Vehicle Trajectory Dataset Research Articles

The dynamic tradeoff between safety and efficiency in discretionary lane-changing behavior: A random parameters logit approach with heterogeneity in means and variances

The discretionary lane-changing behavior is highly associated with crash occurrence. This paper investigates the dynamic tradeoff between safety and efficiency and influencing factors of the decision-making process for the discretionary lane-change. A random parameters logit approach with heterogeneity in means and variances is employed on a microscopic vehicle trajectory dataset. Four vehicles are analyzed as a discretionary lane-changing group, and both the distance-related and speed-related factors are taken into account to examine their dynamic effects. The modeling results indicate that (1) two distance-related variables produce significant random parameters with heterogeneity in means and variances, while there is only one speed-related variable having significant result for heterogeneity in means and variances; (2) the relationship between the speed of lane-changing vehicle and the heterogeneity in mean of vehicle clearance distance between leading vehicle in current lane and lane-changing vehicle quantitatively demonstrates that when the driver decides whether to change lanes, the driver considers the tradeoff between distance (i.e., safety) and speed (i.e., efficiency) at the same time; and (3) the number of significant variables is gradually decreased during the lane-changing decision-making process. It demonstrates that the lane-changing decision is a dynamic process where drivers gradually select more critical factors and pay more attention to them.

A Vehicle Trajectory Extraction Method for Traffic Simulating Modeling

Vehicle driving behavior in urban traffic environment is interactive and complex. In order to simulate the driving behavior of real vehicles, it needs to rely on a large number of real vehicle trajectory data. However, making the simulated model more consistent with the real situation, a large vehicle trajectory dataset is needed. Therefore, this paper proposes a method to automatically extract accurate vehicle trajectory from video captured by UAV camera. The proposed computer vision method obtains the trajectory by detecting the vehicle and then tracking the trajectory, and calculates the relevant parameters of the vehicle, such as position, speed and direction. Therefore, the extracted data can be used as real driving behavior data for simulation modeling. The experimental results show that the proposed method is effective and reliable.

Tight Approximation Ratios of Two Greedy Algorithms for Optimal RSU Deployment in One-Dimensional VANETs

For addressing the One-Dimensional Road side unit Deployment (D1RD) problem, a greedy approximate algorithm named Greedy2P3E was proposed two years ago, and its approximation ratio was proved to be at least 2/3 for the D1RD problem with EQual-radius RSUs (D1RD-EQ problem). Can better or even tight approximations for Greedy2P3E be found? In this paper, approximation ratio of Greedy2P3E is re-inspected and tight approximation ratio is found. To this end, a greedy algorithm named Greedy3P4 is first proposed and proved to have a tight approximation ratio of 3/4 for the D1RD-EQ problem. Then, by using Greedy3P4 as a bridge, 3/4 is also proved to be the tight approximation ratio of Greedy2P3E and it is tight for all $n{\geq }2$ . Comparative evaluations are performed on real cases using a real vehicle trajectory dataset. The results show that these greedy algorithms usually return near optimal solutions with a profit more than 98% of the optimal solutions, and the greedy algorithms well outperform the other typical algorithms tested.

Dynamic clustering analysis for driving styles identification

For intelligent driving systems, the ability to recognize different driving styles of surrounding vehicles is crucial in determining the safest, yet more efficient driving decisions especially in the context of the mixed driving environment. Knowing for instance if the vehicle in the adjacent lane is aggressive or cautious can greatly assist in the decision making of ego vehicle in terms of whether and when it is appropriate to make particular manoeuvres (e.g. lane change). In addition, vehicles behave differently under different surrounding environments, making the driving styles identification highly challenging. To this end, in this paper we propose a dynamic clustering based driving styles identification and profiling approach where clusters vary in response to the changing surrounding environment. To better capture dynamic driving patterns and understand the driving style switch behaviours and more complicated driving patterns, a position-dependent dynamic clustering structure is developed where a driver is assigned to a cluster sequence rather than a single cluster. To the best of our knowledge, this is the first research paper of its kind on the dynamic clustering of driving styles. The usefulness of the proposed method is demonstrated on a real-world vehicle trajectory dataset where results show that driving style switches and more complex driving behaviours can be better captured. The potential applications in intelligent driving systems are also discussed.

