Search

Computer vision has advanced research methodologies, enhancing system services across various fields. It is a core component in traffic monitoring systems for improving road safety; however, these monitoring systems don't preserve the privacy of pedestrians who appear in the videos, potentially revealing their identities. Addressing this issue, our paper introduces Video-to-Text Pedestrian Monitoring (VTPM), which monitors pedestrian movements at intersections and generates real-time textual reports, including traffic signal and weather information. VTPM uses computer vision models for pedestrian detection and tracking, achieving a latency of 0.05 seconds per video frame. Additionally, it detects crossing violations with 90.2% accuracy by incorporating traffic signal data. The proposed framework is equipped with Phi-3 mini-4k to generate real-time textual reports of pedestrian activity while stating safety concerns like crossing violations, conflicts, and the impact of weather on their behavior with latency of 0.33 seconds. To enhance comprehensive analysis of the generated textual reports, Phi-3 medium is fine-tuned for historical analysis of these generated textual reports. This fine-tuning enables more reliable analysis about the pedestrian safety at intersections, effectively detecting patterns and safety critical events. The proposed VTPM offers a more efficient alternative to video footage by using textual reports reducing memory usage, saving up to 253 million percent, eliminating privacy issues, and enabling comprehensive interactive historical analysis.

This paper introduces our solution for Track 2 in AI City Challenge 2024. The task aims to solve traffic safety description and analysis with the dataset of Woven Traffic Safety (WTS), a real-world Pedestrian-Centric Traffic Video Dataset for Fine-grained Spatial-Temporal Understanding. Our solution mainly focuses on the following points: 1) To solve dense video captioning, we leverage the framework of dense video captioning with parallel decoding (PDVC) to model visual-language sequences and generate dense caption by chapters for video. 2) Our work leverages CLIP to extract visual features to more efficiently perform cross-modality training between visual and textual representations. 3) We conduct domain-specific model adaptation to mitigate domain shift problem that poses recognition challenge in video understanding. 4) Moreover, we leverage BDD-5K captioned videos to conduct knowledge transfer for better understanding WTS videos and more accurate captioning. Our solution has yielded on the test set, achieving 6th place in the competition. The open source code will be available at https://github.com/UCF-SST-Lab/AICity2024CVPRW

Previous studies in predicting crash risk primarily associated the number or likelihood of crashes on a road segment with traffic parameters or geometric characteristics of the segment, usually neglecting the impact of vehicles' continuous movement and interactions with nearby vehicles. Advancements in communication technologies have empowered driving information collected from surrounding vehicles, enabling the study of group-based crash risks. Based on high-resolution vehicle trajectory data, this research focused on vehicle groups as the subject of analysis and explored risk formation and propagation mechanisms considering features of vehicle groups and road segments. Several key factors contributing to crash risks were identified, including past high-risk vehicle-group states, complex vehicle behaviors, high percentage of large vehicles, frequent lane changes within a vehicle group, and specific road geometries. A multinomial logistic regression model was developed to analyze the spatial risk propagation patterns, which were classified based on the trend of high-risk occurrences within vehicle groups. The results indicated that extended periods of high-risk states, increase in vehicle-group size, and frequent lane changes are associated with adverse risk propagation patterns. Conversely, smoother traffic flow and high initial crash risk values are linked to risk dissipation. Furthermore, the study conducted sensitivity analysis on different types of classifiers, prediction time intervalsss and adaptive TTC thresholds. The highest AUC value for vehicle-group risk prediction surpassed 0.93. The findings provide valuable insights to researchers and practitioners in understanding and prediction of vehicle-group safety, ultimately improving active traffic safety management and operations of Connected and Autonomous Vehicles., Comment: 14 pages, 8 figures

