PUBLICATIONS
* Corresponding author. A full list is also available on Google Scholar.
Journal Publications(26)
Edge-based multimodal sensor data fusion with Vision-Language-Action (VLA) model for real-time autonomous vehicle accident avoidance
Journal ArticleSep 2026 • Engineering Applications of Artificial Intelligence • 180 • Elsevier
Fengze Yang, Bo Yu, Yang Zhou, Xuewen Luo, Zhengzhong Tu, Chenxi Liu*
DOI: 10.1016/j.engappai.2026.115186
Autonomous driving (AD) systems relying solely on onboard sensors may fail to detect distant or occluded hazards, potentially causing preventable collisions; however, existing transformer-based Vehicle-to-Everything (V2X) approaches either lack effective multimodal fusion and reasoning or struggle to meet real-time performance requirements under complex traffic conditions. This paper proposes the Real-time Edge-based Autonomous Co-pilot Trajectory planner (REACT), a V2X-integrated trajectory optimization framework for AD based on a fine-tuned lightweight Vision–Language–Action (VLA) model explicitly grounded in low-level control generation. REACT integrates infrastructure-provided hazard alerts with onboard sensor data, capturing surrounding traffic dynamics and vehicle intents through visual embeddings, interpreting numerical data from symbolic inputs, and employing contextual reasoning to predict control-aware residual action sequences. To enable robust real-time deployment on edge devices, REACT employs a Residual Trajectory Fusion (RTF) module that predicts discrete residual controls and fuses them with a nominal plan, together with edge-adaptation strategies to reduce model complexity and improve inference efficiency. Evaluated on the DeepAccident benchmark, REACT achieves state-of-the-art performance, a 84.9% collision rate reduction, a 50.6% Video Panoptic Quality, and a 0.57-second inference latency on the Jetson AGX Orin. Ablation studies validate the contribution of each input modality, architectural component, and edge adaptation strategy. These results highlight the effectiveness of lightweight control-grounded VLA reasoning for real-time cooperative planning on edge platforms and underscore the potential of language-guided residual control generation for improving traffic safety.
Enhancing Deep Learning–Based Segmentation for Obstructive Roadside Element Detection in Point Cloud Data from Mobile Laser Scanning in Rural Road Environments
Journal ArticleAug 2026 • Journal of Transportation Engineering, Part A: Systems • 152(8) • American Society of Civil Engineers (ASCE)
Nutvara Jantarathaneewat, Chenxi Liu, Shucheng Zhang, Yinhai Wang*
DOI: 10.1061/JTEPBS.TEENG-9479
This study presents a deep learning-based approach for the semantic segmentation of wall-like roadside structures, such as buildings and fences, from mobile laser scanning point cloud data collected in rural road environments. These fixed roadside elements, particularly when positioned close to the roadway, pose significant safety hazards in run-off-road incidents. Traditional field-based identification methods are labor-intensive and inefficient for large-scale applications. To address this limitation, we investigate the effectiveness of transfer learning by fine-tuning pretrained segmentation models on a custom dataset representing approximately 20 km of rural roads in Thailand. Two state-of-the-art models, KPConv and RandLA-Net, originally trained on the SemanticKITTI dataset, were evaluated. To enhance task relevance and reduce class imbalance, the segmentation task was restructured to predict three classes: ground, nonground, and wall-like objects. Fine-tuning was carried out by modifying the output layer to match the new class definitions and retraining selected layers on the custom dataset. The fine-tuned KPConv model outperformed alternatives, achieving an overall accuracy of 73.2%, mean intersection over union of 68.2%, and an 66.7% detection rate for wall-like structures. Additional experiments investigated the impact of including intensity as an input feature; however, no significant improvements in performance were observed, likely due to variability in object materials and inconsistent reflectivity in rural scenes. The findings demonstrate that transfer learning, when combined with targeted fine-tuning, allows pretrained models to be effectively adapted to novel environments with limited annotated data. However, as the fine-tuning was conducted on five road segments representing different rural contexts, the results primarily reflect feasibility rather than full generalization across diverse sites. This approach provides a cost-efficient and scalable solution for rural road infrastructure analysis and supports applications such as automated asset inventory, safety audits, and lateral clearance assessments.
ATSDAS: A Vision-Based System for Automated Traffic Sign Management and Condition Assessment
Journal ArticleAug 2026 • IEEE Transactions on Intelligent Transportation Systems • 27(8) • Institute of Electrical and Electronics Engineers (IEEE)
Shucheng Zhang, Chenxi Liu, Nutvara Jantarathaneewat, Yinhai Wang*
DOI: 10.1109/TITS.2026.3683626
The maintenance and management of traffic signal assets are crucial for ensuring the safety and smooth operation of road transportation. However, timely repairs and updates of damaged traffic signs remain a significant challenge due to the inefficiency of manual measurements. This study proposes an innovative system that automates traffic sign collection and assessment using deep learning techniques. The system performs three key tasks, including traffic sign detection and tracking, condition assessment, and inventory building, by processing dashboard camera videos with GPS data or open-source street imagery. The system achieves a $\text {mAP}_{50}$ of 93.50% for detection and recognition and an accuracy of 93.56% for damage assessment by utilizing a retroreflectivity-related image feature extraction pipeline. The outputs are formalized into a visual inventory, enabling transportation agencies to manage traffic assets more efficiently. Field tests conducted in collaboration with the Washington State Department of Transportation validate the practical effectiveness of the framework. To further address the lack of publicly available condition datasets, this study also introduces the Damaged Traffic Sign Detection and Assessment (DTSDA) dataset, which combines real and synthetic images of damaged signs. Overall, the proposed approach offers a cost-effective and scalable alternative to the traditional labor-intensive method, thereby enhancing traffic asset management efficiency and contributing to safer transportation systems.
From Camera-Based Sensing to Reasoning: A Comprehensive Review Toward Proactive Vulnerable Road User Safety
Journal ArticleJul 2026 • IEEE Transactions on Intelligent Transportation Systems • Institute of Electrical and Electronics Engineers (IEEE)
Shucheng Zhang, Yan Shi, Bingzhang Wang, Yuang Zhang, Muhammad Monjurul Karim, Kehua Chen, Chenxi Liu, Mehrdad Nasri, Yinhai Wang*
DOI: 10.1109/TITS.2026.3707876
Ensuring the safety of vulnerable road users (VRUs), such as pedestrians and cyclists, remains a critical challenge, as conventional infrastructure-based measures are often insufficient in dynamic urban environments. Recent advances in learning-based visual sensing systems have created new opportunities for more adaptive and context-aware VRU protection. This paper presents a comprehensive review of camera-based methods for proactive VRU safety, with a focus on developments over the past five years. Unlike prior surveys that primarily emphasize detection, we organize the literature into three interconnected components: visual perception (detection and classification), motion modeling (tracking and trajectory prediction), and behavior understanding (intent recognition and reasoning). These components form a unified hierarchical pipeline that enables early risk anticipation and timely intervention. Furthermore, this survey systematically incorporates emerging AI paradigms, including Vision Transformers (ViTs), Large Language Models (LLMs), and diffusion models, highlighting their roles in representation learning, uncertainty modeling, and semantic reasoning. Finally, we identify four key challenges specific to VRU safety, including data scarcity, behavioral uncertainty, edge deployment efficiency, and real-world sensing constraints, and discuss corresponding research directions. This work provides a unified foundation for the development of reliable, scalable, and deployable VRU safety systems.
