Loop Detector Data Research Articles

A theory-informed multivariate causal framework for trustworthy short-term urban traffic forecasting

Traffic forecasting using Deep Learning has been a remarkably active and innovative research field during the last decades. However, there are still several barriers to real-world, large-scale implementation of Deep Learning forecasting models, including their data requirements, limited explainability and low efficiency. In this paper, we propose a novel theory-driven framework that is based on a Granger causality-inspired feature selection method and a multitask LSTM to jointly predict two traffic variables. Traffic flow theory intuition is induced in the training process by an enhanced Traffic Flow Theory-Informed loss function (TFTI loss), which includes the divergence of the joint prediction of two traffic variables from the actual fundamental diagram of the corresponding location. The theory-informed, Granger causal, multitask LSTM is trained for one step ahead volume and speed forecasting using loop detector data coming from the extended Athens road network (Greece). Findings indicate that the models trained using the TFTI loss and a reduced input space, which includes only causal information, achieve a significantly improved performance, compared to the models using the classic Mean Squared Error loss function. Moreover, we introduce a dedicated trustworthiness evaluation framework that indicates that the proposed approach enhances the trustworthiness of the predictions, as well as the models’ transparency and resilience to noisy data.

A traffic incident detection method of highway based on spatial-temporal characteristics

In this study, we propose a method for detecting traffic incidents by leveraging tensor decomposition in conjunction with spatial-temporal constraints. The method comprises three main stages. Firstly, we develop an initialization and noise reduction procedure for the traffic data pre-processing model, which includes Tucker decomposition and truncating higher-order Singular Value Decomposition (SVD). Subsequently, we construct spatial and temporal coefficient matrices based on the Pearson test and introduce a Laplace penalty term to quantify the disparity between predicted and actual data. Next, we validate the method using traffic loop detector data and incident records from I90 in Seattle, USA, in 2015, comparing its performance with seven other traffic prediction methods. The results demonstrate the superior prediction accuracy, correct detection rate, and low false alarm rate of our traffic incident detection method. Specifically, the mean square error of speed prediction is 5.22 km / h , the average relative error is 6.88 % , and the detection rate reaches 98.92 % . Our method effectively evaluates the spatial-temporal characteristics of traffic data, enabling accurate prediction and detection of traffic incidents. Its technical applicability holds promise for enhancing the capacity and efficiency of future traffic control systems.

Choice-based macroscopic lane-change prediction model for weaving areas

This paper introduces two macroscopic lane change models, separately for small vehicles (passenger cars and vans) as well as heavy vehicles. The model is structured using a Nested Logit discrete choice framework. The model was estimated and validated using unique trajectory data collected from a busy weaving section in Antwerp, Belgium. The models exhibit an average and maximum rate of inaccurate prediction results, amounting to 11% and 16%, respectively. Additionally, the model underestimates the total number of lane change maneuvers, with a margin of no more than 4.1–7.7%. This study aims to fill the gap in macroscopic lane change models, which mainly depend on theoretical underpinnings, by including empirical data from the analysis of more than 31,000 observed maneuvers. Furthermore, it differentiates among three maneuver intents, a unique alternative-specific discrete choice framework. This differentiation allows for the consideration of drivers’ systematic taste variations. In contrast to other models that utilize empirical datasets and are heavily dependent on expensive video-based data, including the entire traffic flow, this study introduces a novel approach for estimating the position and frequency of maneuvers by using just a small fraction (1–2%) of vehicle trajectories, supplemented by loop detector data, therefore eliminating the need for information from surrounding vehicles. Additionally, the work’s uniqueness is its validation methods extending beyond just numerical evaluations. The practical usability and usefulness of the model are shown in real-world traffic circumstances. Finally, the model is integrated in the context of a macroscopic simulation that represents a highly consistent lane balance based on the Scalable Quality Value statistics.

