No file available
To read the file of this research,
you can request a copy directly from the authors.
CitySim: A Drone-Based Vehicle Trajectory Dataset for Safety Oriented Research and Digital Twins
Preprints and early-stage research may not have been peer reviewed yet.
Abstract
The development of safety-oriented research ideas and applications requires fine-grained vehicle trajectory data that not only has high accuracy but also captures a substantial number of critical safety events. This paper introduces the CitySim Dataset, which was devised with a core objective of facilitating safety-based research and applications. CitySim has vehicle trajectories extracted from 1140-minutes of drone videos recorded at 12 different locations. It covers a variety of road geometries including freeway basic segments, weaving segments, expressway merge/diverge segments, signalized intersections, stop-controlled intersections, and intersections without sign/signal control. CitySim trajectories were generated through a five-step procedure which ensured the trajectory accuracy. Furthermore, the dataset provides vehicle rotated bounding box information which is demonstrated to improve safety evaluation. Compared to other video-based trajectory datasets, the CitySim Dataset has significantly more critical safety events with higher severity including cut-in, merge, and diverge events. In addition, CitySim facilitates research towards digital twin applications by providing relevant assets like the recording locations'3D base maps and signal timings. These features enable more comprehensive conditions for safety research and applications such as autonomous vehicle safety and location-based safety analysis. The dataset is available online at https://github.com/ozheng1993/UCF-SST-CitySim-Dataset.
... Vehicle bounding box feature description is shown in Figure 1. [23] CitySim is a publicly available drone-based vehicle trajectory dataset that contains detailed driving data, vehicle data, and supporting information for the study of driving trajectory and driving intention [23]. A sub-dataset Freeway-B with six lanes in two directions was used in this research. ...
... Vehicle bounding box feature description is shown in Figure 1. [23] CitySim is a publicly available drone-based vehicle trajectory dataset that contains detailed driving data, vehicle data, and supporting information for the study of driving trajectory and driving intention [23]. A sub-dataset Freeway-B with six lanes in two directions was used in this research. ...
Accurately detecting and predicting lane change (LC)processes can help autonomous vehicles better understand their surrounding environment, recognize potential safety hazards, and improve traffic safety. This paper focuses on LC processes and compares different machine learning methods' performance to recognize LC intention from high-dimensionality time series data. To validate the performance of the proposed models, a total number of 1023 vehicle trajectories is extracted from the CitySim dataset. For LC intention recognition issues, the results indicate that with ninety-eight percent of classification accuracy, ensemble methods reduce the impact of Type II and Type III classification errors. Without sacrificing recognition accuracy, the LightGBM demonstrates a sixfold improvement in model training efficiency than the XGBoost algorithm.
... The FWHA recommended threshold values for traffic conflict studies from 0 to 5 s for PET and 0 to 1.5 s for TTC [26]. In addition, a PET value of 2.5s is severely used as a threshold between serious and non-serious conflicts, while the threshold value of 1.0s is strongly correlated with severe crashes (i.e., crashes result in injuries or fatalities [15,16,31]. ...
The advent of technology has brought forth substantial changes in various industries, significantly transforming their operational capabilities and enhancing their productivity. A notable innovation shaping the future of many sectors is the use of Unmanned Aerial Vehicles (UAVs), commonly referred to as drones. Among various fields, the construction industry, and in particular the oil & gas sector, has significantly benefited from integrating drone technology, especially in project construction monitoring [1].
Overview of Project Construction Monitoring in the Oil & Gas Industry
In the oil & gas industry, project construction monitoring plays a crucial role, ensuring that projects progress as planned and any issues are promptly identified and addressed. These monitoring processes traditionally involve numerous on-site workers, heavy machinery, and extensive time and resources [2]. However, the industry's dynamic environment, characterized by sprawling sites, hazardous conditions, and complex projects, necessitates a more efficient approach to construction monitoring. In response to these challenges, the industry is increasingly turning to drone technology [3].
Importance of Efficient Project Construction Monitoring
Efficient construction monitoring is pivotal for the timely completion of projects, maintaining budget constraints, and ensuring safety standards. Any delays or cost overruns in large-scale oil & gas projects can have substantial financial and operational implications. Moreover, efficient monitoring can help identify potential risks and issues early, allowing for preventative measures to be implemented, reducing the likelihood of accidents and promoting a safer working environment [4].