Enhancing Urban Paratransit Reliability: Spatial–Temporal and Causal Analysis of Service Inefficiencies
Journal ArticleJun 2026 • Journal of Urban Planning and Development • 152(2) • American Society of Civil Engineers (ASCE)
Arman Malekloo, Xiaoyue Cathy Liu, Nikola Marković, Chenxi Liu, Jeff Phillips
DOI: 10.1061/JUPDDM.UPENG-6200
Reliable paratransit services are essential for urban mobility, particularly for individuals with disabilities who depend on demand-responsive transportation. However, service inefficiencies in urban paratransit systems, such as travel time variability, congestion impacts, and scheduling constraints, continue to pose significant challenges. This study introduces a multitier analytical framework that integrates spatial–temporal modeling, machine learning–based outlier detection, and causal inference to systematically evaluate the reliability of paratransit services in an urban context. Leveraging trip transaction data from a metropolitan paratransit system, we develop the paratransit efficiency index (PEI) to assess travel time reliability at both system-wide and individual trip levels. We then analyze PEI's spatial–temporal variability using geographically and temporally weighted regression and identify outlier trips with high PEI using XGBoost to pinpoint service unreliability and systemic inefficiencies. Later, we utilize causal inference techniques to show that peak-hour pick-ups causally contribute to travel time inefficiency (a 19.4% relative increase in outlier probability), whereas subscription-based bookings causally improve service consistency (a 33.9% relative decrease). The findings provide actionable insights for urban planners and transit agencies to optimize scheduling, mitigate congestion effects, and explore innovative strategies such as integrating transportation network companies for high-cost or unreliable trips. By addressing critical urban transportation equity issues, this study offers data-driven solutions to enhance the efficiency and resilience of paratransit services in growing metropolitan areas.
Traffic co-simulation framework empowered by infrastructure camera sensing and reinforcement learning
Journal ArticleSep 2025 • Journal of Intelligent Transportation Systems • Taylor & Francis
Talha Azfar, Kaicong Huang, Andrew Tracy, Sandra Misiewicz, Chenxi Liu, Ruimin Ke
DOI: 10.1080/15472450.2025.2559410
Traffic simulations are commonly used to optimize urban traffic flow, with reinforcement learning (RL) showing promising potential for automated traffic signal control, particularly in intelligent transportation systems involving connected automated vehicles. Multi-agent reinforcement learning (MARL) is particularly effective for learning control strategies for traffic lights in a network using iterative simulations. However, existing methods often assume perfect vehicle detection, which overlooks real-world limitations related to infrastructure availability and sensor reliability. This study proposes a co-simulation framework integrating CARLA and SUMO, which combines high-fidelity 3D modeling with large-scale traffic flow simulation. Cameras mounted on traffic light poles within the CARLA environment use a YOLO-based computer vision system to detect and count vehicles, providing real-time traffic data as input for adaptive signal control in SUMO. MARL agents trained with four different reward structures leverage this visual feedback to optimize signal timings and improve network-wide traffic flow. Experiments in a multi-intersection test-bed demonstrate the effectiveness of the proposed MARL approach in enhancing traffic conditions using real-time camera based detection. The framework also evaluates the robustness of MARL under faulty or sparse sensing and compares the performance of YOLOv5 and YOLOv8 for vehicle detection. Results show that while better accuracy improves performance, MARL agents can still achieve significant improvements with imperfect detection, demonstrating scalability and adaptability for real-world scenarios.
A Self-Supervised Multi-Agent Large Language Model Framework for Customized Traffic Mobility Analysis Using Machine Learning Models
Journal ArticleJul 2025 • Transportation Research Record Journal of the Transportation Research Board • 2679(7) • SAGE Publications
Fengze Yang, Xiaoyue Cathy Liu, Lingjiu Lu, Bingzhang Wang, Chenxi Liu*
DOI: 10.1177/03611981251322468
With the urbanization process, an increasing number of sensors are being deployed in transportation systems, leading to an explosion of big data. To harness the power of this vast transportation data, various machine learning (ML) and artificial intelligence (AI) methods have been introduced to address numerous transportation challenges. However, these methods often require significant investment in data collection, processing, storage, and the employment of professionals with expertise in transportation and ML. Additionally, privacy issues are a major concern when processing data for real-world traffic control and management. To address these challenges, the research team proposes an innovative multi-agent framework called Independent Mobility Generative Pre-Trained Transformers (IDM-GPT) based on large language models (LLMs) for customized traffic analysis, management suggestions, and privacy preservation. IDM-GPT efficiently connects users, transportation databases, and ML models economically. IDM-GPT trains, customizes, and applies various LLM-based AI agents for multiple functions, including user query comprehension, prompts optimization, data analysis, model selection, and performance evaluation and enhancement. With IDM-GPT, users without any background in transportation or ML can efficiently and intuitively obtain data analysis and customized suggestions in near real-time based on their questions. Experimental results demonstrate that IDM-GPT delivers satisfactory performance across multiple traffic-related tasks, providing comprehensive and actionable insights that support effective traffic management and urban mobility improvement.
