Using Artificial Intelligence to Uncover How Safety Perception Influences Travel Behavior Shifts: Comparative & Longitudinal Analysis for the Future of Autonomous Vehicle, Transit and Ride-hailing Services
Transit agencies and cities are increasingly overwhelmed by large volumes of unstructured data; yet they lack methodical, validated tools to turn safety narratives into operational indicators. This project addresses that gap by measuring and comparing public safety perception for autonomous-vehicle services (robotaxis), public transit, and ride-hailing services. It will assess how these perceptions relate to traveler profiles and mode choice in San Francisco and San Jose over a six-month period. San Francisco as a mature setting where robotaxis may compete with ride-hailing and transit, and San Jose as a newer coming deployment that provides a baseline for comparison and forward-looking extrapolation. The research team will use artificial intelligence with human-audited classification to analyze public discourse drawn from news-comment threads and social-media posts, for example, discussions of disengagements, curb conflicts, yielding behavior, and interpersonal harm such as unwanted contact, theft, or assault. Validation will include human audit with inter-rater reliability (aiming for Cohen’s kappa of at least 0.60), time- and city-based cross-validation, and an error taxonomy with documented adjustments. The project will deliver (1) a transparent safety-perception taxonomy, (2) traveler-persona profiles linked to safety perceptions, (3) a lightweight dashboard for agencies and cities to explore time, place, and topic trends, and (4) operational and policy frameworks for improvements across all modes, organized into vehicle-level safety measures, station and hub operating practices, reporting and response mechanisms, and rider communication standards. The approach and workflow are replicable and can be extended to additional cities. The innovation lies in a reusable tool bridging research and practice providing concrete, methodical steps to turn qualitative narratives into consistent indicators they can trust. Agencies can adopt it to sort and prioritize incoming signals, rerun it with new data, and compare results across time and places to support day-to-day decisions and longer-term planning.
Language
- English
Project
- Status: Active
- Funding: $73,703.00
-
Contract Numbers:
69A3552348323
-
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:
2400 6th Street, NW
Washington, DC United States 20059 -
Project Managers:
Bruner, Britain
-
Performing Organizations:
1 Washington Sq
San Jose, California United States 95192 -
Principal Investigators:
Cornet, Henriette
- Start Date: 20260202
- Expected Completion Date: 20201231
- Actual Completion Date: 0
- USDOT Program: University Transportation Centers Program
Subject/Index Terms
- TRT Terms: Artificial intelligence; Autonomous vehicles; Data analysis; Public opinion; Public transit; Ridesourcing; Safety; Taxi services; Travel behavior
- Geographic Terms: San Francisco (California); San Jose (California)
- Subject Areas: Data and Information Technology; Highways; Passenger Transportation; Planning and Forecasting; Public Transportation; Safety and Human Factors; Vehicles and Equipment;
Filing Info
- Accession Number: 01976550
- Record Type: Research project
- Source Agency: Research and Education for Promoting Safety (REPS) University Transportation Center
- Contract Numbers: 69A3552348323
- Files: UTC, RIP
- Created Date: Jan 19 2026 4:09PM