Monitoring traffic streams through vehicle trajectories offers valuable insights into traffic flow characteristics. In recent years, there has been a surge in the availability of vehicle trajectory datasets. At the same time, the number of autonomously-driven vehicles on the road is increasing, largely due to the adoption of systems like adaptive cruise control. However, distinguishing system-controlled vehicles from human-driven vehicles remains challenging, despite its potential to enable valuable applications and informed policy-making. The differences between the driving behavior of human-driven (HDs) and automated (ADs) vehicles in the longitudinal direction are highlighted in the literature and hold promise for novel methodologies that exploit them to identify the type of driver. Here, we propose a novel online-offline framework with three key contributions. First, a feature design component performs feature disentanglement to increase the performance of downstream deep learning models. Second, a bidirectional LSTM (bLSTM) architecture demonstrates excellent accuracy in differentiating between HD and AD vehicles. Third, a data drift detection component identifies changes in data distributions, enabling the framework to generalize effectively to unseen datasets with minimal new labeled observations.

Network-wide Traffic State Estimation (TSE), which aims to infer a complete image of network traffic states with sparsely deployed sensors, plays a vital role in intelligent transportation systems. With the development of data-driven methods, traffic dynamics modeling has advanced significantly. However, TSE poses fundamental challenges for data-driven approaches, since historical patterns cannot be learned locally at sensor-free segments. Although graph representation learning shows promise in estimating states at locations without sensors, existing methods typically handle unobserved locations by filling them with zeros, introducing bias to the sensitive graph message propagation. The recently proposed Dirichlet Energy-based Feature Propagation (DEFP) method achieves State-Of-The-Art (SOTA) performance in unobserved node classification by eliminating the need for zero-filling. However, applying it to TSE faces three key challenges: inability to handle directed traffic networks, strong assumptions in traffic spatial correlation modeling, and overlooking distinct propagation rules of different patterns (e.g., congestion and free flow). We propose DGAE, a novel inductive graph representation model that addresses these challenges through theoretically derived DEFP for Directed graph (DEFP4D), enhanced spatial representation learning via DEFP4D-guided latent space encoding, and physics-guided propagation mechanisms that separately handle congested and free-flow patterns. Experiments on three traffic datasets demonstrate that DGAE outperforms existing SOTA methods and exhibits strong cross-city transferability. Furthermore, DEFP4D can serve as a standalone lightweight solution, showing superior performance under extremely sparse sensor conditions. The code of this work is publicly available at:https://github.com/ZJU-TSELab/DGAE

Inference of detailed vehicle trajectories is crucial for applications such as traffic flow modeling, energy consumption estimation, and traffic flow optimization. Static sensors can provide only aggregated information, posing challenges in reconstructing individual vehicle trajectories. Shockwave theory is used to reproduce oscillations that occur between sensors. However, as the emerging of connected vehicles grows, probe data offers significant opportunities for more precise trajectory reconstruction. Existing methods rely on Eulerian observations (e.g., data from static sensors) and Lagrangian observations (e.g., data from connected vehicles) incorporating shockwave theory and car-following modeling. Despite these advancements, a prevalent issue lies in the static assignment of shockwave speed, which may not be able to reflect the traffic oscillations in a short time period caused by varying response times and vehicle dynamics. Moreover, driver dynamics while reconstructing the trajectories are ignored. In response, this paper proposes a novel framework that integrates Eulerian and Lagrangian observations for trajectory reconstruction on freeways. The approach introduces a calibration algorithm for time-varying shockwave speed. The shockwave speed calibrated by the CV is then utilized for trajectory reconstruction of other non-connected vehicles based on shockwave theory. Additionally, vehicle and driver dynamics are introduced to optimize the trajectory and estimate energy consumption by applying a vehicle movement model. The proposed method is evaluated using real-world datasets, demonstrating superior performance in terms of trajectory accuracy, reproducing traffic oscillations, and estimating energy consumption.

Recently, efforts have been made to standardize signal phase and timing (SPaT) messages. These messages contain signal phase timings of all signalized intersection approaches. This information can thus be used for efficient motion planning, resulting in more homogeneous traffic flows and uniform speed profiles. Despite efforts to provide robust predictions for semi-actuated signal control systems, predicting signal phase timings for fully-actuated controls remains challenging. This paper proposes a time series prediction framework using aggregated traffic signal and loop detector data. We utilize state-of-the-art machine learning models to predict future signal phases’ duration. The performance of a Linear Regression (LR), Random Forest (RF), a light gradient-boosting machine (LightGBM), a bidirectional Long-Short-Term-Memory neural network (BiLSTM) and a Temporal Convolutional Network (TCOV) are assessed against a naive baseline model. Results based on an empirical data set from a fully-actuated signal control system in Zurich, Switzerland, show that state of the art machine learning models outperform conventional prediction methods.