Attention-Based Feature Fusion Empowered Encoder-Decoder Framework for Nighttime Traffic Perception From High-Altitude Surveillance System
Journal ArticleJul 2025 • IEEE Transactions on Intelligent Transportation Systems • 26(7) • Institute of Electrical and Electronics Engineers (IEEE)
Qianxia Cao, Zhenyu Shan, Chenxi Liu*, Mingxin Yan
DOI: 10.1109/TITS.2025.3560539
Traffic perception is essential for intelligent transportation systems (ITS) to improve road safety and facilitate traffic management. High-altitude video surveillance systems, known for their wide coverage, provide an effective solution for continuous traffic monitoring and are increasingly deployed worldwide. However, the application of high-altitude video surveillance faces significant challenges, such as the small size and high density of vehicles, and low ambient brightness at night or low-light conditions. In this study, we propose a novel lightweight model specifically for reliable traffic perception in low-light conditions, utilizing an encoder-decoder architecture. The proposed framework begins with an advanced image enhancement algorithm designed to amplify the semantic content of low-brightness nighttime images. Subsequently, we employ a lightweight, deep two-channel neural network to extract image features both pre- and post-enhancement. To effectively manage the noise introduced during image enhancement, we introduce a feature fusion module based on an attention mechanism, which adaptively merges the features extracted by the two-channel network. The final step involves a lightweight decoder that synergizes high-level semantic information with low-level features. Extensive evaluation on our nighttime dataset shows that the proposed method can reduce error by 40% compared to the methods without image enhancement and by 28% compared to those relying solely on enhancement, outperforming all baseline methods. Notably, the lightweight module embedded in our model empowers the framework to perform real-time traffic perception at night, providing an efficient solution for traffic monitoring.
Advancing Automatic Asset Management: An edge-based US-specific Traffic Sign Detection and Recognition (TSDR) System based on Image Processing
Journal ArticleMar 2025 • Journal of Transportation Engineering Part A Systems • 151(3) • American Society of Civil Engineers (ASCE)
Chenxi Liu, Nutvara Jantarathaneewat, Shucheng Zhang, Hao Yang, Xin Fu, Yinhai Wang*
DOI: 10.1061/JTEPBS.TEENG-8473
Effective asset management is crucial in transportation systems, guaranteeing the optimal use, upkeep, and durability of infrastructure, which in turn boosts the safety, efficiency, and sustainability of travel networks. Traffic signs management, as a vital component of asset management, plays a key role in maintaining road safety and efficiency; however, their maintenance demands substantial time and resources investment. The advent of advanced sensing technologies in intelligent transportation systems (ITS) presents an opportunity for more effective and precise asset management. Yet, the challenge lies in the lack of localized traffic sign data in the US, which hinders the implementation of these technologies. To address this gap, our research introduces a traffic sign detection and recognition (TSDR) architecture designed to automatically collect traffic sign information and establish a US-specific data inventory. Recognizing the limitations of existing public traffic sign data sets, which are not tailored for US traffic signs, we collected an additional 5,000 traffic sign images from the Washington State area using Google Map application programming interface (API) and self-installed dash cameras. These signs were manually labeled into 43 classes for training purposes. With the training process, the proposed TSDR model can achieve impressive accuracy (98.34% in detection and 97.10% in recognition). In summary, we developed an automated pipeline, TSDR, for capturing, detecting, classifying, and storing traffic signs, culminating in the creation of a localized traffic sign data inventory for the US.
Potentials and Challenges of AI-Empowered Solutions to Urban Transportation Infrastructure Systems (UTIS): NSF AI-Transportation Workshop Phase I
Journal ArticleSep 2024 • Journal of Transportation Engineering Part A Systems • 150(9) • American Society of Civil Engineers (ASCE)
Chenxi Liu, Chenlu Pu, Lili Du, Yinhai Wang*
Frontiers of Emerging AI Technologies Best Practices and Workforce Development in Transportation: NSF AI-Transportation Workshop Phase II
Journal ArticleSep 2024 • Journal of Transportation Engineering Part A Systems • 150(9) • American Society of Civil Engineers (ASCE)
Chenlu Pu, Chenxi Liu, Yinhai Wang, Lili Du*
Traffic Performance Score: Measure Urban Mobility and Online Predict Near-term Traffic like Weather Forecast
Journal ArticleAug 2024 • Transportation Research Record Journal of the Transportation Research Board • 2678(8) • SAGE Publications
Zhiyong Cui, Meng-Ju Tsai, Meixin Zhu, Hao Yang, Chenxi Liu, Shuyi Yin, Yinhai Wang*
DOI: 10.1177/03611981231222232
Measuring traffic performance is critical for public agencies which manage traffic and individuals who. This is the topic which the authors attempt to emphasize. One potential challenge for traffic prediction tasks is that short-term-incident-induced traffic pattern changes cannot be timely detected and the deployed model cannot adapt to the new traffic pattern. As for encountering long-term incidents, such as during COVID-19, traffic patterns are gradually changing, and the prediction model also needs to be periodically updated to avoid the so-called out-of-distribution problem. Therefore, the online training and predicting mechanisms can facilitate model updates, deployment of traffic prediction applications, and the planning of trips, especially when special events happen, such as the long-lasting COVID-19 pandemic. However, most existing traffic performance metrics narrowly focus on one aspect of the impacts but not comprehensive changes to the network. Further, during the pandemic, urban traffic patterns and travelers’ trip planning were dramatically affected and, thus, network-wide online traffic prediction became an urgent but more complicated task. To overcome such challenges, this study proposes a traffic performance score (TPS) incorporating multiple parameters for measuring both urban and freeway network-wide traffic performance. The TPS is compared with other metrics to show its superiority. To solve the challenging network-wide online traffic prediction task, this study also proposes an online training and updating strategy to predict network-wide traffic performance. Experimental results indicate that the proposed model with the online learning strategy outperforms existing methods in prediction accuracy and learning efficiency. In addition, the TPS measurement and its related online prediction functions are implemented on a publicly accessible platform and applied in real practice, which is another contribution of this work.
Mitigating biases in big mobility data: a case study of monitoring large-scale transit systems
Journal ArticleJul 2024 • Transportation Letters • 17(4) • Taylor & Francis
Feilong Wang, Xuegang (Jeff) Ban, Peng Chen, Chenxi Liu, Rong Zhao*
DOI: 10.1080/19427867.2024.2379703
Big mobility data (BMD) have shown many advantages in studying human mobility and evaluating the performance of transportation systems. However, the quality of BMD remains poorly understood. This study evaluates biases in BMD and develops mitigation methods. Using Google and Apple mobility data as examples, this study compares them with benchmark data from governmental agencies. Spatio-temporal discrepancies between BMD and benchmark are observed and their impacts on transportation applications are investigated, emphasizing the urgent need to address these biases to prevent misguided policymaking. This study further proposes and tests a bias mitigation method. It is shown that the mitigated BMD could generate valuable insights into large-scale public transit systems across 100+ US counties, revealing regional disparities of the recovery of transit systems from the COVID-19. This study underscores the importance of caution when using BMD in transportation research and presents effective mitigation strategies that would benefit practitioners.