The EPC GEN 2 communication protocol for Ultra-high frequency Radio Frequency Identification (RFID) has offered a promising avenue for advancing the intelligence of transportation infrastructure. With the capability of linking vehicles to RFID readers to crowdsource information from RFID tags on road infrastructures, the RF-enhanced road infrastructure (REI) can potentially transform data acquisition for urban transportation. Despite its potential, the broader adoption of RFID technologies in building intelligent roads has been limited by a deficiency in understanding how the GEN 2 protocol impacts system performance under different transportation settings. This paper fills this knowledge gap by presenting the system architecture and detailing the design challenges associated with REI. Comprehensive real-world experiments are conducted to assess REI's effectiveness across various urban contexts. The results yield crucial insights into the optimal design of on-vehicle RFID readers and on-road RFID tags, considering the constraints imposed by vehicle dynamics, road geometries, and tag placements. With the optimized designs of encoding schemes for reader-tag communication and on-vehicle antennas, REI is able to fulfill the requirements of traffic sign inventory management and environmental monitoring while falling short of catering to the demand for high-speed navigation. In particular, the Miller 2 encoding scheme strikes the best balance between reading performance (e.g., throughput) and noise tolerance for the multipath effect. Additionally, we show that the on-vehicle antenna should be oriented to maximize the available time for reading on-road tags, although it may reduce the received power by the tags in the forward link.

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

Large Language Models (LLMs) have shown remarkable effectiveness in various general-domain natural language processing (NLP) tasks. However, their performance in transportation safety domain tasks has been suboptimal, primarily attributed to the requirement for specialized transportation safety expertise in generating accurate responses [1]. To address this challenge, we introduce TrafficSafetyGPT, a novel LLAMA-based model, which has undergone supervised fine-tuning using TrafficSafety-2K dataset which has human labels from government produced guiding books and ChatGPT-generated instruction-output pairs. Our proposed TrafficSafetyGPT model and TrafficSafety-2K train dataset are accessible at https://github.com/ozheng1993/TrafficSafetyGPT.

The real-time crash likelihood prediction model is an essential component of the proactive traffic safety management system. Over the years, numerous studies have attempted to construct a crash likelihood prediction model in order to enhance traffic safety, but mostly on freeways. In the majority of the existing studies, researchers have primarily employed a deep learning-based framework to identify crash potential. Lately, Transformer has emerged as a potential deep neural network that fundamentally operates through attention-based mechanisms. Transformer has several functional benefits over extant deep learning models such as LSTM, CNN, etc. Firstly, Transformer can readily handle long-term dependencies in a data sequence. Secondly, Transformers can parallelly process all elements in a data sequence during training. Finally, a Transformer does not have the vanishing gradient issue. Realizing the immense possibility of Transformers, this paper proposes inTersection-Transformer (inTformer), a time-embedded attention-based Transformer model that can effectively predict intersection crash likelihood in real-time. The proposed model was evaluated using connected vehicle data extracted from Signal Analytics Platform. Acknowledging the complex traffic operation mechanism at intersection, this study developed zone-specific models by dividing the intersection region into two distinct zones: within-intersection and approach zone. The best inTformer models in 'within-intersection,' and 'approach' zone achieved a sensitivity of 73%, and 70%, respectively. The zone-level models were also compared to earlier studies on crash likelihood prediction at intersections and with several established deep learning models trained on the same connected vehicle dataset., Comment: 29 pages, 10 figures, 8 tables

Accurately detecting and predicting lane change (LC)processes of human-driven vehicles can help autonomous vehicles better understand their surrounding environment, recognize potential safety hazards, and improve traffic safety. This paper focuses on LC processes, first developing a temporal convolutional network with an attention mechanism (TCN-ATM) model to recognize LC intention. Considering the intrinsic relationship among output variables, the Multi-task Learning (MTL)framework is employed to simultaneously predict multiple LC vehicle status indicators. Furthermore, a unified modeling framework for LC intention recognition and driving status prediction (LC-IR-SP) is developed. The results indicate that the classification accuracy of LC intention was improved from 96.14% to 98.20% when incorporating the attention mechanism into the TCN model. For LC vehicle status prediction issues, three multi-tasking learning models are constructed based on MTL framework. The results indicate that the MTL-LSTM model outperforms the MTL-TCN and MTL-TCN-ATM models. Compared to the corresponding single-task model, the MTL-LSTM model demonstrates an average decrease of 26.04% in MAE and 25.19% in RMSE.