... Experimental data were obtained from the publicly available CitySim dataset (46). The CitySim dataset is a collection of vehicle trajectory data obtained from drones. ...
Accurate Vehicle Trajectory Prediction is critical for automated vehicles and advanced driver assistance systems. Vehicle trajectory prediction consists of two essential tasks, i.e., longitudinal position prediction and lateral position prediction. There is a significant correlation between driving intentions and vehicle motion. In existing work, the three tasks are often conducted separately without considering the relationships between the longitudinal position, lateral position, and driving intention. In this paper, we propose a novel Temporal Multi-Gate Mixture-of-Experts (TMMOE) model for simultaneously predicting the vehicle trajectory and driving intention. The proposed model consists of three layers: a shared layer, an expert layer, and a fully connected layer. In the model, the shared layer utilizes Temporal Convolutional Networks (TCN) to extract temporal features. Then the expert layer is built to identify different information according to the three tasks. Moreover, the fully connected layer is used to integrate and export prediction results. To achieve better performance, uncertainty algorithm is used to construct the multi-task loss function. Finally, the publicly available CitySim dataset validates the TMMOE model, demonstrating superior performance compared to the LSTM model, achieving the highest classification and regression results. Keywords: Vehicle trajectory prediction, driving intentions Classification, Multi-task
The appropriate threshold selection is pivotal in near-crashes detections. Nevertheless, most cutting-edge warning systems utilize thresholds that are directly derived from general traffic scenarios. There is no evidence support that thresholds acquired from general scenarios can be applied to specific scenarios. Thus, the purpose of this study is to investigate whether thresholds of three surrogate measures of safety (SMoS) differ between the general clear scenario and the special rainy scenario. A total of 1048 rear-end events (724 in clear conditions and 324 in light rain conditions) were obtained from the CitySim dataset. Two threshold selection methods were used to recommend appropriate thresholds. The findings indicated that only MTTC had a statistical difference in two weather scenarios, hence, a threshold was recommended under each of two weather conditions. Furthermore, the MTTC threshold for rainy conditions was three times higher than for clear conditions. Contrarily, the threshold was statistically insignificant for the SPDRF and DRAC under two conditions, so only one integration threshold per SMoS was recommended, even in two scenarios. These findings could help safety-system designers set more targeted and appropriate thresholds when encounter specific characteristics (e.g., weather) for active safety systems or safety warning infrastructures in future smart cities.
As the pedestrian’s rapid progress, pedestrian safety has recently assumed greater importance in the research population activities. To assess road user interactions at an unsignalized junctions with heterogeneous traffic complexity, innovative trajectory-based data was used and to make urban intersections safer for road users, the proposed severity levels will be used to test and evaluate numerous infrastructures and control upgrades. Study authors suggests an advanced pattern-based technique to characterize pedestrian-vehicle interactions based on road user behaviors. Surrogate safety measures (SSM) can be estimated more accurately with this study than with the regular grid-based analysis. In order to evaluate SSM (Speed, Time to Collision (TTC), Gap Time of the pedestrian and vehicle interactions) at an unsignalized crossing, Safe Distance and Stopping sight distance with trajectory data are to be estimated. Support Vector Machine (SVM) is used to classify severity grades based on specified indicators generated at an unsignalized junction in India. From the analysis, Severity at the intersection found as 8.12, 8.91sec respectively, Average gap time, stopping sight distance also calculated. Plots between Time to Collision and Gap Time for pedestrian passing first (PPF), vehicle passing first (VPF) are compared by developing a linear regression model and R ² =0.684, 0.656 developed with some independent parameters respectively. Concluded that TTC for PPF is higher than VPF and by considering all the evaluated values this research had improved the methods for analysing and improving the safety of uncontrolled intersections.
