News

  • Our latest research on crash risk measure for SVROR crashes is published on Analytic Methods in Accident Research (IF=12.9)!

    2024-04-22

    Our team introduces a groundbreaking study on modeling the risk of single-vehicle run-off-road (SVROR) crashes using connected vehicle data. This study proposes a SVROR crash risk measure (SVROR-CRM) based on the concept of tetraquark in particle physics to model SVROR crash risks. It utilizes high-resolution connected vehicle data, which is getting increasingly available, to mathematically model the interactions between a single vehicle and the roadway. The proposed SVROR-CRM is validated using connected vehicle data (over 2 million trajectories) and SVROR crash data from sixteen horizontal curves on I-80 in Wyoming. The validation results suggest that the risk estimates generated by SVROR-CRM match the observed SVROR crashes very well. Also, the validation study shows the importance of considering both position deviation risk and attitude deviation risk. The proposed approach bridges an important gap in crash risk measure research and can be used to identify unsafe trajectories and high-crash locations and/or periods to SVROR crashes on highway horizontal curves. For more details, please refer to the DOI link: https://doi.org/10.1016/j.amar.2024.100333.
    Image Alt Text

  • Our latest research on AV longitudinal control modeling and safety evaluation is published on Accident Analysis and Prevention (IF=5.9)!

    2024-01-20

    Our team unveils a groundbreaking study on AV safety and stability, emphasizing the critical role of damping characteristics in longitudinal control. By integrating an Adaptive Cruise Control model with damping behavior analysis, our research highlights the importance of prioritizing safety alongside stability in AV operations. Published findings in Accident Analysis and Prevention shed light on optimal control parameters for enhanced safety in autonomous vehicle systems.
    Image Alt Text

  • Our latest research on nighttime image enhancement is published on IEEE T-ITS(IF=8.5)!

    2023-11-07

    Our team introduces a groundbreaking approach in vehicle detection for Intelligent Transportation Systems. Addressing the challenge of nighttime conditions, we present a novel framework, utilizing an illumination-adjustable GAN (IA-GAN) for all-day vehicle detection. By transforming daytime images into diverse nighttime scenarios, our framework, coupled with the Day-Night Balanced EfficientDet (DNBED) detector, showcases superior performance in vehicle detection across all lighting conditions. The privacy-sanitized dataset and model are open-sourced, revolutionizing the field of surveillance video analysis.
    Image Alt Text

  • Our latest research on pedestrian crossing intention prediction is published on IEEE T-ITS(IF=8.5)!

    2023-09-22

    Our team pioneers pedestrian crossing intention prediction using surveillance cameras, enhancing pedestrian safety with over-the-horizon warnings. Introducing an innovative learning framework with a pedestrian-centric environment graph and Graph Convolutional Network (GCN) technology, our model excels in capturing visual variations and spatiotemporal relationships for accurate intention prediction. By surpassing existing methods, our framework showcases superior performance and sets a new standard in pedestrian safety research.
    Image Alt Text

  • Our latest research on traffic participant detection is published on IEEE T-ITS(IF=8.5)!

    2023-08-22

    Our team introduces a game-changing solution in urban mixed traffic participant detection. Leveraging the SEU_PML dataset and the innovative YOLO SOD detector, we address the challenge of detecting small objects with increased accuracy and efficiency. Discover our cutting-edge approach and access the SEU_PML dataset at https://github.com/vvgoder/SEU_PML_Dataset.
    Image Alt Text