Models for predicting vehicle emissions: A comprehensive review
Journal ArticleMay 2024 • The Science of The Total Environment • 923 • Elsevier
Hui Zhong, Kehua Chen, Chenxi Liu, Meixin Zhu*, Ruimin Ke
DOI: 10.1016/j.scitotenv.2024.171324
Air pollution is a primary concern, causing around 7 million premature deaths annually, with traffic-related sources contributing 23 %-45 % of emissions. While several studies have surveyed vehicle emission models, they are either outdated or focus on specific data-driven models. This paper systematically reviews vehicle emission prediction models, comparing traditional approaches with data-driven emission models. The traditional emission models can be divided into average-speed, modal, and other models, noting their reliance on empirical assumptions and parameters that may not be universally applicable. In contrast, we delve into data-driven models utilizing dynamometer and on-road test data for time-series and spatial-temporal predictions. The application of these models is discussed across various scenarios, highlighting the progress and gap. We observed that traditional models, primarily estimating total traffic emissions in study regions, lack micro-level detail crucial for tailored decisions. The direct link between road emission model accuracy and input data quality poses challenges in disaggregating on-road vehicle emission inventories. Due to unique transportation instruments, traffic fleet components, and patterns, exploring the effects of emission-reduction policies in specific cities or regions is urgent. Vehicle characteristics, environmental conditions, traffic scenarios, and prediction scales are common effect factors, while instantaneous driving profiles prove effective in model calibration. In data-driven models, ANN outperforms in estimating emissions and performance of low-power diesel engines with errors not exceeding 5 %. However, no single data-driven method performed excellently in predicting all pollutants. Besides, integrated methods utilizing LSTM, GRU, and RNN outperform individual models. To enhance prediction accuracy considering the inherent connectivity of road networks and spatiotemporal variation patterns of vehicle emissions, GCN is an emerging approach for capturing spatial-temporal relationships based on remote sensing data. Moreover, limited data-driven studies have been performed to forecast particle matter emissions, the main contributors to urban pollution, calling for more attention for future research.
Cooperative and Comprehensive Multi-task Surveillance Sensing and Interaction System Empowered by Edge Artificial Intelligence
Journal ArticleSep 2023 • Transportation Research Record Journal of the Transportation Research Board • 2677(9) • SAGE Publications
Chenxi Liu, Hao Yang, Ruimin Ke, Wei Sun, Julia Wang, Yinhai Wang*
DOI: 10.1177/03611981231160174
Numerous sensors were introduced to intelligent transportation systems (ITS) in the past decade. Consequently, new sensing technologies and the generated data attracted more and more attention, which brought new challenges, such as redundant sensors, huge maintenance costs, and data explosion, to ITS. To satisfy the core demands of traffic agencies in a more effective and efficient way, therefore, this paper proposes the idea of “Sensing as a Service” (SaaS) and implements it in the Cooperative and Comprehensive Smart Edge Node for Sensing and OpeRation (COCO SENSOR) system for practical deployments. COCO SENSOR is an innovative multi-task sensing system, which was developed to address key practical applications, including real-time vehicle counting and recognition, road-surface-condition classification, visibility estimation, and live communication among traffic controllers and road users in a single unit. COCO SENSOR introduced customized cooperative sensing and a parallel computation mechanism to increase perception accuracy and efficiency with limited computation resources on the edge device. In collaboration with the Washington State Department of Transportation and the City of Bellevue, a field experiment was conducted to test the system’s performance. The results of the experiment show that the COCO SENSOR effectively fills the gap between sensing and traffic services and that it successfully executed four practical applications, including traffic-volume counting by vehicle type, traffic-status detection, road-visibility estimation, and road-surface-condition classification, with high accuracy. Additionally, a mobile app was developed for both traffic managers and users to access comprehensive traffic information and live warning messages.
Real-time Multi-task Environmental Perception System for Traffic Safety Empowered by Edge Artificial Intelligence
Journal Article2023 • IEEE Transactions on Intelligent Transportation Systems • 25(1) • Institute of Electrical and Electronics Engineers (IEEE)
Chenxi Liu, Hao Yang, Meixin Zhu, Feilong Wang, Torgeir Vaa, Yinhai Wang*
DOI: 10.1109/TITS.2023.3309100
Traffic safety, reliability, and resilience are significantly influenced by environmental factors such as visibility, road surface, and weather conditions. Yet, current monitoring methods, including weather stations and onboard environmental sensors, often fall short due to their high costs, significant latency, and limited dissemination. This paper presents the Edge-based Multi-task Safety-oriented Environmental (Edge-MuSE) sensing system, designed to address these traffic safety challenges associated with environmental factors. Edge-MuSE departs from traditional single-task sensing methods by performing multidimensional traffic environment perception tasks. It estimates key safety-related environmental factors exclusively through camera inputs and incorporates four innovative sensing tasks: visibility estimation, image dehazing, road segmentation, and road surface condition classification. The system is tailored to edge devices to transition computational loads from central servers to distributed nodes, thereby enhancing privacy and reducing latency. Additionally, Edge-MuSE integrates communication functions based on TCP/IP and Wi-Fi protocols, enabling rapid dissemination of sensing results and warning messages to local road users. System structures and data streaming have been optimized to accommodate the constraints of edge devices, ensuring high-efficiency edge computing. Field testing of Edge-MuSE in multiple testbeds in Bellevue (WA, US) and Oslo (Norway) has demonstrated its reliable and precise performance in perception tasks (92.15% accuracy in visibility estimation and 92.25% in road surface condition classification) as well as an impressive processing speed of 21.3 FPS. As such, Edge-MuSE presents a promising solution for enhancing roadway safety, efficiency, and resilience.
Toward a Dynamic Reversible Lane Management Strategy by Empowering Learning-Based Predictive Assignment Scheme
Journal ArticleDec 2022 • IEEE Transactions on Intelligent Transportation Systems • 23(12) • Institute of Electrical and Electronics Engineers (IEEE)
Chenxi Liu, Hao Yang, Ruimin Ke, Yinhai Wang*
DOI: 10.1109/TITS.2022.3202487
Traffic congestion is a long-lasting worldwide problem and even becoming more severe in well-developed regions. Reversible lanes have been used worldwide on various road types to mitigate the effects of congestion and optimize mobility since the 1930s. However, with the limitation of traditional control and management methods, existing solutions can not meet the increasing travel demands. Therefore, the paper introduces a Predictive Empowered Assignment scheme for Reversible Lane (PEARL). By integrating the advanced traffic flow prediction module and bi-level optimization model, PEARL can be a more flexible dynamic lane assignment strategy with foresight compared to traditional lane management methods. In the prediction module, taking advantage of the development of machine learning technologies, the input of PEARL covers not only historical sequence and real-time data but also the environmental conditions in the region. The advanced Bi-directional Long Short-term Memory (Abi-LSTM) model is employed for short-term traffic flow prediction. Then, the study introduces a bi-level optimization method to maximize the total throughput in both directions and minimize the total user costs which determine the lane deployment. The iterations on the prediction module and optimization module can help PEARL coordinate the future lane control plan and make the best decision. Finally, the paper builds up an experiment platform to simulate PEARL in a real-world scenario with heavy input flow for its performance evaluation.