Driving behavior modeling is of great importance for designing safe, smart, and personalized autonomous driving systems. In this paper, an internal reward function-based driving model that emulates the human's decision-making mechanism is utilized. To infer the reward function parameters from naturalistic human driving data, we propose a structural assumption about human driving behavior that focuses on discrete latent driving intentions. It converts the continuous behavior modeling problem to a discrete setting and thus makes maximum entropy inverse reinforcement learning (IRL) tractable to learn reward functions. Specifically, a polynomial trajectory sampler is adopted to generate candidate trajectories considering high-level intentions and approximate the partition function in the maximum entropy IRL framework. An environment model considering interactive behaviors among the ego and surrounding vehicles is built to better estimate the generated trajectories. The proposed method is applied to learn personalized reward functions for individual human drivers from the NGSIM highway driving dataset. The qualitative results demonstrate that the learned reward functions are able to explicitly express the preferences of different drivers and interpret their decisions. The quantitative results reveal that the learned reward functions are robust, which is manifested by only a marginal decline in proximity to the human driving trajectories when applying the reward function in the testing conditions. For the testing performance, the personalized modeling method outperforms the general modeling approach, significantly reducing the modeling errors in human likeness (a custom metric to gauge accuracy), and these two methods deliver better results compared to other baseline methods. Moreover, it is found that predicting the response actions of surrounding vehicles and incorporating their potential decelerations caused by the ego vehicle are critical in estimating the generated trajectories, and the accuracy of personalized planning using the learned reward functions relies on the accuracy of the forecasting model.
The behavior decision-making algorithm plays an important role in ensuring the safe driving of autonomous vehicles. However, existing behavior decision-making methods lack the capability to cope with future motion uncertainty in traffic, because the historical state of vehicles are not considered. This paper proposes a novel driving behavior decision-making method EnMFO-ImGRU based on Gated Recurrent Unit (GRU) and Moth-Flame Optimization algorithm (MFO). Four improvements are proposed in EnMFO-ImGRU. First, to consider the driving information of the vehicles on the road, ImGRU is designed based on a double-layer GRU. Second, to promote decisions accuracy, Support Vector Machine (SVM), which has good performance in classification problems, replaces the softmax classifier to train the output of the ImGRU. Third, to promote the classification capability of SVM, MFO is introduced to optimize the key parameters that affect the performance of SVM. Finally, to promote the optimization capability of MFO, we propose the Enhanced Moth-Flame Optimization algorithm (EnMFO). A new position updating method is proposed in EnMFO. The experimental results on the NGSIM dataset show that EnMFO-ImGRU brings higher accuracy than existing methods for the behavior decision-making results of autonomous vehicles.
As an effective method, Traffic Conflict Technology (TCT) is widely applied to estimate the safety level of some risky areas, especially for the merging areas in the urban roads. Most of researchers prefer to just exploit the promising safety assessment models using realistic traffic data to predict the numbers of conflicts. There are a few types of research focusing on how to predict traffic conflict assessment indexes precisely in merging areas. Despite some related studies have realized this critical dilemma, significant lane-change characteristics are usually ignored and it is worthwhile devoting much effort to this. Hence, a modified Post Encroachment Time (PET) model is proposed in this study, to figure out lane-change characteristics as well as accurately forecast traffic safety of the merging area. Unlike other conventional methods, such as Time to Collision (TTC) or PET models, the proposed model not only fully takes the lane-change characteristics of merging vehicles into consideration, but also it explores the safety requirements in the process of lane changing in details. Moreover, the calculation formula of the modified PET model is gradually deduced by the trajectory of merging behavior and velocity formula. Besides, for the sake of pursuing a high validity, this paper exceptionally adopts two crucial compositions of PET, and eventually gives a unified calculation formula. In order to determine an appropriate threshold, traffic conflict data collected from Guangyuan Road in Guangzhou are analyzed. The results clearly prove that PET < 0.7 means a serious conflict, 0.7. PET <1.31 means a general conflict, 1.31. PET < 2.25 means a slight conflict, and PET. 2.25 means a potential conflict. Finally, 50 groups of PET data and a comparative experiment are collected to demonstrate the effectiveness and reliability of the modified PET model.
In this paper, we review trajectory data-based traffic flow studies that have been conducted over the last 15 years. Our purpose is to provide a roadmap for readers who have an interest in the latest developments of traffic flow theory that have been stimulated by the availability of trajectory data. We first highlight the critical role of trajectory data (especially the next generation simulation (NGSIM) trajectory dataset) in the recent history of traffic flow studies. Then, we summarize new traffic phenomena/models at the microscopic/mesoscopic/macroscopic levels and provide a unified view of these achievements perceived from different directions of traffic flow studies. Finally, we discuss some future research directions.