Travel Time Estimation for Urban Arterials Based on the Multi-Source Data

Accurate traffic information, such as travel time, becomes more important since it could help provide more efficient traffic management strategies. This paper presents a method for estimating the travel time of segments on urban arterials by leveraging multi-source data from loop detectors and probe vehicles. Travel time is defined into three distinct sections based on floating car trajectories, i.e., accelerating, constant speed, and decelerating. Considering the traffic flow characteristics, different methods are developed using various data for each section. The proposed methodology is validated using field data collected in Shanghai, China. The results validated the proposed method with absolute percentage errors (APEs) of approximately 5% in constrained traffic flow conditions and 10–20% in less constrained traffic flow. The results also show that the proposed method has better performance than the method with loop detector data and another data fusion model. It is expected that the proposed method could help improve traffic management efficiency, such as traffic signal control, by providing more accurate travel time information.

Graph Convolutional Gated Recurrent Unit Network for Traffic Prediction Using Loop Detector Data

Traffic prediction is challenging due to the stochastic nonlinear dependencies in spatiotemporal traffic characteristics. We introduce a Graph Convolutional Gated Recurrent Unit Network (GC-GRU-N) to capture the critical spatiotemporal dynamics. Using 15-min aggregated Seattle loop detector data, we recontextualize the prediction challenge across space and time. We benchmark our model against Historical Average, LSTM, and Transformers. While Transformers outperformed other models, our GC-GRU-N came in a close second with notably faster inference time — six times quicker than Transformers. We offer a comprehensive comparison of all models based on training and inference times, MAPE, MAE, and RMSE. Furthermore, we delve into the spatial and temporal characteristics of each model’s performance.

Traffic Performance of Successive Freeway Merge Segments

At freeway interchanges with high entering flows, implementing two successive merges can improve traffic flow and capacity by distributing the merging maneuvers. The German Guidelines for the Design of Motorways define different types of successive (“double”) merges. According to the quality of service assessment procedure for freeway merge segments given in the German Highway Capacity Manual, successive merge segments can be evaluated separately. However, depending on the distance between the merges, the capacity of the second (downstream) merge might be affected by the entering traffic at the first (upstream) merge, which influences the lane-flow distribution. For a more precise evaluation of traffic flow performance and traffic safety depending on the geometric design characteristics, vehicle interactions, lane changes, the resulting capacity of successive merge segments, accident rates, and crash types were analyzed in the study. The research was based on loop detector data as well as trajectory data obtained from video measurements, which provide an insight into the interactions between mainline and entering vehicles. For the analysis of traffic safety, crash data were evaluated over a period of 3 years. Furthermore, successive merge segments with different geometric and traffic characteristics were modeled with the microscopic traffic simulation tool BABSIM to analyze the influence on traffic flow and capacity. As a result, capacity models and recommendations for the geometric design of successive freeway merge segments are provided.

Integrated self-consistent macro-micro traffic flow modeling and calibration framework based on trajectory data

Calibrating microscopic car-following (CF) models is crucial in traffic flow theory as it allows for accurate reproduction and investigation of traffic behavior and phenomena. Typically, the calibration procedure is a complicated, non-convex optimization issue. When the traffic state is in equilibrium, the macroscopic flow model can be derived analytically from the corresponding CF model. In contrast to the microscopic CF model, calibrated based on trajectory data, the macroscopic representation of the fundamental diagram (FD) primarily adopts loop detector data for calibration. The different calibration approaches at the macro- and microscopic levels may lead to misaligned parameters with identical practical meanings in both macro- and micro-traffic models. This inconsistency arises from the difference between the parameter calibration processes used in macro- and microscopic traffic flow models. Hence, this study proposes an integrated multiresolution traffic flow modeling framework using the same trajectory data for parameter calibration based on the self-consistency concept. This framework incorporates multiple objective functions in the macro- and micro-dimensions. To expeditiously execute the proposed framework, an improved metaheuristic multi-objective optimization algorithm is presented that employs multiple enhancement strategies. Additionally, a deep learning technique based on attention mechanisms was used to extract stationary-state traffic data for the macroscopic calibration process, instead of directly using the entire aggregated data. We conducted experiments using real-world and synthetic trajectory data to validate our self-consistent calibration framework.