Vehicles equipped with automated driving capabilities have shown the potential to improve safety and operations. Advanced driver assistance systems (ADAS) and automated driving systems (ADS) have been widely developed to support vehicular automation. Although the studies on the injury severity outcomes that involve automated driving systems are ongoing, there is limited research investigating the difference between injury severity outcomes of the ADAS and ADS vehicles using real-world crash data. To ensure comprehensive analysis, a multi-source dataset that includes the NHTSA crash database (752 cases), CA DMV crash reports (498 cases), and news outlet data (30 cases) is used. Two random parameters multinomial logit models with heterogeneity in the means and variances are estimated to gain a better understanding of the variables impacting the crash injury severity outcome for the ADAS (SAE Level 2) and ADS (SAE Levels 3-5) vehicles. We found that while 56 percent of crashes involving ADAS vehicles took place on a highway, 84 percent of crashes involving ADS took place in more urban settings. The model estimation results indicate that the weather indicators, traffic incident or work zone indicator, differences in the system sophistication that are captured by both manufacture year and high or low mileage, type of collision, as well as rear and front impact indicators all play a significant role in the crash injury severity. The results offer an exploratory assessment of the safety performance of the ADAS and ADS equipped vehicles in the real-world environment and can be used by the manufacturers and other stakeholder to dictate the direction of their deployment and usage.

Speeding has been and continues to be a major contributing factor to traffic fatalities. Various transportation agencies have proposed speed management strategies to reduce the amount of speeding on arterials. While there have been various studies done on the analysis of speeding proportions above the speed limit, few studies have considered the effect on the individual's journey. Many studies utilized speed data from detectors, which is limited in that there is no information of the route that the driver took. This study aims to explore the effects of various roadway features an individual experiences for a given journey on speeding proportions. Connected vehicle trajectory data was utilized to identify the path that a driver took, along with the vehicle related variables. The level of speeding proportion is predicted using multiple learning models. The model with the best performance, Extreme Gradient Boosting, achieved an accuracy of 0.756. The proposed model can be used to understand how the environment and vehicle's path effects the drivers' speeding behavior, as well as predict the areas with high levels of speeding proportions. The results suggested that features related to an individual driver's trip, i.e., total travel time, has a significant contribution towards speeding. Features that are related to the environment of the individual driver's trip, i.e., proportion of residential area, also had a significant effect on reducing speeding proportions. It is expected that the findings could help inform transportation agencies more on the factors related to speeding for an individual driver's trip.

The application of Computer Vision (CV) techniques massively stimulates microscopic traffic safety analysis from the perspective of traffic conflicts and near misses, which is usually measured using Surrogate Safety Measures (SSM). However, as video processing and traffic safety modeling are two separate research domains and few research have focused on systematically bridging the gap between them, it is necessary to provide transportation researchers and practitioners with corresponding guidance. With this aim in mind, this paper focuses on reviewing the applications of CV techniques in traffic safety modeling using SSM and suggesting the best way forward. The CV algorithm that are used for vehicle detection and tracking from early approaches to the state-of-the-art models are summarized at a high level. Then, the video pre-processing and post-processing techniques for vehicle trajectory extraction are introduced. A detailed review of SSMs for vehicle trajectory data along with their application on traffic safety analysis is presented. Finally, practical issues in traffic video processing and SSM-based safety analysis are discussed, and the available or potential solutions are provided. This review is expected to assist transportation researchers and engineers with the selection of suitable CV techniques for video processing, and the usage of SSMs for various traffic safety research objectives.

Crash data of autonomous vehicles (AV) or vehicles equipped with advanced driver assistance systems (ADAS) are the key information to understand the crash nature and to enhance the automation systems. However, most of the existing crash data sources are either limited by the sample size or suffer from missing or unverified data. To contribute to the AV safety research community, we introduce AVOID: an open AV crash dataset. Three types of vehicles are considered: Advanced Driving System (ADS) vehicles, Advanced Driver Assistance Systems (ADAS) vehicles, and low-speed autonomous shuttles. The crash data are collected from the National Highway Traffic Safety Administration (NHTSA), California Department of Motor Vehicles (CA DMV) and incident news worldwide, and the data are manually verified and summarized in ready-to-use format. In addition, land use, weather, and geometry information are also provided. The dataset is expected to accelerate the research on AV crash analysis and potential risk identification by providing the research community with data of rich samples, diverse data sources, clear data structure, and high data quality.