To solve the unacceptable issue caused by the inconsistency of lane‐changing behaviour between autonomous vehicles and actual drivers. A lane‐changing behaviour decision‐making model based on the Gauss mixture hidden Markov model (GM‐HMM) is proposed according to the characteristic of a driver's lane changing behaviour. The proposed model is tested and verified based on the database of Next‐Generation Simulation (NGSIM). The results show that the GM‐HMM is 95.4% similar to the real driver's behaviour. To further verify the proposed model, the proposed algorithm is compared with some machine learning techniques from literature in different test scenarios. The comparison and analysis indicate that the GM‐HMM method can more accurately simulate the real driver's lane‐change behaviour, thus improving the trust of the passengers and other vehicles around autonomous vehicles.
The great success that deep models have achieved in the past is mainly owed to large amounts of labeled training data. However, the acquisition of labeled data for new tasks aside from existing benchmarks is both challenging and costly. Active learning can make the process of labeling new data more efficient by selecting unlabeled samples which, when labeled, are expected to improve the model the most. In this paper, we combine a novel method of active learning for object detection with an incremental learning scheme (Käding et al., 2016b) to enable continuous exploration of new unlabeled datasets. We propose a set of uncertainty-based active learning metrics suitable for most object detectors. Furthermore, we present an approach to leverage class imbalances during sample selection. All methods are evaluated systematically in a continuous exploration context on the PASCAL VOC 2012 dataset (Everingham et al., 2010).
The majority of the prevailing collision detection and avoidance systems render evasive maneuvers for averting front and rear vehicle collisions. This paper introduces a mid vehicle collision detection and avoidance system with a constraint-free condition that produces mid vehicle maneuvers, particularly when jammed between the front and rear vehicles. Accordingly, three novel path estimation models based on crisp, fuzzy and fuzzy regression logic are blended to formulate a mid (host) vehicle collision detection and avoidance system. At the onset, the crisp model fits an offset-based curvilinear path for the mid vehicle to avoid edge collisions. The realized crisp model is later fuzzified to address more antecedents and accordingly deliver consequents for enhancing path estimation. Finally, the fuzzified model is regressed to obtain a good fitness for real-road conditions. The fused fuzzy regression renders an approximate version of the actual road strategy to obtain collision-free trajectories. This concept is later intrinsically extended to parallel parking in reverse direction. Simulation studies using coefficient of determination \((R^{2} )\) and mean square error on the field, real (next generation simulation-NGSIM) dataset reveals the goodness and the path closeness of the proposed system rendered by the novelly blended models tuned at each observation using the optimum h-uncertain factor. Also, relative mean square error analysis with state-of-art CDAS reveals the superiority of MCDAS, thereby making it amicable for real-road scenarios.
Highlights the shortcoming of using aggregate data as a whole with the assumption of similar lane-use-pattern over lanes, traffic variations are attentively investigated to explore traffic dynamics distribution both in lane-utility in efficiency and safety performance lane-by-lane. In this paper, above two dimensions are examined for each of five mainline lanes on Highway 101 in Angeles using Next Generation Simulation (NGSIM) data. The lane-specific data is statistically processed to reflect lane utility in lane flow parameters along the location and identify Lane Risk Coefficient by verifying degree of consistence vehicles operation with 'Assured Clear Distance Ahead' criterion. It appears completely new that the most-middle lanes are similar utilized under congested conditions compared to two sides of the lanes, as well as the quite close values of LRC to reveals an uniformly safety level between Lane 2 and Lane 3. The findings of this study are intended to be relevant with the calibration of microscopic simulation models and suggest implementing a proper variable speed towar d under-utilized shoulder Lane 5 but with worst safety performance.
Conflict analysis using surrogate safety measures (SSMs) has become an efficient approach to investigate safety issues. The state-of-the-art studies largely resort to video images taken from high buildings. However, it suffers from heavy labor work, high cost of maintenance, and even security restrictions. Data collection and processing remains a common challenge to traffic conflict analysis. Unmanned Aerial Systems (UASs) or Unmanned Aerial Vehicles (UAVs), known for easy maneuvering, outstanding flexibility, and low costs, are considered to be a novel aerial sensor. By taking full advantage of the bird's eye view offered by UAV, this study, as a pioneer work, applied UAV videos for surrogate safety analysis of pedestrian-vehicle conflicts at one urban intersection in Beijing, China. Aerial video sequences for a period of one hour were analyzed. The detection and tracking systems for vehicle and pedestrian trajectory data extraction were developed, respectively. Two SSMs, that is, Postencroachment Time (PET) and Relative Time to Collision (RTTC), were employed to represent how spatially and temporally close the pedestrian-vehicle conflict is to a collision. The results of analysis showed a high exposure of pedestrians to traffic conflict both inside and outside the crosswalk and relatively risking behavior of right-turn vehicles around the corner. The findings demonstrate that UAV can support intersection safety analysis in an accurate and cost-effective way.