Lane‐level short‐term travel speed prediction for urban expressways: An attentive spatio‐temporal deep learning approach

AbstractNumerous efforts have been made to address the section‐level travel speed prediction problem. However, section‐level predictions can hardly be used for fine‐grained applications, such as lane management and lane‐level navigation. The main reason for this is that significant speed heterogeneity exists among the lanes within one section. Thus, this study proposes a three‐dimensional (3D) dual attention convolution‐based deep learning model for predicting the lane‐level travel speed. 3D convolutions are designed to learn high‐dimensional spatiotemporal traffic flow features, that is, the relationships between different sections, lanes, and periods. Dual attention modules are used to focus on the traffic flow propagation patterns and to explain the model's mechanisms. To evaluate the proposed model, an indicator is introduced to assess the spatio‐temporal learning ability, based on targeting the lane‐level case. Evaluation experiments are conducted based on loop detector data in Shanghai, China. The results show that high accuracy is obtained by the proposed model, with a 2.9 km/h mean absolute error, thereby outperforming several existing methods. Finally, an in‐depth analysis is provided regarding the attention coefficients and interpretation of real‐world lane‐level traffic flow propagation patterns, so as to gain insights into the model's mechanism when capturing dynamic lane‐level traffic flow.

Estimation of a Fundamental Diagram with Heterogeneous Data Sources: Experimentation in the City of Santander

The reduction of urban congestion represents one of the main challenges for increasing sustainability. This implies the necessity to increase our knowledge of urban mobility and traffic. The fundamental diagram (FD) is a possible tool for analyzing the traffic conditions on an urban road link. FD is commonly associated with the links of a transport network, but it has recently been extended to the whole transport network and named the network macroscopic fundamental diagram (NMFD). When used at the link or network level, the FD is important for supporting the simulation, design, planning, and control of the transport system. Recently, floating car data (FCD), which are based on vehicles’ trajectories using GPS, are able to provide the trajectories of a number of vehicles circulating on the network. The objective of this paper is to integrate FCD with traffic data obtained from traditional loop-detector technology for building FDs. Its research contribution concerns the proposal of a methodology for the extraction of speed data from taxi FCD, corresponding to a specific link section, and the calibration of FDs from FCD and loop detector data. The methodology has been applied to a real case in the city of Santander. The first results presented are encouraging, supporting the paper’s thesis that FCD can be integrated with data obtained from loop detectors to build FD.

Flow count data-driven static traffic assignment models through network modularity partitioning

AbstractAccurate static traffic assignment models are important tools for the assessment of strategic transportation policies. In this article we present a novel approach to partition road networks through network modularity to produce data-driven static traffic assignment models from loop detector data on large road systems. The use of partitioning allows the estimation of the key model input of Origin–Destination demand matrices from flow counts alone. Previous network tomography-based demand estimation techniques have been limited by the network size. The amount of partitioning changes the Origin–Destination estimation optimisation problems to different levels of computational difficulty. Different approaches to utilising the partitioning were tested, one which degenerated the road network to the scale of the partitions and others which left the network intact. Applied to a subnetwork of England’s Strategic Road Network and other test networks, our results for the degenerate case showed flow and travel time errors are reasonable with a small amount of degeneration. The results for the non-degenerate cases showed that similar errors in model prediction with lower computation requirements can be obtained when using large partitions compared with the non-partitioned case. This work could be used to improve the effectiveness of national road systems planning and infrastructure models.

Effects of loop detector position on the macroscopic fundamental diagram

Loop detectors are probably the widest-used technology for traffic state estimation. Previous research has shown that loop detector positions within the link significantly affect the estimation of the macroscopic fundamental diagram (MFD) of a given network. This paper examines the biases produced by the positioning of loop detectors on the MFD, using both analytical and simulation methods, as well as empirical data from UTD19. We confirm earlier results that a uniform distribution of loop detector positions reduces the bias. We discovered that: (i) subsets of the MFD determined by the loop detector position can help estimate whether the loop detector MFD will have a bias; (ii) non-uniform distribution of loop detectors is more likely to cause a discrepancy in the position subsets of the MFD, particularly if detectors in the network are positioned more downstream with a greater variation; and (iii) a lower ratio of link length to green signal time increases the possibility of bias in loop detector MFD, while the impact of the aggregation interval was found to be negligible. This research opens the possibility for the bias of MFD induced by the loop detector data to be corrected by only using itself.

