Toward Smarter Mobility: AI-Powered Safety Insights for AVs and Vulnerable Road Users

This project investigates the safety dynamics between autonomous vehicles (AVs) and vulnerable road users (VRUs)—including pedestrians, cyclists, and e-scooter riders—by applying advanced artificial intelligence (AI) and data fusion (DF) methods to high-resolution, real-world datasets. By understanding how AVs interact with diverse VRUs in complex urban environments, this project will generate critical insights into where and how conflicts occur, what environmental factors contribute to unsafe conditions, and how different VRUs respond to perceived threats. These findings will inform safety improvements that reduce crashes and injuries, leading to significant public health benefits such as fewer hospitalizations, reduced long-term disabilities, and lower healthcare costs.  Moreover, safer streets will encourage more pedestrian and active transportation activity, promoting healthier lifestyles and improving community well-being. As AVs become more integrated into urban mobility systems, their potential to provide efficient, reliable, and low-stress transportation will further enhance public health by reducing traffic congestion, energy use, and travel-related stress.  The first dataset used for this study  Argoverse 2 3D Tracking, captures interactions between AVs and pedestrians/cyclists in Austin, Texas. The second dataset includes sensor data collected from e-scooters in San Antonio, Texas, by ScooterLab at the University of Texas at San Antonio.  The project will begin by training an AI model on the Argoverse data to identify close encounters (e.g., within 2 meters), spatially aggregate them to locate dense near-miss zones, and analyze built environment features and vehicle movement characteristics. Next, the e-scooter data will be used to detect abrupt rider responses—such as hard braking, sudden acceleration, or sharp turning—using anomaly detection algorithms, which signal perceived or actual hazards. These data streams will be fused to perform a comparative analysis across different VRU types, enabling researchers to identify common risk patterns and mode-specific vulnerabilities.   This project will advance the scientific understanding of AV-VRU safety interactions, and discuss how mobility and efficiency can be co-optimized with safety. It will lay the groundwork for future research and transportation interventions that support healthier, safer, and more efficient communities. 

Language

• English

Project

• Status: Active
• Funding: $75,000.00
• Contract Numbers:

  69A3552348329

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Center for Efficient Mobility

  1111 Rellis Parkway
  Bryan, Texas  United States  77807
• Project Managers:

  Ocon, Monica

• Performing Organizations:

  Texas A&M Transportation Institute

  Texas A&M University System
  3135 TAMU
  College Station, TX  United States  77843-3135
• Principal Investigators:

  Geedipally, Srinivas

  Bai, Shunhua

• Start Date: 20260301
• Expected Completion Date: 20261231
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers
• Source Data: 03-04-TTI

Subject/Index Terms

• TRT Terms: Autonomous vehicles; Bicycle safety; Datasets; Pedestrian safety; Pedestrian vehicle interface; Traffic conflicts; Vulnerable road users
• Subject Areas: Data and Information Technology; Operations and Traffic Management; Pedestrians and Bicyclists; Safety and Human Factors;

Filing Info

• Accession Number: 01979444
• Record Type: Research project
• Source Agency: Center for Advancing Research in Transportation Emissions, Energy, and Health (CARTEEH)
• Contract Numbers: 69A3552348329
• Files: UTC, RIP
• Created Date: Feb 12 2026 3:28PM