Traffic-Informed Multi-Camera Sensing (TIMS) System Based on Vehicle Re-Identification
Journal ArticleOct 2022 • IEEE Transactions on Intelligent Transportation Systems • 23(10) • Institute of Electrical and Electronics Engineers (IEEE)
Hao Yang, Jiarui Cai, Meixin Zhu, Chenxi Liu, Yinhai Wang*
DOI: 10.1109/TITS.2022.3154368
Surveillance cameras are widely deployed traffic sensors, due to their affordable prices and being able to capture rich information. However, current surveillance systems have not been fully exploited: these cameras are isolated and can only extract information from their own fixed views. To enable a collaborative sensing system, we propose a novel framework called Traffic-Informed Multi-camera Sensing (TIMS) system for network-level traffic information estimation. By pushing multi-camera Re-IDentification (ReID) workflow towards network-wide traffic information extraction, TIMS system integrates a customized metric-learning vision-based vehicle ReID method (TIM-ReID) and establishing a traffic-informed workflow. To integrate the traffic network connection information along with visual and vehicle attributes features, the road network is extracted as a weighted graph through the Spatial-temporal Camera Graph Inference Model (StCGIM) and serves for matching and re-ranking ReID candidates. Moreover, an Accuracy Model (AAM) is designed to provide accurate, reliable and comprehensive traffic information estimation, including both the values and distribution of parameters under a high penetration rate. In experiments based on real-world multi-camera datasets captured in the city of Seattle, the customized TIM-ReID outperforms existing state-of-the-art methods, and delivers accurate cross-camera information estimation, whose value error is less than 8% and the Kullback-Leibler (KL) distance between the estimated and real distribution is less than 3.42 among all the evaluated camera pairs. TIMS system empowers cameras to work collaboratively through an interactive brain, and provides users with valuable and comprehensive traffic information.
Truck Parking Pattern Aggregation and Availability Prediction by Deep Learning
Journal ArticleAug 2022 • IEEE Transactions on Intelligent Transportation Systems • 23(8) • Institute of Electrical and Electronics Engineers (IEEE)
Hao Yang, Chenxi Liu, Yifan Zhuang, Wei Sun, Karthik Murthy, Ziyuan Pu, Yinhai Wang*
DOI: 10.1109/TITS.2021.3117290
With the significant increase of e-commerce, freight transportation demand has surged significantly over the past decade. Most of the demand has been served by trucks in the United States. One major problem commonly identified across the country is the worsening truck parking availability because the increase of truck parking facilities has lagged behind the growth of trucking activities. The lack of parking spaces and real-time parking availability information greatly exacerbate the uncertainty of trips, and often results in illegal and potentially dangerous parking or overtime driving. This paper elaborates on pilot research on improving truck parking facilities cooperated with the Washington State Department of Transportation (WSDOT), building and testing the advanced Truck Parking Information and Management System (TPIMS) with the real-time user visualization and prediction function empowered by artificial intelligence. Furthermore, by analyzing the activities of truck drivers, the researchers aggregated the regularity of truck parking patterns by a customized sequential similarity methodology. A Truck Parking Occupancy Prediction (TPOP) neural network for time-variant occupancy prediction by deep learning and attributes embedding is proposed and integrated into the TPIMS. The TPOP achieves 5.82%, 5.07%, 4.84%, and 4.19% mean average percentage error (MAPE) for 16, 8, 4, and 2 minutes ahead of occupancy prediction respectively, significantly outperforms other state-of-the-art methods. Clearly, the proposed solutions can benefit both the truck drivers and government agencies by a more efficient and smart TPIMS.
Real-time Crash Identification Using Connected Electric Vehicle Operation Data
Journal ArticleAug 2022 • Accident Analysis & Prevention • 173 • Elsevier
Meixin Zhu, Hao (Frank) Yang, Chenxi Liu, Ziyuan Pu, Yinhai Wang*
DOI: 10.1016/j.aap.2022.106708
As the automobile market gradually develops towards intelligence, networking, and information-orientated, intelligent identification based on connected vehicle data becomes a key technology. Specifically, real-time crash identification using vehicle operation data can enable automotive companies to obtain timely information on the safety of user vehicle usage so that timely customer service and roadside rescue can be provided. In this paper, an accurate vehicle crash identification algorithm is developed based on machine learning techniques using electric vehicles' operation data provided by SAIC-GM-Wuling. The point of battery disconnection is identified as a potential crash event. Data before and after the battery disconnection is retrieved for feature extraction. Two different feature extraction methods are used: one directly extracts the descriptive statistical features of various variables, and the other directly unfolds the multivariate time series data. The AdaBoost algorithm is used to classify whether a potential crash event is a real crash using the constructed features. Models trained with the two different features are fused for the final outputs. The results show that the final model is simple, effective, and has a fast inference speed. The model has an F1 score of 0.98 on testing data for crash classification, and the identified crash times are all within 10 s around the true crash times. All data and code are available at https://github.com/MeixinZhu/vehicle-crash-identification.
How Fast You Will Drive? Predicting Speed of Customized Paths By Deep Neural Network
Journal ArticleMar 2022 • IEEE Transactions on Intelligent Transportation Systems • 23(3) • Institute of Electrical and Electronics Engineers (IEEE)
Hao Yang, Chenxi Liu, Meixin Zhu, Xuegang Ban, Yinhai Wang*
DOI: 10.1109/TITS.2020.3031026
Customized path-based speed prediction is an eventful tool for congestion avoidance, route optimization and travel time prediction for navigation apps, cab-hailing companies and autonomous vehicles. Traditionally, the speed prediction algorithms are based on road segments and can only support several main roads. Path-based speed prediction is very challenging since the speed is always changing in different path locations and is jointly affected by lots of complicated factors. This article presents a novel deep learning framework for customized path-based speed prediction. A Path-based Speed Prediction Neural Network (PSPNN) is designed to achieve speed predictions for a given path and attributes information. A hierarchical Convolutional Neural Network (CNN) and deep Bidirectional Long Short-Term Memory (Bi-LSTM) structure for different kinds of feature extraction are applied for multiple levels: the path cell, sub-path and the whole path. The method narrows down the prediction unit from road segments to customized path cells (mean length: 59.52m) and achieves a mean absolute error (MAE) of 1.94 m/s and Mean Absolute Percentage Error (MAPE) of 18.14%, showing the potential of serving rigorous data-driven applications. So far, PSPNN is the first made-to-order path-based speed prediction algorithm and can help both travelers and managers to obtain large-scale bespoke paths speed information in advance.