Multivariate time-varying Kalman filter approach for cycle-based maximum queue length estimation

Real-time queue length is a crucial information, increasingly needed for signal operation and signal optimization purposes. This paper proposes a novel multivariate time-varying Kalman filter approach to estimate the cycle-based maximum queue lengths in real-time by only using high-resolution vehicle loop detector data and signal timing data. The modeling framework builds on state changing points that can be identified at detector site and shockwave theory that models the propagation of traffic states in the space–time domain. The times at which traffic states change at the detector site, known as break points (e.g., free flow to jam condition, jam to saturation condition, etc.) are identified using high resolution detector data and the end of the queue is estimated using linear models that build on the shockwave theory. Nevertheless, this deterministic approach, which is previously proposed in the literature, does not account for measurements errors associated with traffic state changing points or the assumptions in the shockwave theory. Therefore, we propose a multivariate time-varying Kalman filter approach to produce a robust estimate for the cycle-based maximum queue length in both undersaturated and oversaturated conditions. Further, the developed algorithm can estimate the residual queue length and inform its occurrence in case of oversaturation. The proposed methodology can easily be adapted in practice, as it relies solely on measurements from a single loop detector that may be located close to the stop line.

Analytical formulation for explaining the variations in traffic states: A fundamental diagram modeling perspective with stochastic parameters

Despite the simplicity and practicality of (deterministic) fundamental diagram models in highway traffic flow theory, the wide scattering effect observed in empirical data remains highly controversial, particularly for explaining traffic state variations. Owing to the analytical properties of the fundamental diagram modeling approach, in this study, we proposed an analytical and quantitative method for analyzing traffic state variations. We investigated the scattering effect in the fundamental diagram and proposed two stochastic fundamental diagram (SFD) models with lognormal and skew-normal distributions to explain the variations in traffic states. The first SFD model assumes that the scattering effect results from stochasticity in both the free-flow speed and the speed at critical density. Both random variables were assumed to follow the lognormal distribution. In the second SFD model, an integrated error term that was assumed to follow the skew-normal distribution over different density ranges was appended to the deterministic fundamental diagram. The properties of these two SFD models were analyzed and compared, and the parameters in these SFD models were calibrated using real-world loop detector data. The observed scatters from the empirical data were reproduced well by the simulated fundamental diagram model, indicating the validity of the proposed SFD models for explaining traffic state variations. Using these two analytical SFD models, we can analyze the stochastic capacity of freeways with closed forms. More importantly, the sources of stochasticity in freeway capacity can be traced in terms of randomly distributed parameters in fundamental diagram models.

Anti-circulant dynamic mode decomposition with sparsity-promoting for highway traffic dynamics analysis

Highway traffic state data collected from a network of sensors can be considered as a high-dimensional nonlinear dynamical system. In this paper, we develop a novel data-driven method–anti-circulant dynamic mode decomposition with sparsity-promoting (circDMDsp)–to study the dynamics of highway traffic speed data. Particularly, circDMDsp addresses several issues that hinder the application of existing DMD models: limited spatial dimension, presence of both recurrent and non-recurrent patterns, high level of noise, and known mode stability. The proposed circDMDsp framework allows us to numerically extract spatial–temporal coherent structures with physical meanings/interpretations: the dynamic modes reflect coherent spatial bases, and the corresponding temporal patterns capture the temporal oscillation/evolution of these dynamic modes. Our result based on Seattle highway loop detector data showcases that traffic speed data is governed by a set of periodic components, e.g., mean pattern, daily pattern, and weekly pattern, and each of them has a unique spatial structure. The spatiotemporal patterns can also be used to recover/denoise observed data and predict future values at any timestamp by extrapolating the temporal Vandermonde matrix. Our experiments also demonstrate that the proposed circDMDsp framework is more accurate and robust in data reconstruction and prediction than other DMD-based models. The code for the algorithm and experiment is available at https://github.com/mcgill-smart-transport/circDMDsp.

