Ensuring Safety in an AI-Enhanced PTC System

Kabra, Aditi; Mitsch, Stefan; Platzer, Andre; Budnik, Christof; Nagaraja, Parinitha

i

Ensuring Safety in an AI-Enhanced PTC System

2026-08-25
By Kabra, Aditi ; Mitsch, Stefan ; Platzer, Andre

File Language:

English

Details

Creators:

Kabra, Aditi ; Mitsch, Stefan ; Platzer, Andre ; Budnik, Christof ; Nagaraja, Parinitha
Corporate Creators:

Carnegie Mellon University. School of Computer Science
Corporate Contributors:

United States. Department of Transportation. Federal Railroad Administration. Office of Research, Development and Technology
Subject/TRT Terms:
Publication/ Report Number:

DOT/FRA/ORD-26/18
Resource Type:

Tech Report
Geographical Coverage:

United States
Edition:

Final Report
Contracting Officer:

Withers, Jared
Corporate Publisher:

United States. Department of Transportation. Federal Railroad Administration
Abstract:

Embedded software for train control is safety-critical because errors can have disastrous consequences. To ensure the safety of controllers, formal verification with computer-checked, repeatable mathematical proofs presents a particularly trustworthy method for controller design. The Federal Railroad Administration contracted a research team from Carnegie Mellon University to develop a provably safe, machine-learning based, predictive train control algorithm to control acceleration and braking along rail tracks. This research was conducted from January 2021 to December 2023. The team used formal verification as a tool to design and verify a symbolic train controller, and a testing-based approach to fill in appropriate values for the symbolic parameters of the formal model. The formalization addresses complex dynamics with transcendental arithmetic, competing forces with subtle interaction, and effects whose exact magnitude is unknown at proof time.
Format:

PDF
Download URL:

https://rosap.ntl.bts.gov/view/dot/93169/dot_93169_DS1.pdf
File Type:

[PDF - 5.52 MB]
Collection(s):

Federal Railroad Administration
Main Document Checksum:

urn:sha-512:27b365ce0bb5c2b9caeb53a641121f155751abf5a7c1879c7dafb666e6515506db863a745a6efb3683125a597f99f1499a035fa96356214021b02eda549e2f49

File Language:

English

ON THIS PAGE

Details

Found an issue?

Send us an email at:

publishinghd@cdc.gov

ROSA P serves as an archival repository of USDOT-published products including scientific findings, journal articles, guidelines, recommendations, or other information authored or co-authored by USDOT or funded partners. As a repository, ROSA P retains documents in their original published format to ensure public access to scientific information.