The main objective of this study was to identify near crashes in vehicle trajectory data with interdriver heterogeneity and situation dependency considered. Several efforts have been made to evaluate the effects of near-crash events on safety with the use of naturalistic driving data, driving simulators, and test tracks. However, these efforts have faced some challenges because the observations reflected only the equipped vehicles. The development of connected vehicle technology provided the essential data to study high-risk maneuvers in the entire traffic stream. In this study, two near-crash detection algorithms were proposed. One algorithm had its basis in fixed thresholds, while the other considered interdriver heterogeneity and estimates driver-specific thresholds. The models were tested against two NGSIM trajectory data sets. Initial results showed that consideration of driver preferences resulted in more realistic identification of near crashes than otherwise.
Digital inpainting provides a means for reconstruction of small damaged portions of an image. Although the inpainting basics are straightforward, most inpainting techniques published in the literature are complex to understand and implement. We present here a new algorithm for digital inpainting based on the fast marching method for level set applications. Our algorithm is very simple to implement, fast, and produces nearly identical results to more complex, and usually slower, known methods. Source code is available online.
Capturing variability within flow is an important task for traffic flow models. The linearity of the congested part of the fundamental diagram induces a linear speed-spacing relationship at an individual level, characterized by two parameters. This study assumes that most intervehicle variability can be accounted for by estimating these two parameters for each vehicle. Two methods are presented to quantify individual linear speed-spacing relationships. The first method is based on data: it estimates the speed-spacing relationship by fitting the experimental speed-spacing scatter plot with a straight line. The second method is based on simulation: it computes the optimum parameters so that the simulated trajectories obtained by Newell's car-following algorithm reproduce as closely as possible the experimental vehicle's trajectories. Both proposed methods are implemented on the Next Generation Simulation trajectory data set recorded on I-80. The individual parameters for the speed-spacing relationship are quantified, and their distributions are specified. The need to distinguish driver behavior on a lane-by-lane basis is discussed. The results tend to prove that taking into account individual variability between drivers can improve the accuracy of simulated trajectories.
Real-time segmentation of moving regions in image sequences is a fundamental step in many vision systems including automated visual surveillance, human-machine interface, and very low-bandwidth telecommunications. A typical method is background subtraction. Many background models have been introduced to deal with different problems. One of the successful solutions to these problems is to use a multi-colour background model per pixel proposed by Grimson et al [1,2,3]. However, the method suffers from slow learning at the beginning, especially in busy environments. In addition, it can not distinguish between moving shadows and moving objects. This paper presents a method which improves this adaptive background mixture model. By reinvestigating the update equations, we utilise different equations at different phases. This allows our system learn faster and more accurately as well as adapt effectively to changing environments. A shadow detection scheme is also introduced in this paper. It is based on a computational colour space that makes use of our background model. A comparison has been made between the two algorithms. The results show the speed of learning and the accuracy of the model using our update algorithm over the Grimson et al's tracker. When incorporate with the shadow detection, our method results in far better segmentation than that of Grimson et al.
Traffic conflict can be identified by the presence of evasive actions or the amount of temporal (spatial) proximity measures like time-to-collision (TTC). However, it is not enough to use only one kind of measures in some scenarios and it is hard to set a threshold for those measures. This paper proposed a method to identify traffic conflict by learning the representation of TTC and driver maneuver profiles with deep unsupervised learning and clustering the representations into traffic conflict and non-conflict clusters. We first trained a transformer encoder to encode sequences of surrogate safety measures into some latent space with unsupervised pre-training. Second, we identified informative clusters in the latent space by calculating the statistic summaries and visualizing trajectory pairs of each cluster. Some clusters are interpreted as traffic conflict clusters because they have small TTC, large deceleration rate and intertwining trajectories and they can be further interpreted as rear-end or angle conflicts. Moreover, the identified traffic conflicts contain critical conditions from the two vehicles in an interaction and one vehicle perceives them as abnormal and takes evasive action to avoid crashes.