Real-Time Traffic and Road Surveillance with Parallel Edge Intelligence
Journal ArticleJan 2022 • IEEE Journal of Radio Frequency Identification • 6 • Institute of Electrical and Electronics Engineers (IEEE)
Ruimin Ke, Chenxi Liu, Hao Yang, Wei Sun, Yinhai Wang*
DOI: 10.1109/JRFID.2022.3211262
Edge computing, which is an emerging concept to complement cloud computing, processes data closer to where the data is generated at the network edge. Transportation system is a critical component of our cities and is generating more and more data. To leverage the power of the increasing amount of data from traffic sensors, researchers have started exploring the feasibility of edge computing for traffic big data analytics. Given the resource constraints on edge devices, it is very challenging to achieve high intelligence and real-time capability at the same time. There are few edge transportation systems that function in a multi-tasking mode with high efficiency and intelligence on the edge. This paper proposes a multi-thread parallel edge system architecture and several lightweight algorithms for traffic and road surveillance. The designs ensure robustness and real-time capability for the key modules of the system.
Cooperative multi-camera vehicle tracking and traffic surveillance with edge artificial intelligence and representation learning
Journal Article2022 • Transportation Research Part C Emerging Technologies • 148 • Elsevier
Hao (Frank) Yang, Jiarui Cai, Chenxi Liu, Ruimin Ke, Yinhai Wang*
DOI: 10.1016/j.trc.2022.103982
Traffic surveillance cameras are the eyes of the Intelligent Transportation Systems (ITS). However, they are currently isolated and can only extract information from each of their fixed views. To track vehicles across multiple cameras and help public agencies collect link travel time and speed information, an Edge-empowered Cooperative Multi-camera Sensing (ECoMS) System is proposed. ECoMS system presents a novel algorithmic and edge-server cooperative system construct to push edge computing and multi-camera re-identification workflow serving for traffic sensing based on Internet of Things (IoT) architecture. On the algorithm side, ECoMS system proposes a featherlight edge-based computer vision framework for vehicle detection, tracking, and features selection process in a real-time manner. Then, by only sending the objects’ representations to the server, the high-bandwidth data transmission and the heavy post-processing system can be abandoned. Furthermore, a hierarchical clip-based deep vehicle re-identification framework is proposed and integrated into the ECoMS system, and significantly outperforms other state-of-the-art methods by 4%–8% on Rank-1 accuracy. Finally, to balance the accuracy level of different camera pairs, a collaborative cross-camera traffic information estimation framework based on kernel density estimation with kernel smoother is implemented, which can get the precise link and region traffic information together with distributions (less than 1.01 KL distance). By maximizing the cooperation of the computational resources, orchestrating the data transmission and integrating road network graph features in the system, the ECoMS can precisely model the network-scale traffic information in a flexible, cost-effective, and scalable workflow. To the author’s best knowledge, ECoMS is the first multi-camera vehicle tracking and traffic monitoring system based on cooperative IoT architecture.
When Intelligent Transportation Systems Sensing Meets Edge Computing: Vision and Challenges
Journal ArticleOct 2021 • Applied Sciences • 11(20) • MDPI
Xuan Zhou, Ruimin Ke*, Hao Yang, Chenxi Liu
The widespread use of mobile devices and sensors has motivated data-driven applications that can leverage the power of big data to benefit many aspects of our daily life, such as health, transportation, economy, and environment. Under the context of smart city, intelligent transportation systems (ITS), as a main building block of modern cities, and edge computing (EC), as an emerging computing service that targets addressing the limitations of cloud computing, have attracted increasing attention in the research community in recent years. It is well believed that the application of EC in ITS will have considerable benefits to transportation systems regarding efficiency, safety, and sustainability. Despite the growing trend in ITS and EC research, a big gap in the existing literature is identified: the intersection between these two promising directions has been far from well explored. In this paper, we focus on a critical part of ITS, i.e., sensing, and conducting a review on the recent advances in ITS sensing and EC applications in this field. The key challenges in ITS sensing and future directions with the integration of edge computing are discussed.
Resilience modeling method of airport network affected by global public health events
Journal ArticleFeb 2021 • Mathematical Problems in Engineering • 2021(1) • Hindawi
Jiuxia Guo, Xinping Zhu, Chenxi Liu, ShuzhiSam Ge
The COVID-19 global pandemic hit the aviation industry hard since the end of 2019. It has had an immediate, dramatic impact on airport traffic and revenue. Airports are the important nodes in the aviation system network, and the failure of a single airport can often affect the surrounding airports. The purpose of our analysis is to show how is the resilience and recovery of airports in the global public health crisis. Much research on resilience can be found in air transportation networks facing natural hazards or extreme weather, which focus on the robustness of the airport network. These methods are not suitable for the global public health event. Therefore, based on the collection of existing data, we combined with existing resilience measurement methods to analyze the resilience and recovery of airports during the global public health crisis. The resilience metrics results reflect the recovery of airports very well under different strategies. Here, we analyze airport network resilience by considering the performance-based methods. We integrate some metrics such as aircraft movements, passenger throughput, and freight throughput in the resilience metrics model, comparing the resilience evaluation under different preventive and control strategies, which can reflect the airport’s recovery speed during the COVID-19 pandemic. Our analysis indicates that the aviation system network deteriorates soon after the COVID-19 outbreak, but the recovery level of the aviation industry depends on what measures are taken to prevent and control the COVID-19 epidemic. In particular, the recovery of the aviation system network in Europe takes longer than in China, due to different prevention and control strategies for COVID-19. The study proves that the emergency response ability of the country for the public health crisis has a significant positive impact on speeding up the recovery of the aviation system.
A hybrid neural network for large-scale expressway network OD prediction based on toll data
Journal ArticleMay 2019 • PLOS ONE • 14(5) • Public Library of Science (PLoS)
Xin Fu, Hao Yang, Chenxi Liu, Jianwei Wang, Yinhai Wang*
DOI: 10.1371/journal.pone.0217241
Accurate Origin-Destination (OD) prediction is significant for effective traffic monitor, which can support operation decision in traffic planning and management field. The enclosed expressway network system like toll gates system in China can collect mounts of trip records which can be gathered for OD prediction. The paper develops a novel neural network, which is named Expressway OD Prediction Neural Network (EODPNN) for toll data-based prediction. The network consists of the following three modules: The Feature Extension Module, the Memory Module, and the Prediction Module. In the process, the attributes data which can reflect the city attribute such as GDP, population, and the number of vehicles are considered to embeded into the notwork to increase the accuracy of the model. For the applicability improvment of the model, we categorize the cities in multiple classes based on their economy and population scales in this paper, which can provide a higher accurate prediction of OD by EODPNN. The results shows that, comparing to the traditional model like ARIMA and SVM, or typical neural networks like Bidirectional Long Short-term Memory, the EODPNN delivers a better prediction performance. The method proposed in this paper has been fully verified and has a potential to transplant to the other OD data-based management systems for a more accurate and flexible prediction.