A coordinated ramp metering framework based on heterogeneous causal inference

AbstractThe coordinated ramp metering strategy aims to enhance the traffic flow on freeways by integrating the effects of multiple on‐ramps. The success of this strategy depends heavily on the metering effects of each on‐ramp and the cooperation pattern. To ensure the effectiveness of coordination control, it is essential to design the coordinated strategy based on the causal effect of each ramp flow on the freeway bottleneck. Therefore, this study investigated the heterogeneous causal effects of ramp flow on freeway bottleneck and proposed a coordinated ramp metering framework based on these causal effects. Causal inference was conducted using loop detector data. The estimated heterogeneous causal effect, which varies across traffic conditions, measures the change of bottleneck occupancy resulting from a unit increase in each on‐ramp outflow. This information quantifies the importance of each ramp in real time and can be used to update the ramp weight and metering rate of each ramp controller. To comprehensively test the proposed coordination framework, two instances with different road networks were selected. In the first instance, limiting the on‐ramp flow downstream of the bottleneck was found to be helpful in mitigating the bottleneck. Compared with isolated ramp metering, the proposed algorithm reduced travel delay by 26.0% by considering the heterogeneous causal effects of on‐ramp flow both upstream and downstream of the freeway bottleneck. In the next instance, the proposed algorithm reduced travel delay by 5.8%, compared to a well‐performed feedback‐based coordinated ramp metering algorithm. The results of this study suggest that the use of heterogeneous causal inference is effective in improving the performance of coordinated ramp metering algorithms.

Investigating speed-safety association: Considering the unobserved heterogeneity and human factors mediation effects.

The relationship between mean speed and crash likelihood is unclear in the literature. The contradictory findings can be attributed to the masking effects of the confounding variables in this association. Moreover, the unobserved heterogeneity has almost been criticized as a reason behind the current inconclusive results. This research provides an effort to develop a model that analyzes the mean speed-crash frequency relationship by crash severity and type. Also, the confounding and mediation effects of the environment, driver, and traffic-related attributes have been considered. To this end, the loop detector and crash data were aggregated daily for rural multilane highways of Tehran province, Iran, covering two years, 2020-2021. The partial least squares path modeling (PLS-PM) was employed for crash causal analysis along with the finite mixture partial least squares (FIMIX-PLS) segmentation to account for potential unobserved heterogeneity between observations. The mean speed was negatively and positively associated with the frequency of property damage-only (PDO) and severe accidents, respectively. Moreover, driver-related variables, including tailgating, distracted driving, and speeding, played key mediation roles in associating traffic and environmental factors with the crash risk. The higher the mean speed and the lower the traffic volume, the higher odds of distracted driving. Distracted driving was, in turn, associated with the higher vulnerable road users (VRU) accidents and single-vehicle accidents, triggering a higher frequency of severe accidents. Moreover, lower mean speed and higher traffic volume were positively correlated with the percentage of tailgating violations, which, in turn, predicted multi-vehicle accidents as the main predictor of PDO crash frequency. In conclusion, the mean speed effects on the crash risk are entirely different for each crash type through distinct crash mechanisms. Hence, the distinct distribution of crash types in different datasets might have led to current inconsistent results in the literature.

Cycle-Based Estimation on Lane-Level Queue Length at Isolated Signalized Intersection Using License Plate Recognition Data

Queue length is an important parameter to evaluate the congestion level of urban intersections. Abundant methods based on different data sources, such as loop detector data and mobile sensor data, for estimating queue length have received wide attention. A license plate recognition (LPR) system recording vehicle information can be used as a data source to estimate lane-level queue length. In this study, an improved method is proposed for cycle-based queue length estimation at an isolated signalized intersection using LPR data, considering the residual queued vehicles under oversaturated conditions. First, a modified interpolation method is developed to infer the travel time of unmatched vehicles. Then, the complete arrival and departure information are processed as three characteristic parameters of vehicles, which are the input to the maximum probability function for queue length estimation at each lane. Finally, the proposed model is validated using actual cycle maximum queue length collected from a video camera and LPR data in Changsha city, China. The numerical results showed that the proposed model can achieve accurate estimation for lane-level queue length.