A new method of temporal and spatial risk estimation for lane change considering conventional recognition defects

Risky lane change behavior of drivers normally will pose some negative impacts on traffic safety. To ensure a lane change safe and prevent potential accidents, it is important to recognize some lane-changing conditions with potential risks in advance. Despite the researches on traffic safety assessment have been developed for decades, most of the existing researches are mainly interesting in how to estimate the overall safety of the lane-changing process based on historical data. These methods tend to ignore the interactive impacts between lane-changing vehicle and its surrounding vehicles, and have the disadvantages of the long-term evaluation period and single evaluation index. To address these gaps, this study presents a temporal and spatial risk estimation (TSRE) to recognize lane-changing risk in real-time. However, this study concentrates on the instantaneous risk for a lane change event, considering temporal and spatial dimensions for the current lane change circumstance on the highway. After processing the realistic vehicle trajectory dataset, this study extracted 1444 groups of lane change samples, and then incorporates the temporal risk level (TRL) and spatial risk level (SRL) into a comprehensive risk index by applying fault tree analysis. Furthermore, SRL and comprehensive risk index are both used to determine whether the traffic condition of a lane change is safe, and it can effectively overcome the conventional recognition defects that existed in other methods. To achieve a better evaluation effect, the sensitivity tests of recognition accuracy for various risk threshold combinations were carried out. Ultimately, experimental results showed that the proposed TSRE model achieved 97.45 %, 97.79 %, 84.1 % and 85.05 % accuracy rate in terms of classifying the risky lane change samples, traffic conflicts, risky frames and safe frames, when an appropriate risk threshold combination was selected. This encouraging finding can provide the basis for algorithm design of the lane change warning system for connected vehicles.

Studying Car-Following Dynamics on the Basis of the HighD Dataset

A large-scale naturalistic vehicle trajectory dataset from German highways, highD, was used to investigate the car-following behavior of individual drivers. These data include trajectories of 110,000 vehicles recorded for a duration of 16.5 h. Solving a nonlinear optimization problem, the intelligent driver model and the optimal velocity model with two leaders in interaction were calibrated by minimizing the deviations between the observed and simulated gaps when following the prescribed leading vehicle. The obtained calibration errors ranged between 5.2% and 6.9%, which were slightly lower than previous findings. This was explained by the shorter highD trajectories, predominantly free-flow traffic, and the good precision metrics of this dataset. The optimal velocity model with multivehicle anticipation resulted in lower calibration errors. This confirmed that natural drivers take into account several leading vehicles ahead. The ratio between interdriver and intradriver variability was investigated by performing global and platoon calibrations. Intradriver variation accounted for a larger portion of the calibration errors than interdriver variation. We analyzed the acceleration time-series of the natural highD and artificial drivers using simulations of two car-following models. A new cumulative measure, proportional to the energy of the follower’s position time-series curve, was calculated both for natural and modeled drivers. Human drivers had higher energy and demonstrated more acceleration fluctuations, sometimes behaving irrationally. In contrast, artificial drivers followed the logical rules incorporated in the model, resulting in a smoother acceleration profile. This led to less fuel consumption and gas emissions.