Hybrid electric vehicles (HEVs) have reached the market share required to meaningfully affect many aspects of the road transport system, including traffic behaviour, energy consumption, and emissions. However, traffic models for hybrids remain insufficiently addressed in microscopic simulation because traditional models ignore vehicle dynamics, and therefore, cannot capture driving differences that exist among the hybrid, conventional, and electric vehicles. This study extends the lightweight microsimulation free-flow acceleration (MFC) model and fills the above gap in the literature by introducing hybrid vehicle dynamics into traffic simulation. First, the methodology underlying the MFC model to reproduce hybrid vehicle dynamics is described, for both charge depleting (CD) and charge sustaining (CS) modes. Then, the experimental setup for model validation and implementation was introduced. The simulations suggest the proposed MFC model can ensure smooth speed and acceleration profiles while converging to the steady state. The results show the MFC model can accurately capture the dynamics of the hybrid vehicle tested on the chassis dynamometer. The MFC model is compared with the Gipps’ model and the intelligent driver model (IDM) regarding their abilities to reproduce driving trajectories of the hybrid vehicle. It was found, in CD mode, the MFC model leads to reductions in both speed and acceleration root mean square errors (RMSEs). In CS mode, the MFC model yields even greater accuracy gains. When predicting the 0-100 km/h acceleration specifications, the MFC model also outperforms the Gipps’ and the IDM, reducing RMSE by 45.8 % and 51.9 %, respectively.

Evidence in the literature shows that automated and human driving modes demonstrate different driving characteristics, i.e., headway policy, spacing policy, reaction time, comfortable acceleration, and others. These differences alter observed traffic dynamics and have an impact on energy consumption. This paper assesses the energy footprint of commercially implemented adaptive cruise control (ACC) systems and human drivers in car-following formation via different models using empirical observations on very similar driving cycles and/or routes. Most importantly, it initiates a critical discussion of the findings under the behavioral properties of each mode. Findings show that: ACC systems propagate an increasing energy consumption upstream, while human drivers do not; they succeed in maintaining a constant time-headway policy, operating very reliably; they develop a strong bond with their leader compared to their human counterparts; the two modes (humans and ACCs) are operating in different phase-space areas with room for improvement. Overall, findings show that ACC systems must be optimized to achieve a trade-off between functional requirements and eco-driving instructions.

This paper analyzes trajectory observations from vehicles driving in platoon formation and they are equipped with Adaptive Cruise Control (ACC) and Cooperative Adaptive Cruise Control (CACC) systems; aiming to quantify response delays. When a preceding vehicle induces a perturbation, the delay until the reaction of the following vehicle, often quoted as observable response time, can have negative implication to the traffic flow and other factors such as energy consumption, stability and safety. Quantifying such delays can help towards more realistic traffic simulation modeling. The analysis is performed based on empirical observations from three well-known experimental campaigns in the literature with data from ACC-driven and CACC-driven vehicle platoons. Three state-of-the-art techniques were implemented to provide quantitative results for the observed response times. The benefits and downsides of each technique are discussed as well. The results show that ACC systems do not exhibit a significant improvement compared to human drivers, yet, it can be concluded that the additional vehicle-to-vehicle communication incorporated in the CACC systems allows for a substantially higher traffic flow and possibly other benefits.

Road vehicles are characterized by increasing levels of automation and it is vital to understand the future impact on transport efficiency. Adaptive Cruise Control (ACC) is one of the first and most common automated functionalities available in privately owned vehicles. The effect of ACC on traffic flow has been widely studied by making assumptions on its operating strategy and on some of its important parameters such as the response time and the desired time headway. In the literature, these parameters are usually set to low values, based on the vehicle controller's theoretical ability to respond within a very short time frame. Response time is known to be an important parameter in defining the capacity of the road and therefore, assuming a very short response time, studies usually conclude that systems like the ACC will contribute increasing the road capacity significantly. The present study aims at measuring the actual response time of an ACC-enabled vehicle in car-following conditions. A new methodology for the estimation of the controller's response time and the desired time-gap was developed to this objective. Results show that the response time of the particular ACC controller was in the range 0.8s-1.2s, which is similar to what is commonly assumed for human drivers. In this light, the results of the present study question the common assumption that ACC or other automation technologies necessarily improve traffic flow and increase road capacity.