Understanding sudden traffic jams: From emergence to impact

Road traffic jams are a major problem in most cities of the world, resulting in massive delays, increased fuel wastage, and monetary and productivity losses. Unlike conventional computer networks, which experience congestion due to excessive traffic, road transportation networks can experience traffic jams over prolonged periods due to traffic bursts over short time scales that push the traffic density beyond a threshold jam density. We observe that the emergence of such jams can happen over a very short duration, hence we term them as sudden traffic jams. We provide a formalism for understanding the phenomena of sudden traffic jams and show evidence of its existence using loop detector data from New York City. Further, we show the signature of sudden jams when observed at hourly resolution. We also provide a method to compute the traffic curve in a situation where we do not have access to fine-grained flow and density information. With this method, using only hourly speed data from Uber, we compute traffic curves for the road segments in Nairobi, São Paulo, and New York City, which is, by our knowledge, the first attempt to do so for signalized road networks. Running our analysis on the Uber movement speed data for the three cities, we show numerous instances of jams that last several hours, and sometimes as long as 2–3 days. Empirically, we find that Nairobi experiences 3x the mean jam time per road segment as compared to São Paulo and New York City. Based on key development metrics, we find that the ratio of traffic load per road segment for São Paulo, New York City, and Nairobi is approximately 1:2:3. We propose that chaotic driving patterns and traffic mismanagement in the developing world cities lead to tighter traffic curves, more intense jams and overall lower road capacity utilization, which explains the observed data. We posit that the problem of traffic congestion in developing countries cannot be solved entirely by building new infrastructure, but also requires smart management of existing road infrastructure.

Lane-level trajectory reconstruction based on data-fusion

While lane-changing movements are performed on the entire motorway network for various reasons, such as overtaking, their intensity is substantially greater near complex segments, such as weaving areas. Aside from mandatory lane changes, some drivers also conduct lateral maneuvers for cooperation or anticipation near network nodes. Unlike longitudinal driver behavior (car-following models), lateral driver behavior (lane-changing movements) has received fewer research efforts. The scarcity of suitable data resources to analyze these behaviors and movements might be a crucial cause for this research gap. This paper presents a four-step approach for reconstructing and correcting lateral bias in trajectories collected by a commercial traffic information application running on everyday smartphones. The resulting lateral position is accurate enough to allow for identification of the driving lane, and thus, the lane changes. The algorithm's core is built on a data fusion method using trajectory and loop detector data. The evaluation and validation of the proposed algorithm using drones and closed-circuit television (CCTV) data demonstrate that the core of the algorithm can correctly match more than 94% of trajectory and loop detector data. Between each pair of successive detector stations, the lateral position error has been significantly corrected and reduced to less than half the width of a standard lane of motorway networks. As a result, more than 90% of processed trajectory sample points are in the correct lane. The algorithm requires just two calibration parameters, so it is relatively simple to apply to other test networks.

Deep representation of imbalanced spatio‐temporal traffic flow data for traffic accident detection

AbstractAutomatic detection of traffic accidents has a crucial effect on improving transportation, public safety, and path planning. Many lives can be saved by the consequent decrease in the time between when the accidents occur and when rescue teams are dispatched, and much travelling time can be saved by notifying drivers to select alternative routes. This problem is challenging mainly because of the rareness of accidents and spatial heterogeneity of the environment. This paper studies deep representation of loop detector data using long‐short term memory (LSTM) network for automatic detection of freeway accidents. The LSTM‐based framework increases class separability in the encoded feature space while reducing the dimension of data. The experiments on real accident and loop detector data collected from the Twin Cities Metro freeways of Minnesota demonstrate that deep representation of traffic flow data using LSTM network has the potential to detect freeway accidents in less than 18 min with a true positive rate of 0.71 and a false positive rate of 0.25 which outperforms other competing methods in the same arrangement.