Adaptive simplification of GPS trajectories with geographic context – a quadtree-based approach

ABSTRACT Big GPS trajectory datasets can have redundant spatio-temporal information for applications, which requires simplification as a key preprocessing for modeling. Many existing simplification methods focus on the geometric information from a trajectory per se. Conversely, methods considering geographic context often fail to provide spatially adaptive simplification, or require complex parameter settings to achieve this task. This study proposes a novel two-stage adaptive trajectory simplification method embedding spatial indexing, enrichment, and aggregation in an integrated process. The first stage employs a quadtree for the subdivision depending on the density of geographic context features (i.e. POIs), leading to a variable-resolution representation of the area. The second stage aggregates trajectory waypoints locating in the same quadtree leaf node into a representative point, making the aggregation adapting to the spatial layout of the geographic feature in the first stage. Evaluation with a real-world vehicle trajectory dataset shows that the proposed approach can automatically simplify trajectory segments at variable compression ratios with greater simplification in areas with sparse context features (e.g. rural) and less simplification in areas with dense context features (e.g. urban). More importantly, the method can still preserve inter-trajectory distances between original trajectories and simplified ones, while significantly reducing the computing time.

A novel facility location problem for taxi hailing platforms

PurposeMotivated by a problem in the context of DiDi Travel, the biggest taxi hailing platform in China, the purpose of this paper is to propose a novel facility location problem, specifically, the single source capacitated facility location problem with regional demand and time constraints, to help improve overall transportation efficiency and cost.Design/methodology/approachThis study develops a mathematical programming model, considering regional demand and time constraints. A novel two-stage neighborhood search heuristic algorithm is proposed and applied to solve instances based on data sets published by DiDi Travel.FindingsThe results of this study show that the model is adequate since new characteristics of demand can be deduced from large vehicle trajectory data sets. The proposed algorithm is effective and efficient on small and medium as well as large instances. The research also solves and presents a real instance in the urban area of Chengdu, China, with up to 30 facilities and demand deduced from 16m taxi trajectory data records covering around 16,000 drivers.Research limitations/implicationsThis study examines an offline and single-period case of the problem. It does not consider multi-period or online cases with uncertainties, where decision makers need to dynamically remove out-of-service stations and add other stations to the selected group.Originality/valuePrior studies have been quite limited. They have not yet considered demand in the form of vehicle trajectory data in facility location problems. This study takes into account new characteristics of demand, regional and time constrained, and proposes a new variant and its solution approach.

VTDP: Privately Sanitizing Fine-grained Vehicle Trajectory Data with Boosted Utility

With the rapidly growing deployment of intelligent transportation systems (ITS) and smart traffic applications, vehicle trajectory data are ubiquitously generated, e.g., from GPS navigation systems, mobile applications, and urban traffic cameras. Analyzing such fine-grained data would greatly benefit the development of ITS and smart cities, yet pose severe privacy risks due to the recorded drivers’ visited locations, routes, and driving habits. Recently, some privacy enhancing techniques were proposed to sanitize such data. However, such schemes have some major limitations–they either lack formal privacy notions to quantify and bound the privacy risks, or result in very limited utility, e.g., only a sequence of locations or aggregated information can be released (without retaining the speeds, accelerations and the timestamps of vehicles). In this article, we propose a novel framework to sanitize the fine-grained vehicle trajectories with differential privacy (VTDP), which provides rigorous privacy protection against adversaries who possess arbitrary background knowledge. Our VTDP technique involves three phases of differentially private sampling, which sequentially generate all the three categories of data (besides a pseudo identity for each vehicle)– position, moving, and timestamps . It also includes a vehicle trajectory interpolation procedure to further improve the output utility with the properties of fine-grained vehicle trajectory data. We conducted experiments on real vehicle trajectory datasets to validate the performance of our approach.