The utilization of traffic conflict indicators is crucial for assessing traffic safety, especially when the crash data is unavailable. To identify traffic conflicts based on traffic flow characteristics across various traffic states, we propose a framework that utilizes unsupervised learning to automatically establish surrogate safety measures (SSM) thresholds. Different traffic states and corresponding transitions are identified with the three-phase traffic theory using high-resolution trajectory data. Meanwhile, the SSMs are mapped to the corresponding traffic states from the perspectives of time, space, and deceleration. Three models, including k-means, GMM, and Mclust, are investigated and compared to optimize the identification of traffic conflicts. It is observed that Mclust outperforms the others based on the evaluation metrics. According to the results, there is a variation in the distribution of traffic conflicts among different traffic states, wide moving jam (phase J) has the highest conflict risk, followed by synchronous flow (phase S), and free flow (phase F). Meanwhile, the thresholds of traffic conflicts cannot be fully represented by the same value through different traffic states. It reveals that the heterogeneity of thresholds is exhibited across traffic state transitions, which justifies the necessity of dynamic thresholds for traffic conflict analysis.

ChatGPT embarks on a new era of artificial intelligence and will revolutionize the way we approach intelligent traffic safety systems. This paper begins with a brief introduction about the development of large language models (LLMs). Next, we exemplify using ChatGPT to address key traffic safety issues. Furthermore, we discuss the controversies surrounding LLMs, raise critical questions for their deployment, and provide our solutions. Moreover, we propose an idea of multi-modality representation learning for smarter traffic safety decision-making and open more questions for application improvement. We believe that LLM will both shape and potentially facilitate components of traffic safety research., Comment: Submitted to Nature - Machine Intelligence (Revised and Extended)

Digital Twin is an emerging technology that replicates real-world entities into a digital space. It has attracted increasing attention in the transportation field and many researchers are exploring its future applications in the development of Intelligent Transportation System (ITS) technologies. Connected vehicles (CVs) and pedestrians are among the major traffic participants in ITS. However, the usage of Digital Twin in research involving both CV and pedestrian remains largely unexplored. In this study, a Digital Twin framework for CV and pedestrian in-the-loop simulation is proposed. The proposed framework consists of the physical world, the digital world, and data transmission in between. The features for the entities (CV and pedestrian) that need digital twining are divided into external state and internal state, and the attributes in each state are described. We also demonstrate a sample architecture under the proposed Digital Twin framework, which is based on Carla-Sumo Co-simulation and Cave automatic virtual environment (CAVE). A case study that investigates Vehicle-Pedestrian (V2P) warning system is conducted to validate the effectiveness of the presented architecture. The proposed framework is expected to provide guidance to the future Digital Twin research, and the architecture we build can serve as the testbed for further research and development of ITS applications on CV and pedestrians.

Surrogate safety measures in the form of conflict indicators are indispensable components of the proactive traffic safety toolbox. Conflict indicators can be classified into past-trajectory-based conflicts and predicted-trajectory-based conflicts. While the calculation of the former class of conflicts is deterministic and unambiguous, the latter category is computed using predicted vehicle trajectories and is thus more stochastic. Consequently, the accuracy of prediction-based conflicts is contingent on the accuracy of the utilized trajectory prediction algorithm. Trajectory prediction can be a challenging task, particularly at intersections where vehicle maneuvers are diverse. Furthermore, due to limitations relating to the road user trajectory extraction pipelines, accurate geometric representation of vehicles during conflict analysis is a challenging task. Misrepresented geometries distort the real distances between vehicles under observation. In this research, a prediction-based conflict identification methodology was proposed. A sequence-to-sequence Recurrent Neural Network was developed to sequentially predict future vehicle trajectories for up to 3 seconds ahead. Furthermore, the proposed network was trained using the CitySim Dataset to forecast both future vehicle positions and headings to facilitate the prediction of future bounding boxes, thus maintaining accurate vehicle geometric representations. It was experimentally determined that the proposed method outperformed frequently used trajectory prediction models for conflict analysis at intersections. A comparison between Time-to-Collision (TTC) conflict identification using vehicle bounding boxes versus the commonly used vehicle center points for geometric representation was conducted. Compared to the bounding box method, the center point approach often failed to identify TTC conflicts or underestimated their severity.

Intersection safety often relies on the correct modelling of signal phasing and timing parameters. A slight increase in yellow time or red time can have significant impact on the rear end crashes or conflicts. This paper aims to identify the relationship between surrogate safety measures and signal phasing. Unmanned Aerial Vehicle (UAV) video data has been used to study an intersection. Post Encroachment Time (PET) between vehicles was calculated from the video data as well as speed, heading and relevant signal timing parameters such as all red time, red clearance time, yellow time, etc. Random Parameter Ordered Logit Model was used to model the relationship between PET and these signal timing parameters. Overall, the results showed that yellow time and red clearance time is positively related to PETs. The model was also able to idendity certain signal phases that could be a potential safety hazard and would need to be retimed by considering the PETs. The odds ratios from the models also indicates that increasing the yellow and red clearance times by one second can improve the PET levels by 16% and 3% respectively.

