Towards Building a Foundation AI Model for Railway Safety
Foundation models, including large language models (LLMs), are revolutionizing multiple aspects of everyday life and work, with the use of LLMs such as ChatGPT being now commonplace and transforming life, work, and scientific discovery as we know it. Despite impressive performances in conversations, such systems still suffer from hallucinations and pose safety risks, which are critical for domain-specific applications, such as railway safety. In this project, the research team proposes to harness the power of such foundation models in order to transform railway safety, by building a foundation model that can support various critical tasks of railway safety practice, such as understanding and summarizing possible causes for an accident, comparing different accidents and understanding commonalities and risk factors, and coming up with policy recommendations that can improve safety in a grade crossing or a locality at large. To do so, the team outlines a number of fascinating and hard research challenges that need to be addressed, and as part of this one-year project, the team sets out to build a prototype proof-of-concept that will demonstrate the viability of foundation models for railway safety.
- Record URL:
-
Supplemental Notes:
- The University of Texas Rio Grande Valley is a partner for this project.
Language
- English
Project
- Status: Active
- Funding: $142500
-
Contract Numbers:
69A3552348340
-
Sponsor Organizations:
Office of the Assistant Secretary for Research and Technology
University Transportation Centers Program
Department of Transportation
Washington, DC United States 20590University of California, Riverside
1084 Columbia Ave.
Riverside, CA United States 92507 -
Managing Organizations:
University of California, Riverside
1084 Columbia Ave.
Riverside, CA United States 92507 -
Project Managers:
Stearns, Amy
-
Performing Organizations:
University of California, Riverside
1084 Columbia Ave.
Riverside, CA United States 92507 -
Principal Investigators:
Papalexakis, Evangelos
Chen, Jia
- Start Date: 20240601
- Expected Completion Date: 20250831
- Actual Completion Date: 0
- USDOT Program: University Transportation Centers Program
Subject/Index Terms
- TRT Terms: Artificial intelligence; Language; Prototypes; Railroad safety
- Subject Areas: Data and Information Technology; Planning and Forecasting; Railroads; Safety and Human Factors;
Filing Info
- Accession Number: 01924857
- Record Type: Research project
- Source Agency: University Transportation Center for Railway Safety
- Contract Numbers: 69A3552348340
- Files: UTC, RIP
- Created Date: Jul 22 2024 8:50AM