A Scalable Framework for Trajectory Prediction

Trajectory prediction (TP) is of great importance for a wide range of location-based applications in intelligent transport systems, such as location-based advertising, route planning, traffic management, and early warning systems. In the last few years, the widespread use of GPS navigation systems and wireless communication technology enabled vehicles has resulted in huge volumes of trajectory data. The task of utilizing these data employing spatio-temporal techniques for TP in an efficient and accurate manner is an ongoing research problem. Existing TP approaches are limited to the short-term predictions. Moreover, they cannot handle a large volume of trajectory data for long-term prediction. To address these limitations, we propose a scalable clustering and Markov chain-based hybrid framework, called Traj-clusiVAT-based TP, for both short- and long-term TPs, which can handle a large number of overlapping trajectories in a dense road network. Traj-clusiVAT can also determine the number of clusters, which represent different movement behaviors in input trajectory data. In our experiments, we compare our proposed approach with a mixed Markov model-based scheme and a trajectory clustering, NETSCAN-based TP method for both short- and long-term TPs. We performed our experiments on two real, vehicle trajectory datasets, including a large-scale trajectory dataset consisting of 3.28 million trajectories obtained from 15 061 taxis in Singapore over a period of one month. The experimental results on two real trajectory datasets show that our proposed approach outperforms the existing approaches in terms of both short- and long-term prediction performances, based on the prediction accuracy and distance error (in km).

Integrated deep learning and stochastic car-following model for traffic dynamics on multi-lane freeways

The current paper proposes a novel stochastic procedure for modelling car-following behaviours on a multi-lane motorway. We develop an integrated multi-lane stochastic continuous car-following model where a deep learning architecture is used to estimate a probability of lane-changing (LC) manoeuvres. To the best of our knowledge, this work is among the very few papers which exploit deep learning to model driving behaviour on a multi-lane road. The objective of this study is to establish a coupled stochastic continuous multi-lane car-following model using Langevin equations to cope with probabilistic characteristics of LC manoeuvres. In particular, a stochastic volatility, derived from LC manoeuvres is introduced in a multi-lane stochastic optimal velocity model (SOVM). In additions, Convolutional Neural Network (CNN) is applied to estimate a probability of LC manoeuvres in the integrated multi-lane car-following model. Furthermore, imaged second-based trajectories of the lane-changer and surrounding vehicles are used to identify whether LC manoeuvres occur by using the CNN. Finally, the proposed method is validated using a real-world high-resolution vehicle trajectory dataset. The results indicate that the prediction of the integrated SOVM is almost identical to the observed trajectories of the lane-changers and the following vehicles in the initial and the target lane. It has been found that the proposed multi-lane SOVM can tackle the unpredictable fluctuations in the velocity of the vehicles in the acceleration/deceleration zone.

A Novel Car-Following Control Model Combining Machine Learning and Kinematics Models for Automated Vehicles

The machine learning-based car-following models are widely adopted to control the longitudinal movements of automated vehicles, such as Google Car and Apple Car, by mimicking the human drivers’ car-following maneuver. However, like human drivers, the models easily produce unsafe maneuvers for automated vehicles and has low robustness, especially in uncommon situations. To improve the machine learning-based car-following models, this paper proposes to combine the machine learning models with the kinematics-based car-following models that can overcome the shortcomings of machine learning models, using an optimal combination prediction method, which is called the combination car-following model in the paper. The selected kinematics-based car-following model is the Gipps model that has an intrinsic crash-avoidance mechanism, and the used machine learning-based models are the Back-Propagation Neural Networks (BPNN) model and Random Forest (RF) model, producing the two CCF models, the Gipps-RF model and Gipps-BPNN model. The real vehicle trajectory data sets are applied to calibrate and validate the proposed models, and simulations are conducted to evaluate the model performances. The results display that the proposed CCF models can enhance safety level and robustness of the car-following control of automated vehicles. Both the two CCF models have better performance than the BPNN and RF car-following models in reducing congestion, stabilizing traffic, and avoiding crashes, especially the Gipps-BPNN model.

