudden Braking

Understanding conditions and situations causing abnormal driving behaviors like sudden braking or sudden acceleration is important for preventing traffic accidents. Previous studies have used probe vehicle data to detect risky situations where sudden braking frequently occurred. However, they have mainly focused on location and vehicle-related factors. In this paper, we build models which discriminate sudden braking using a machine learning method. The models use weather-related information as well as probe data. To investigate how weather-related factors help to determine sudden braking, we conducted extensive experiments using probe data obtained from dashboard cameras and two types of weather-related information obtained from meteorological observatories (MO) and AMeDAS. Experimental results illustrate that using weather-related information improves performance in determining sudden braking and that the temporally and spatially denser characteristics of weather-related factors from AMeDAS help to compensate for insufficiencies in the model with MO data.

Sensors on the Move: Onboard Camera-Based Real-Time Traffic Alerts Paving the Way for Cooperative Roads.

European road safety has improved greatly in recent decades. However, the current numbers are still far away to reach the European Commission’s road safety targets. In this context, Cooperative Intelligent Transport Systems (C-ITS) are expected to significantly improve road safety, traffic efficiency and comfort of driving, by helping the driver to make better decisions and adapt to the traffic situation. This paper puts forward two vision-based applications for traffic sign recognition (TSR) and real-time weather alerts, such as for fog-banks. These modules will support operators in road infrastructure maintenance tasks as well as drivers, giving them valuable information via C-ITS messages. Different state-of-the-art methods are analysed using both publicly available datasets (GTSB) as well as our own image databases (Ceit-TSR and Ceit-Foggy). The selected models for TSR implementation are based on Aggregated Chanel Features (ACF) and Convolutional Neural Networks (CNN) that reach more than 90% accuracy in real time. Regarding fog detection, an image feature extraction method on different colour spaces is proposed to differentiate sunny, cloudy and foggy scenes, as well as its visibility level. Both applications are already running in an onboard probe vehicle system.

Dynamic Path Flow Estimation Using Automatic Vehicle Identification and Probe Vehicle Trajectory Data: A 3D Convolutional Neural Network Model

Dynamic path flows, referring to the number of vehicles that choose each path in a network over time, are generally estimated with the partial observations as the input. The automatic vehicle identification (AVI) system and probe vehicle trajectories are now popular and can provide rich and complementary trip information, but the data fusion was rarely explored. Therefore, in this paper, the dynamic path flow estimation is based on these two data sources and transformed into a feature learning problem. To fuse the two data sources belonging to different detection ways at the data level, the virtual AVI points, analogous to the real AVI points (turning movements at nodes with AVI detectors), are defined and selected to statically observe the dynamic movement of the probe vehicles. The corresponding selection principles and a programming model considering the distribution of real AVI points are first established. The selected virtual AVI points are used to construct the input tensor, and the turning movement-based observations from both the data sources can be extracted and fused. Then, a three-dimensional (3D) convolutional neural network (CNN) model is designed to exploit the hidden patterns from the tensor and establish the high-dimensional correlations with path flows. As the path flow labels commonly with noises, the bootstrapping method is adopted for model training and the corresponding relabeling principle is defined to purify the noisy labels. The entire model is extensively tested based on a realistic road network, and the results show that the designed CNN model with the presented data fusion method can perform well in training time and estimation accuracy. The robustness of a model to noisy labels is also improved through the bootstrapping method. The dynamic path flows estimated by the trained model can be applied to travel information provision, proactive route guidance, and signal control with high real-time requirements.