STANDARDS OF ARTIFICIAL INTELLIGENCE IN ROAD TRANSPORT
Authors:
1.
Voronezh State University of Forestry and Technologies named after G.F. Morozov
Russian Federation
Russian Federation
2.
Voronezh State University of Forestry and Technologies named after G.F. Morozov
(department of Computer Technology and Microelectronic Engineering, docent)
from 01.01.1921 until now Russian Federation
from 01.01.1921 until now Russian Federation
3.
Voronezh State University of Forestry and Technologies named after G.F. Morozov
Type:
Сonference article
Pages:
from 651
to 654
Published:
30.06.2026
Subject area:
UDC 681.5.015
Language:
Russian
Keywords:
artificial intelligence, autonomous vehicles, functional safety, ISO 26262, SOTIF, ISO PAS 8800, deep learning, risk management
Abstract:
The article analyzes modern approaches to standardization of artificial intelligence systems in the automotive industry. Key functional safety standards ISO 26262, Safety of the Intended Functionality SOTIF (ISO 21448) and the prospective ISO/PAS 8800 standard focused directly on artificial intelligence systems are considered. Limitations of existing standards when applied to non-deterministic machine learning models are identified and ways to overcome them through a combined risk management approach are proposed. Special attention is paid to verification and validation challenges of deep neural networks in the context of autonomous vehicle safety assurance.
The article analyzes modern approaches to standardization of artificial intelligence systems in the automotive industry. Key functional safety standards ISO 26262, Safety of the Intended Functionality SOTIF (ISO 21448) and the prospective ISO/PAS 8800 standard focused directly on artificial intelligence systems are considered. Limitations of existing standards when applied to non-deterministic machine learning models are identified and ways to overcome them through a combined risk management approach are proposed. Special attention is paid to verification and validation challenges of deep neural networks in the context of autonomous vehicle safety assurance.
Keywords:
artificial intelligence, autonomous vehicles, functional safety, ISO 26262, SOTIF, ISO PAS 8800, deep learning, risk management
artificial intelligence, autonomous vehicles, functional safety, ISO 26262, SOTIF, ISO PAS 8800, deep learning, risk management
1. Vysockaya I.A. Poisk mnozhestva dopustimyh upravlencheskih resheniy v tehnicheskih sistemah // Modelirovanie sistem i processov. – 2024. – T. 17, № 1. – S. 35–42. – DOIhttps://doi.org/10.12737/2219-0767-2024-17-1-35-42.
2. Mirzarazi F., Danishvar S., Mousavi A. The Safety Risks of AI-Driven Solutions in Autonomous Road Vehicles // World Electr. Veh. J. – 2024. – Vol. 15. – P. 438. DOI: https://doi.org/10.3390/wevj15100438
3. Yu H., Hulse D., Irshad L. Understanding AI Risks in Transportation // AI Assurance for Transportation Whitepaper Series. – U.S. DOT HASS COE, NASA SWS, 2024.