High-granularity vehicle trajectory data can help researchers develop traffic simulation models, understand traffic flow characteristics, and thus propose insightful strategies for road traffic management. This paper proposes a novel vehicle trajectory extraction method that can extract high-granularity vehicle trajectories from aerial videos. The proposed method includes video calibration, vehicle detection and tracking, lane marking identification, and vehicle motion characteristics calculation. In particular, the authors propose a Monte-Carlo-based lane marking identification approach to identify each vehicle's lane. This is a challenging problem for vehicle trajectory extraction, especially when the aerial videos are taken from a high altitude. The authors applied the proposed method to extract vehicle trajectories from several high-resolution aerial videos recorded from helicopters. The extracted dataset is named by the High-Granularity Highway Simulation (HIGH-SIM) vehicle trajectory dataset. To demonstrate the effectiveness of the proposed method and understand the quality of the HIGH-SIM dataset, we compared the HIGH-SIM dataset with a well-known dataset, the NGSIM US-101 dataset, regarding the accuracy and consistency aspects. The comparison results showed that the HIGH-SIM dataset has more reasonable speed and acceleration distributions than the NGSIM US-101 dataset. Also, the internal and platoon consistencies of the HIGH-SIM dataset give lower errors compared to the NGSIM US-101 dataset. To benefit future research, the authors have published the HIGH-SIM dataset online for public use.
To fully harvest the benefits of vehicular automation and connectivity in the mixed traffic, a cooperative longitudinal control strategy named Cooperative Adaptive Cruise Control with Unconnected vehicle in the loop (CACCu) has been proposed. When encountering an unconnected preceding vehicle, CACCu enables a Connected and Automated Vehicle (CAV) to benefit from communicating with a connected vehicle further ahead, rather than completely falling back to Adaptive Cruise Control (ACC). To validate the feasibility of CACCu, this study developed and tested a CACCu system with real vehicles in the field. A speed-command-based CACCu controller is designed and parameterized for optimizing the anticipated string stability. The experiment was conducted with two automated vehicles equipped with Mobileye sensors and Wi-Fi modules. The car-following performance of CACCu, in comparison with ACC and human driving (as the ego vehicle), was evaluated in the real-traffic scenarios constructed using NGSIM vehicle trajectory data. Over the 6 test runs for each control method, it was found that CACCu reduced 10.8% acceleration Root Mean Square (RMS), 60.8% spacing error RMS and 6.2% fuel consumption from ACC’s, indicating advantages of CACCu in control accuracy, ride comfort and energy efficiency. Compared with human driving, CACCu also reduced 17.6% acceleration and 13.4% fuel consumption. More importantly, the CACCu was able to efficiently avoid the traffic disturbance amplifications that frequently happened to ACC and human driving, which means the string stability has been significantly improved by the CACCu.
Motion prediction for the leading vehicle is a critical task for connected autonomous vehicles. It provides a method to model the leading-following vehicle behavior and analysis their interactions. In this study, a joint time-series modeling approach for leading vehicle trajectory prediction considering different driving styles is proposed. The proposed method enables a precise and personalized trajectory prediction for the leading vehicle based on limited inter-vehicle communication signals, such as the vehicle speed and acceleration of the front vehicles. Three different driving styles are first recognized based on an unsupervised clustering algorithm, namely, Gaussian Mixture Model (GMM). The GMM generates a specific driving style for each vehicle based on the speed, acceleration, jerk, time, and space headway features of the leading vehicle. The feature importance of driving style recognition is also evaluated based on the Maximal Information Coefficient (MIC) algorithm. Then, a personalized joint time series modeling (JTSM) method based on the Long Short-Term Memory (LSTM) Recurrent Neural Network model (RNN) is proposed to predict the front vehicle trajectories. The JTSM contains a common LSTM layer and different fully connected regression layers for different driving styles. The proposed method is tested with the Next Generation Simulation (NGSIM) data on the US101, and I-80 highway dataset. The JTSM is tested for making predictions one to five seconds ahead. Results indicate that the proposed personalized JTSM approach shows a significant advantage over the baseline algorithms.