The development of safety-oriented research and applications requires fine-grain vehicle trajectories that not only have high accuracy, but also capture substantial safety-critical events. However, it would be challenging to satisfy both these requirements using the available vehicle trajectory datasets do not have the capacity to satisfy both.This paper introduces the CitySim dataset that has the core objective of facilitating safety-oriented research and applications. CitySim has vehicle trajectories extracted from 1140 minutes of drone videos recorded at 12 locations. It covers a variety of road geometries including freeway basic segments, signalized intersections, stop-controlled intersections, and control-free intersections. CitySim was generated through a five-step procedure that ensured trajectory accuracy. The five-step procedure included video stabilization, object filtering, multi-video stitching, object detection and tracking, and enhanced error filtering. Furthermore, CitySim provides the rotated bounding box information of a vehicle, which was demonstrated to improve safety evaluations. Compared with other video-based critical events, including cut-in, merge, and diverge events, which were validated by distributions of both minimum time-to-collision and minimum post-encroachment time. In addition, CitySim had the capability to facilitate digital-twin-related research by providing relevant assets, such as the recording locations' three-dimensional base maps and signal timings., Comment: Transportation Research Record (2023)

Toll plaza is a designated area of controlled-access roads like expressway, bridge, and tunnel for toll collection. A number of toll booths are often placed at the toll plaza accommodating high passing traffic and multiple payment methods. Traffic and safety characteristics of toll plazas are different from that of other road entities. Different conflict risk indicators, which are usually longitudinal, have been adopted for real-time safety assessment. In this study, correlated grouped random parameter logit models with heterogeneity in the means are established to capture the unobserved heterogeneity, with additional flexibility, at road user level for the association between conflict risk and influencing factors. In addition, modified conflict risk indicator is developed to assess the safety of diverging, merging, and weaving movements of traffic, with which vehicles' dimensions (width and length), and longitudinal and angular movements are considered. Also, prevalence and severity of both rear-end and sideswipe conflicts are assessed. Results indicate that toll collection type, vehicle's location, average longitudinal speed, angular speed, acceleration, and vehicle class all affect the risk of traffic conflicts. Furthermore, there are significant correlation among the random parameters of severe traffic conflicts. Proposed analytic method can accommodate the conflict risk analysis for different conflict types and account for the correlation of unobserved heterogeneity. Findings should shed light on appropriate remedial measures like traffic signs, road markings, and advanced traffic management system that can improve the safety at tunnel toll plazas.

Crash events identification and prediction plays a vital role in understanding safety conditions for transportation systems. While existing systems use traffic parameters correlated with crash data to classify and train these models, we propose the use of a novel sensory unit that can also accurately identify crash events: microphone. Audio events can be collected and analyzed to classify events such as crash. In this paper, we have demonstrated the use of a deep Convolutional Neural Network (CNN) for road event classification. Important audio parameters such as Mel Frequency Cepstral Coefficients (MFCC), log Mel-filterbank energy spectrum and Fourier Spectrum were used as feature set. Additionally, the dataset was augmented with more sample data by the use of audio augmentation techniques such as time and pitch shifting. Together with the feature extraction this data augmentation can achieve reasonable accuracy. Four events such as crash, tire skid, horn and siren sounds can be accurately identified giving indication of a road hazard that can be useful for traffic operators or paramedics. The proposed methodology can reach accuracy up to 94%. Such audio systems can be implemented as a part of an Internet of Things (IoT) platform that can complement video-based sensors without complete coverage.

In this work, we focus on detecting emergency vehicles using only audio data. Improved and quick detection can help in faster preemption of these vehicles at signalized intersections thereby reducing overall response time in case of emergencies. Important audio features were extracted from raw data and passed into extreme learning machines (ELM) for training. ELMs have been used in this work because of its simplicity and shorter run-time which can therefore be used for online learning. Recently, there have been many studies that focus on sound classification but most of the methods used are complex to train and implement. The results from this paper show that ELM can achieve similar performance with exceptionally shorter training times. The accuracy reported for ELM is about 97% for emergency vehicle detection (EVD).