Conference Proceedings(8)
V2X-UniPool: Unifying Multimodal Perception and Knowledge Reasoning for Autonomous Driving
Conference PaperJun 2026 • Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
X. Luo, F. Yang, D. Fan, X. Gao, B. Yu, Z. Li, Z. Tu, Y. Zhou, C. Liu*
Autonomous driving (AD) has achieved significant progress, yet single-vehicle perception remains constrained by sensing range and occlusions. Vehicle-to-Everything (V2X) communication addresses these limits by enabling collaboration across vehicles and infrastructure, but it also faces heterogeneity, synchronization, and latency constraints. Language models offer strong knowledge-driven reasoning and decision-making capabilities, but they are not inherently designed to process raw sensor streams and are prone to hallucination. We propose V2X-UniPool, the first framework that unifies V2X perception with language-based reasoning for knowledge-driven AD. It transforms multimodal V2X data into structured, language-based knowledge, organizes it in a time-indexed knowledge pool for temporally consistent reasoning, and employs Retrieval-Augmented Generation (RAG) to ground decisions in real-time context. Experiments on the real-world DAIR-V2X dataset show that V2X-UniPool achieves state-of-the-art planning accuracy and safety while reducing communication cost by more than 80\%, achieving the lowest overhead among evaluated methods. These results highlight the promise of bridging V2X perception and language reasoning to advance scalable and trustworthy driving. Our code is available at: https://github.com/Xuewen2025/V2X-UniPool
LangCoop: Collaborative Driving with Language
Conference PaperJun 2025 • 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) • IEEE
Xiangbo Gao, Yuheng Wu, Rujia Wang, Chenxi Liu, Yang Zhou, Zhengzhong Tu*
DOI: 10.1109/CVPRW67362.2025.00406
Multi-agent collaboration holds great promise for enhancing the safety, reliability, and mobility of autonomous driving systems by enabling information sharing among multiple connected agents. However, existing multi-agent communication approaches are hindered by limitations of existing communication media, including high bandwidth demands, agent heterogeneity, and information loss. To address these challenges, we introduce LangCoop, a new paradigm for collaborative autonomous driving that leverages natural language as a compact yet expressive medium for interagent communication. LangCoop features two key innovations: Mixture Model Modular Chain-of-thought ($M^{3} \text{CoT}$) for structured zero-shot vision-language reasoning and Natural Language Information Packaging (LangPack) for efficiently packaging information into concise, language-based messages. Through extensive experiments conducted in the CARLA simulations, we demonstrate that LangCoop achieves a remarkable 96 % reduction in communication bandwidth ($<2 ~\text{KB}$per message) compared to image-based communication, while maintaining competitive driving performance in the closed-loop evaluation. Our project page and code are at https://xiangbogaobarry.github.io/LangCoop/.
SenseRAG: Constructing Environmental Knowledge Bases with Proactive Querying for LLM-Based Autonomous Driving
Conference PaperFeb 2025 • 2025 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) • IEEE
Xuewen Luo, Fan Ding, Fengze Yang, Yang Zhou, Junnyong Loo, Hwa Hui Tew, Chenxi Liu*
DOI: 10.1109/WACVW65960.2025.00111
This study addresses the critical need for enhanced situational awareness in autonomous driving (AD) by leveraging the contextual reasoning capabilities of large language models (LLMs). Unlike traditional perception systems that rely on rigid, label-based annotations, it integrates real-time, multimodal sensor data into a unified, LLMs-readable knowledge base, enabling LLMs to dynamically understand and respond to complex driving environments. To over-come the inherent latency and modality limitations of LLMs, a proactive Retrieval-Augmented Generation (RAG) is designed for AD, combined with a chain-of-thought prompting mechanism, ensuring rapid and context-rich under-standing. Experimental results using real-world Vehicle-to-everything (V2X) datasets demonstrate significant improvements in perception and prediction performance, highlighting the potential of this framework to enhance safety, adaptability, and decision-making in next-generation AD systems.
Traffic Performance GPT (TP-GPT): Real-Time Data Informed Intelligent ChatBot for Transportation Surveillance and Management
Conference PaperSep 2024 • 2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC) • IEEE
Bingzhang Wang, Zhiyu Cai, Muhammad Monjurul Karim, Chenxi Liu, Yinhai Wang*
DOI: 10.1109/ITSC58415.2024.10919940
The digitization of traffic sensing infrastructure has significantly accumulated an extensive traffic data warehouse, which presents unprecedented challenges for transportation analytics. The complexities associated with querying large-scale multi-table databases require specialized programming expertise and labor-intensive development. Additionally, traditional analysis methods have focused mainly on numerical data, often neglecting the semantic aspects that could enhance interpretability and understanding. Furthermore, real-time traffic data access is typically limited due to privacy concerns. To bridge this gap, the integration of Large Language Models (LLMs) into the domain of traffic management presents a transformative approach to addressing the complexities and challenges inherent in modern transportation systems. This paper proposes an intelligent online chatbot, TP-GPT, for efficient customized transportation surveillance and management empowered by a large real-time traffic database. The innovative framework leverages contextual and generative intelligence of language models to generate accurate SQL queries and natural language interpretations by employing transportation-specialized prompts, Chain-of-Thought prompting, few-shot learning, multi-agent collaboration strategy, and chat memory. Experimental study demonstrates that our approach outperforms state-of-the-art baselines such as GPT-4 and PaLM 2 on a challenging traffic-analysis benchmark TransQuery. TP-GPT would aid researchers and practitioners in real-time transportation surveillance and management in a privacy-preserving, equitable, and customizable manner.