Traditional statistical crash prediction models oftentimes suffer from poor data quality and require large amount of historical data. In this paper, we propose a crash prediction method based on a bivariate extreme value theory (EVT) framework, considering both drivers' perception-reaction failure and failure to proper evasive actions. An unmanned aerial vehicle (UAV) was utilized to collect videos of ten intersections in Fengxian, China, at representative time periods. High-resolution vehicle trajectory data were extracted by a Kanade-Lucas-Tomasi (KLT) technique, based on four detailed metrics were derived including Time-to-accident (TA), Post-encroachment Time (PET), minimum Time-to-collision (mTTC), and Maximum Deceleration Rate (MaxD). TA was expected to capture the chance of perception-reaction failure, while other three metrics were used to measure the probability of failure to proper evasive actions. Univariate EVT models were applied to obtain marginal crash probability based on each metric. Bivariate EVT models were developed to obtain joint crash probability based on three pairs: TA and mTTC, TA and PET, and TA and MaxD. Thus, union crash probability within observation periods can be derived and the annual crash frequency of each intersection was predicted. The predictions were compared to actual annual crash frequencies, using multiple tests. The findings are three-folds: 1. The best conflict metrics for angle and rear-end crash predictions were different; 2. Bivariate EVT models were found to be superior to univariate models, regarding both angle and rear-end crash predictions; 3. TA appeared to be an important conflict metric that should be considered in a bivariate EVT model framework. In general, the proposed method can be considered as a promising tool for safety evaluation, when crash data are limited.
This study analyzes rear-end collision risk of cars and heavy vehicles on freeways using a surrogate safety measure. The crash potential index (CPI) was modified to reflect driver's reaction time and estimated by types of lead and following vehicles (car or heavy vehicle). CPIs were estimated using the individual vehicle trajectory data from a segment of the US-101 freeway in Los Angeles, U.S.A. It was found that the CPI was generally higher for the following heavy vehicle than the following car due to heavy vehicle's lower braking capability. This study also validates the CPI using the simulated traffic data which replicate the observed traffic conditions a few minutes before the crash time upstream and downstream of the crash locations. The observed data were obtained from crash records and loop detectors on a section of the Gardiner Expressway in Toronto, Canada. The result shows that the values of CPI were consistently higher during the traffic conditions immediately before the crash time (crash case) than the normal traffic conditions (non-crash case). This demonstrates that the CPI can be used to capture rear-end collision risk during car-following maneuver on freeways. The result also shows that rear-end collision risk is lower for heavy vehicles than cars in the crash case due to their shorter reaction time and lower speed when spacing is shorter. Thus, it is important to reflect the differences in driver behavior and vehicle performance characteristics between cars and heavy vehicles in estimating surrogate safety measures. Lastly, it was found that the CPI-based crash prediction model can correctly identify the crash and non-crash cases at higher accuracy than the other crash prediction models based on detectors.
Accurate positioning is a key factor for enabling innovative applications to properly perform their tasks in various areas including: Intelligent Transportation Systems (ITS) and Vehicular Ad Hoc Network (VANET). Vehicle positioning accuracy depends heavily on positioning techniques and the measurements condition in its surroundings. Several approaches which can be used for improving vehicle positioning accuracy have been reported in literature. Although some positioning techniques have achieved high accuracy in a controlled environment, they suffer from dynamic measurement noises in real environments leading to low accuracy and integrity for some VANET applications. To solve this issue, some existing positioning approaches assume the availability of prior knowledge concerning measurement noises, which is not practical for VANET. The aim of this paper is to propose an algorithm for improving accuracy and integrity of positioning information under dynamic and unstable measurement conditions. To do this, a positioning algorithm has been designed based on the Innovation-based Adaptive Estimation Kalman Filter (IAE_KF) by integrating the positioning measurements with vehicle kinematic information. Following that, the IAE_KF algorithm is enhanced in terms of positioning accuracy and integrity (EIAE_KF) in order to meet VANET applications requirements. This enhancement involves two stages which are: a switching strategy between dead reckoning and the Kalman Filter based on the innovation property of the optimal filter; and the estimation of the actual noise covariance based on the Yule–Walker method. An online error estimation model is then proposed to estimate the uncertainty of the EIAE_KF algorithm to enhance the integrity of the position information. Next Generation Simulation dataset (NGSIM) which contains real world vehicle trajectories is used as ground truth for the evaluation and testing procedure. The effectiveness of the proposed algorithm is demonstrated through a comprehensive simulation study. The results show that the EIAE_KF algorithm is more effective than existing solutions in terms of enhancing positioning information accuracy and integrity so as to meet VANET applications requirements.
