Development and Practical Application of an Analysis Process Utilizing Surrogate AI

KAZUTO UEHARA (AISIN)

In the fluid analysis of an electric water pump, the introduction of surrogate AI on the upstream side reduced the number of CAE computations and improved the efficiency of the analysis process. This report presents the practical application and effectiveness of AI technology in the analysis process, including case studies aimed at improving prediction accuracy.

AI-Driven Vehicle Cyber Risk Modeling for Comprehensive Cybersecurity Management Systems

Eugene Shubov (CTO)

In response to the increasing threats and complexity of vehicle architecture, we introduce an AI-driven approach for continuous vulnerability and threat monitoring and risk assessment. This approach utilizes AI to automate the construction of cyber risk models based on the vehicle architecture. Additionally, it facilitates CSMS compliance by optimally defining mandatory controls, cybersecurity tests, and IDS detection use cases. Furthermore, it helps in assessing vulnerabilities to ensure homologation approval for SUMS in accordance with UNR 155/156 and GB 44495/44496.

Improvement of End-of-Life Vehicle Chassis Number Reading Software Using AI-OCR

Masayoshi Nakamura (National Institute of Technology (KOSEN), Akashi College)･Shigeya Ikebo (Nagoya Bunri Univ.)･Satoru Yaseda･Hitoshi Yamasaki (Aratani Shoukai)

In order to comply with the resource recovery incentive system scheduled to be implemented from 2026, we developed VIN number reading software using AI-OCR two years ago. When the VIN numbers of 1,000 vehicles were read using the software, the reading accuracy was 64%. In this research, we worked to improve the reading accuracy of the software and have achieved some results, so we report here.