Learning Realistic and Reactive Traffic Agents
Conference PaperJun 2024 • 2024 IEEE Intelligent Vehicles Symposium (IV) • IEEE
Meixin Zhu, Di Chen, Xinyi Yuan, Zhiwei Shang, Chenxi Liu
DOI: 10.1109/IV55156.2024.10588807
In recent years, remarkable strides have been made in the field of autonomous driving, with a particular focus on enhancing perception and prediction capabilities through the integration of big data and advanced deep learning algorithms. Despite these advancements, the persistent challenge of effectively validating the performance of autonomous vehicles (AVs) remains a critical issue. In the realm of microscopic traffic simulation, a noteworthy challenge persists – that of bridging the behavior gap between simulated scenarios and real-world driving situations. Efforts to define agent behavior in simulations manually or replay observed behaviors have proven to be inefficient and prone to inaccuracies, mainly because simulated agents often fail to authentically react to actions initiated by AVs. While rule-based traffic simulation models offer plausible behaviors, they exhibit limitations in adapting to diverse and data-driven behavior patterns within complex driving interactions. Addressing these challenges, we propose a learning-based method for traffic agent simulation, emphasizing realism and reactivity. This involves learning data-driven agent models from real-world driving data with detailed interaction information and high-definition (HD) maps. Utilizing a convolutional neural network (CNN), we extract features and predict future trajectories, achieving realism and reactivity through closed-loop simulation at inference. The proposed model is evaluated using real-world data, demonstrating its effectiveness in simulating diverse and realistic traffic behaviors, like stopping at red traffic lights, yielding to other vehicles during right-turn-on-red, and car following.
Cost-effective Vehicle Recognition System in Challenging Environment Empowered by Micro-Pulse LiDAR and Edge AI
Conference PaperJun 2024 • 2024 IEEE Intelligent Vehicles Symposium (IV) • IEEE
Junyue Jiang, Hongliang Lu, Chenxi Liu, Meixin Zhu, Yiran Chen, Hao Frank Yang*
DOI: 10.1109/IV55156.2024.10588634
Vehicle recognition and classification are critical for a number of traffic applications, e.g., traffic signal control, traffic flow modeling, tolling, and logistics optimization. Commonly used sensing systems are mainly counted on in-pavement loops or surveillance video cameras, while both of them have their inherent limitations. Leveraging micro high-speed pulse LiDAR mounted overhead of travel lanes, this study proposes Compact LiDAR Empowered Vehicle Enhancing-minority Recognition (CLEVER) system, a real-time cost-effective vehicle detection and classification framework that is empowered by edge Artificial Intelligence (AI). Based on the customized minority-enhancing vehicle classification deep neural network, the CLEVER system outperforms cutting-edge LiDAR-based vehicle classification methods up to 15.98% true-positive rate in classifying ten types of vehicles. Furthermore, by highly integrating the hardware, the pre-processing algorithm and the classification neural network into an edge computing node, the CLEVER system only consumes 10% of the cost in LiDAR systems and works perfectly in a plug-and-play mode with a negligible sub-second inference time (212ms to 459ms). The proposed CLEVER system offers an affordable end-to-end solution that can benefit traffic operators by collecting more accurate and reliable vehicle classification data streams and that can lead to a more efficient and flexible ITS.
An incremental learning-based framework for non-stationary traffic representations clustering and forecasting
Conference PaperOct 2022 • 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) • IEEE
Meng-Ju Tsai, Zhiyong Cui, Chenxi Liu, Hao Yang, Yinhai Wang*
DOI: 10.1109/ITSC55140.2022.9922113
To curb the growth of COVID-19, many rules, including a work-from-home policy, were issued in 2020. While these limits successfully prevented the virus's transmission, they completely altered original mobility patterns, resulting in considerable reductions in travel time and vehicle miles traveled. Under this non-stationary data stream, the US Department of Transportation struggled to anticipate future traffic conditions. Obviously, two essential challenges need to be addressed immediately: 1) it is challenging for transportation agencies to learn representative traffic patterns from the continually changing traffic circumstances. And 2) when and how should the forecasting model be updated to learn new patterns without forgetting previous tasks? We proposed an incremental learning-based framework for non-stationary data clustering and forecasting in transportation scenarios to tackle the issues mentioned above. It is a dual-module architecture that includes a Temporal Neighborhood Clustering module and an Incremental Learning module. The objective of the first component is to dynamically detect the optimal boundary for clustering statistically similar neighbors by enlarging both the in-group similarity and between-group dissimilarity. The second module applies the online-EWC approach, which is commonly used in image classification tasks but rarely in regression models, to learn new tasks and avoid catastrophic forgetting, which is a typical occurrence while training neural networks with multiple tasks. Experiments on the Greater Seattle Area employed loop detector data collected in 2020 yielded reliable prediction performance in both robustness and accuracy. The dual-module framework can generate promising results from pre-COVID-19 to post-COVID-19 time frames. This framework would aid government agencies and the general public in developing long-term policies and strategies for post-pandemic intelligent transportation systems.
Hybrid data-fusion model for short-term road hazardous segments identification based on the acceleration and deceleration information
Conference PaperAug 2020 • International Conference on Transportation and Development 2020 • American Society of Civil Engineers
Hao (Frank) Yang, Chenxi Liu, Meixin Zhu, Wei Sun, Yinhai Wang*
DOI: 10.1061/9780784483145.028
Road hazardous segments (HS) identification is valid and useful to improve traffic safety and save life. Traditionally, researchers always use long-term historical crash data with the simulation model to find the HS, which is always a way of being wise after the tragedy. However, this paper promotes a scheme of short-term real-time HS identification scheme based on the traffic inconstancy features. Using accumulative historical floating vehicle data to show and capture the road-segment level of spatial-temporal inconstancy features, then combined with the real-time loop detector data along the same road segments, his paper uses an innovative hybrid model called the standard deviation accident model (SDAM) to identify the three risky levels of HS, suspected, hazardous, and dangerous. A day was broken into 83-hour short-term periods, and the model is fitted by two weeks crash data and validated by 3 days, and achieve the accuracy over 73% for identifying the short-term hazard on each road segment (average length 0.826 km) of total 177 road segments. The success of the SDAM model is mainly due to the well-use of the inconstancy information captured from the long-term historical trajectory data. It was found that the inconstancy of speed and deceleration are close related to the HS identification (with a p-value less than 0.001) significantly. The features of HS road segments are summarized, and the safety improvement suggestions are provided at the end of the paper.
Patents(4)
Determining a visibility measure based on an image of an environment
PatentFeb 2024 • U.S. patent application • US20240037899A1
C. Liu, R. Ke, and Y. Wang
Traffic Sensor and System
PatentMay 2023 • U.S. patent application • US20230136925A1
Z. Pu, R. Ke, C. Liu, H. Yang, Y. Wang
Sensor mounting device and sensor module
PatentJul 2021 • Chinese utility model patent • CN215261754U
C. Liu, Z. Pu, R. Ke, Y. Zhuang, C. Shi, and H. Yang
Sensor shell and sensor module applying same
PatentFeb 2021 • Chinese utility model patent • CN212539224U
C. Liu, Z. Pu, and C. Shi