A Novel Classifier Exploiting Mobility Behaviors for Sybil Detection in Connected Vehicle Systems

A Sybil attacker is able to obtain more than one identities and disguise as multiple vehicles in order to interfere the normal operations of the connected vehicle system (CVS). In this paper, we propose a novel classifier to detect Sybil attackers according to their mobility behaviors. Specifically, three levels of Sybil attackers are first defined according to their attack abilities. Through analyzing the mobility behaviors of vehicles, a learning-based model is used in the central server (CS) to extract mobility features and distinguish Sybil attackers from benign vehicles. Three classification algorithms are tested and compared, i.e., the naive Bayes, decision tree, and support vector machine. Furthermore, location certificates issued by base stations are used to resist location forgery by attackers. Based on the location certificates, the CS is able to evaluate the credibilities of uploaded locations using the subjective logic theory. In addition, we develop an edge betweenness-based community detection algorithm to handle the collusion among multiple Sybil attackers. Simulations are conducted based on a real-world vehicle trajectory dataset, which indicate that the proposed scheme is effective to resist Sybil attackers in CVS.

Learning Traffic Flow Dynamics Using Random Fields

This paper presents a mesoscopic traffic flow model that explicitly describes the spatio-temporal evolution of the probability distributions of vehicle trajectories. The dynamics are represented by a sequence of factor graphs, which enable learning of traffic dynamics from limited Lagrangian measurements using an efficient message passing technique. The approach ensures that estimated speeds and traffic densities are non-negative with probability one. The estimation technique is tested using vehicle trajectory datasets generated using an independent microscopic traffic simulator and is shown to efficiently reproduce traffic conditions with probe vehicle penetration levels as little as 10\%. The proposed algorithm is also compared with state-of-the-art traffic state estimation techniques developed for the same purpose and it is shown that the proposed approach can outperform the state-of-the-art techniques in terms reconstruction accuracy.

Evaluation of the Gradient Boosting of Regression Trees Method on Estimating Car-Following Behavior

Car-following models, as the essential part of traffic microscopic simulations, have been utilized to analyze and estimate longitudinal drivers’ behavior for sixty years. The conventional car-following models use mathematical formulas to replicate human behavior in car-following phenomenon. The incapability of these approaches to capture the complex interactions between vehicles calls for deploying advanced learning frameworks to consider more detailed behavior of drivers. In this study, we apply the gradient boosting of regression tree (GBRT) algorithm to vehicle trajectory data sets, which have been collected through the Next Generation Simulation (NGSIM) program, to develop a new car-following model. First, the regularization parameters of the proposed method are tuned using cross-validation technique and sensitivity analysis. Second, prediction performance of the GBRT is compared to the world-famous Gazis-Herman-Rothery (GHR) model, when both models have been trained on the same data sets. The estimation results of the models on unseen records indicate the superiority of the GBRT algorithm in capturing the motion characteristics of two successive vehicles.

Anomaly Detection in Road Networks Using Sliding-Window Tensor Factorization

Anomaly detection in road networks is vital for traffic management and emergency response. However, existing approaches do not directly address multiple anomaly types. We propose a tensor-based spatio-temporal model for detecting multiple types of anomalies in road networks. First, we represent network traffic data as a 3rd-order tensor. Next, we acquire spatial and multi-scale temporal patterns of traffic variations via a novel, computationally efficient tensor factorization algorithm: sliding window tensor factorization. Then, from the factorization results, we can identify different anomaly types by measuring deviations from different spatial and temporal patterns. Finally, we discover path-level anomalies by formulating anomalous path inference as a linear program that solves for the best matched paths of anomalous links. We evaluate the proposed methods via both synthetic experiments and case studies based on a real-world vehicle trajectory dataset, demonstrating advantages of our approach over baselines.