Vehicle automation and cooperation is progressively being introduced in traffic networks. As a consequence research to assess its impacts is currently on-going. Adaptive Cruise Control (ACC) is one of the first automated functionalities available for privately owned vehicles. An experimental study has been conducted to investigate the key features of the ACC controller using Global Navigation Satellite System data. The first remarks based on the data focus on the controller's reaction time and desired time gap. Both parameters are essential in order to assess the influence of these technologies to safety and traffic flow. It is a common assumption that autonomous vehicles will have negligible reaction time and desired time gap comparable to that of a human driver. This paper presents an experimental study of an ACC-enabled vehicle on car following mode and a methodology for the estimation of the controller's reaction time that can be used as benchmark in other scenarios. The results show the reaction time to be around l.ls and the time gap to be distinctly larger than that of a human driver. This poses concern on the impact of ACC on traffic flow when a significant number of vehicles will have such systems operating on-board.

Microscopic traffic simulation models are widely used to assess the impact of measures and technologies on the road transportation system. The assessment usually involves several measures of performance, such as overall traffic conditions, travel time, energy demand/fuel consumption, emissions, and safety. In doing so, it is usually assumed that traffic models are able to capture not only traffic dynamics but also vehicle dynamics (especially to compute energy/fuel consumption, emissions, and safety). However, this is not necessarily the case with the possibility of achieving unreliable outcomes when extrapolating from traffic to measures of performance related to the vehicle dynamics. The objective of the present paper is to assess the capability of existing car-following models to reproduce observed vehicle acceleration dynamics. A set of experiments was carried out in the Vehicle Emissions Laboratories of the European Commission Joint Research Centre in order to generate relevant data sets. These experiments are used to test the performance of three well-known car-following models. Although all models have been largely tested against their capability to correctly reproduce traffic dynamics, the findings raise concerns about their capability (and thus of the traffic models using them) to predict the effect on the microscopic vehicle dynamics and thus on emissions and energy/fuel consumption. The results of the present work can be considered valid beyond the analyzed car-following models, as simple acceleration rules are usually assumed in the vast majority of the traffic simulation frameworks. Consequently, it can be concluded that there is a number.

This paper presents a new technique for adaptive binarization of degraded document images. The proposed technique focuses on degraded documents with various background patterns and noise. It involves a preprocessing local background estimation stage, which detects for each pixel that is considered as background one, a proper grayscale value. Then, the estimated background is used to produce a new enhanced image having uniform background layers and increased local contrast. That is, the new image is a combination of background and foreground layers. Foreground and background layers are then separated by using a new transformation which exploits efficiently, both grayscale and spatial information. The final binary document is obtained by combining all foreground layers. The proposed binarization technique has been extensively tested on numerous documents and successfully compared with other well-known binarization techniques. Experimental results, which are based on statistical, visual and OCR criteria, verify the effectiveness of the technique.

This paper proposes a new technique for solving jigsaw puzzles. The novelty of the proposed technique is that it provides an automatic jigsaw puzzle solution without any initial restriction about the shape of pieces, the number of neighbor pieces, etc. The proposed technique uses both curve- and color-matching similarity features. A recurrent procedure is applied, which compares and merges puzzle pieces in pairs, until the original puzzle image is reformed. Geometrical and color features are extracted on the characteristic points (CPs) of the puzzle pieces. CPs, which can be considered as high curvature points, are detected by a rotationally invariant corner detection algorithm. The features which are associated with color are provided by applying a color reduction technique using the Kohonen self-organized feature map. Finally, a postprocessing stage checks and corrects the relative position between puzzle pieces to improve the quality of the resulting image. Experimental results prove the efficiency of the proposed technique, which can be further extended to deal with even more complex jigsaw puzzle problems.

Word segmentation is a crucial step for segmentation-free document analysis systems and is used for creating an index based on word matching. In this paper, we propose a novel methodology for word segmentation in historical and degraded machine-printed documents. The proposed technique faces problems such as having text of different size, having text and non-text areas lying very near and having non-straight and warped text lines. It is based on: (i) a dynamic run length smoothing algorithm that helps grouping together homogeneous text regions, (ii) noise and punctuation marks removal as well as on obstacle detection in order to facilitate the segmentation process and (iv) a draft text line estimation procedure that guides the final word segmentation result. After testing on numerous historical and degraded machine-printed documents, it has turned out that our methodology performs better compared to current state-of-the-art word segmentation techniques for historical and degraded machine-printed documents.

A new technique for solving jigsaw puzzles is proposed, which takes advantage of both geometrical and color features. It is considered that an image is being divided into a number of pieces (sub-images). The proposed technique is based on extraction of a set of boundary characteristic points and on a Kohonen self-organized feature map (KSOFM) color reduction technique. For each characteristic point a set of color and geometrical features are extracted. The technique compares these sets of features and decides whether two sub-images match or not. When a matching pair has been found, a corrective procedure is applied in order for these sub-images to fit exactly. Next, the proposed technique creates a new sub-image, which consists of the two matched sub-images. The whole matching procedure is being repeated until only one sub-image remains or no more matching sub-images can be found.