Trajectories drawn in a common reference system by all the vehicles on a road are the ultimate
empirical data to investigate traffic dynamics. The vast amount of such data made
freely available by the Next Generation SIMulation (NGSIM) program is therefore opening
up new horizons in studying traffic flow theory. Yet the quality of trajectory data and its
impact on the reliability of related studies was a vastly underestimated problem in the traffic
literature even before the availability of NGSIM data. The absence of established methods
to assess data accuracy and even of a common understanding of the problem makes it
hard to speak of reproducibility of experiments and objective comparison of results, in
particular in a research field where the complexity of human behaviour is an intrinsic
challenge to the scientific method. Therefore this paper intends to design quantitative
methods to inspect trajectory data. To this aim first the structure of the error on point measurements
and its propagation on the space travelled are investigated. Analytical evidence
of the bias propagated in the vehicle trajectory functions and a related consistency requirement
are given. Literature on estimation/filtering techniques is then reviewed in light of
this requirement and a number of error statistics suitable to inspect trajectory data are proposed.
The designed methodology, involving jerk analysis, consistency analysis and spectral
analysis, is then applied to the complete set of NGSIM databases.
Modeling Human Driving Behavior through Generative Adversarial Imitation Learning
- R Bhattacharyya
- B Wulfe
- D Phillips
- A Kuefler
- J Morton
- R Senanayake
- M Kochenderfer
Bhattacharyya, R., B. Wulfe, D. Phillips, A. Kuefler, J. Morton, R. Senanayake, and M.
Kochenderfer. Modeling Human Driving Behavior through Generative Adversarial Imitation
Learning. http://arxiv.org/abs/2006.06412. Accessed Aug. 1, 2022.
Next Generation Simulation (NGSIM) Vehicle Trajectories and Supporting Data
U.S. Department of Transportation Federal Highway Administration. (2016). Next Generation
Simulation (NGSIM) Vehicle Trajectories and Supporting Data. http://doi.org/10.21949/1504477.
Driver Car-Following Behavior: From Discrete Event Process to Continuous Set of Episodes
- S H Hamdar
- H S Mahmassani
Hamdar, S. H., and H. S. Mahmassani. Driver Car-Following Behavior: From Discrete Event
Process to Continuous Set of Episodes. 2008.
Enhanced Index of Risk Assessment of Lane-Change on Expressway Weaving Segments(under Review)
- J Zhang
- M A Jaeyoung Lee
- O Aty
- G Zheng
- Xiao
Zhang, J., Jaeyoung Lee, M. A. Aty, O. Zheng, and G. Xiao. Enhanced Index of Risk Assessment of
Lane-Change on Expressway Weaving Segments(under Review).
Modelling the Relationship Between Post Enchroachment Time and Signal Timings Using UAV Video Data(under Review)
- Z Islam
- M Abdel-Aty
- A Goswamy
- A Abdelraouf
- O Zheng
Islam, Z., M. Abdel-Aty, A. Goswamy, A. Abdelraouf, and O. Zheng. Modelling the Relationship
Between Post Enchroachment Time and Signal Timings Using UAV Video Data(under Review).
Trajectory Prediction for Vehicle Conflict Identification at Intersections Using Sequence-to-Sequence Recurrent Neural Networks(under Review)
- Abdelraouf Amr
- Mohamed Abdel-Aty
- Z Wang
- O Zheng
Abdelraouf Amr, Mohamed Abdel-Aty, Z. Wang, and O. Zheng. Trajectory Prediction for Vehicle
Conflict Identification at Intersections Using Sequence-to-Sequence Recurrent Neural
Networks(under Review).
A Review of the Applications of Computer Vision Technology in Surrogate-SafetyMeasures-Based Traffic Safety Analysis
- M Abdel-Aty
- Z Wang
- M R R Jahin
- O Zheng
- A Abdelraouf
Abdel-Aty, M., Z. Wang, M. R. R. Jahin, O. Zheng, and A. Abdelraouf. A Review of the
Applications of Computer Vision Technology in Surrogate-SafetyMeasures-Based Traffic Safety
Analysis(under Review).