Method for investigating intradriver heterogeneity using vehicle trajectory data: A Dynamic Time Warping approach

After first extending Newell’s car-following model to incorporate time-dependent parameters, this paper describes the Dynamic Time Warping (DTW) algorithm and its application for calibrating this microscopic simulation model by synthesizing driver trajectory data. Using the unique capabilities of the DTW algorithm, this paper attempts to examine driver heterogeneity in car-following behavior, as well as the driver’s heterogeneous situation-dependent behavior within a trip, based on the calibrated time-varying response times and critical jam spacing. The standard DTW algorithm is enhanced to address a number of estimation challenges in this specific application, and a numerical experiment is presented with vehicle trajectory data extracted from the Next Generation Simulation (NGSIM) project for demonstration purposes. The DTW algorithm is shown to be a reasonable method for processing large vehicle trajectory datasets, but requires significant data reduction to produce reasonable results when working with high resolution vehicle trajectory data. Additionally, singularities present an interesting match solution set to potentially help identify changing driver behavior; however, they must be avoided to reduce analysis complexity.

Review of Microscopic Lane-Changing Models and Future Research Opportunities

Driver behaviors, particularly lane-changing behaviors, have an important effect on the safety and throughput of the roadway-vehicle-based transportation system. Lane-changing models are a vital component of various microscopic traffic simulation tools, which are extensively used and playing an increasingly important role in Intelligent Transportation Systems studies. The authors conducted a detailed review and systematic comparison of existing microscopic lane-changing models that are related to roadway traffic simulation to provide a better understanding of respective properties, including strengths and weaknesses of the lane-changing models, and to identify potential for model improvement using existing and emerging data collection technologies. Many models have been developed in the last few decades to capture the uncertainty in lane change modeling; however, lane-changing behavior in the real world is very complex due to driver distraction (e.g., texting and cellphone or smartphone use) and environmental (e.g., pavement and lighting conditions) and geometric (e.g., horizontal and vertical curves) factors of the roadway, which have not been adequately considered in existing models. Therefore, large and detailed microscopic vehicle trajectory data sets are needed to develop new lane changing models that address these issues, and to calibrate and validate lane-changing models for representing the real world reliably. Possible measures to improve the accuracy and reliability of lane-changing models are also discussed in this paper.

Technical Note: Wavelet Transform-Based Algorithm for Reduction of Agricultural Vehicle Tracking Data

When handling a vehicle trajectory dataset composed of a large number of points on a digital map, all the collected data may not be required. In particular, when a trajectory is displayed on a computer screen, it is desirable to use the smallest dataset that can accurately describe the actual trajectory. The objective of this study is to develop an algorithm that reduces the amount of vehicle tracking data while retaining the original shape of the vehicle trajectory. The algorithm assigns a degree of importance in shape representation to each data point. The degree of importance is defined in terms of wavelet coefficients, and the algorithm determines the order by which data points are eliminated. To evaluate the performance of the algorithm, a 15,618-point dataset obtained from a GPS unit mounted on an agricultural vehicle was processed. Even when the dataset was reduced to 912 points, which is only 5.8% of the original dataset, the root mean square error was no more than 0.76 m. The computational time for data reduction on the selected computer was 8 ms. The proposed reduction algorithm was embedded in the map software used to display the vehicle trajectory.

Modeling the Lane-Changing Execution of Multiclass Vehicles under Heavy Traffic Conditions

In general, existing lane-changing behavior models focus on drivers' lane-changing decisions and neglect lane-changing execution. However, a lane-changing maneuver is likely to require several seconds for execution. Excluding lane-changing execution may have a significant effect on estimated traffic flow characteristics, particularly under heavy traffic conditions (level of service E). Acceleration and deceleration behaviors of heavy vehicle and passenger car drivers during the execution of a lane-changing maneuver are compared and contrasted. In addition, separate acceleration and deceleration models are developed for heavy vehicles and passenger cars during the lane change. The vehicle trajectory data set used in this research reflects heavy traffic conditions. Analysis of heavy vehicle drivers' lane-changing execution reveals that drivers maintained an almost constant speed during the maneuver; this suggests that they did not accelerate or decelerate to adjust their speeds according to the speed of surrounding traffic in the target lane. However, passenger car drivers do accelerate to adjust their speeds according to the speeds of the lead and lag vehicles in the target lane. The results highlight differences in the behavior of heavy vehicle and passenger car drivers during lane-changing execution.