Subaru Introduces Neural Concept to Revolutionize Automotive Development
.jpg)
- In their effort to accelerate the development of next-generation vehicles, Subaru Corporation, through our partner Cybernet Systems Corporation, evaluated Neural Concept’s predictive capabilities. NC successfully reduced the analysis time for die face shape design from 3 hours to just 2 minutes, while maintaining high accuracy, comparable to traditional CAE tools.
- Based on these impressive results, Subaru integrated NC into their development process. This deep learning-based approach allows Subaru to explore a wide range of designs quickly and accurately, significantly shortening the development period and enhancing product performance.
- Thanks to this innovation by NC, Subaru will bring the next generation of vehicles to market more efficiently by 2028.
The Drive to Shorten Development Timelines
Subaru aims to begin mass production of a diverse range of vehicles by 2028, including the next-generation e-BOXER and BEV models. Achieving this ambitious goal necessitates a significant reduction in the development time for new models. Currently, forming analysis in the stamping area using conventional CAE tools is a time-consuming process, taking up to 48 hours for analysis and one to two weeks for feedback. This delay is a bottleneck in the design process, hindering the swift development of safe and efficient automobiles.
Breakthrough in Forming Analysis with Neural Concept
To address these challenges, Subaru evaluated Neural Concept (NC) ability to predict the thickness reduction rate of blank material required for die face shape design. The results were remarkable: NC reduced the analysis time from 3 hours to just 2 minutes while maintaining prediction accuracy comparable to traditional CAE tools. This dramatic improvement is expected to significantly cut down the man-hours needed for the entire development process.
Expert Insights and Future Prospects
The design team at Subaru's Production Engineering Department expressed high expectations for NC’s 3D AI technology. Conventional AI tools fell short in understanding complex 3D shapes, making them unsuitable for CAE tasks like press molding. However, NC demonstrated strong capability in comprehending and simplifying 3D shape features, leading to faster and more accurate forming predictions.According to the Design Representative of Subaru's Car Production Technology Department, Neural Concept "has a very high capability for capturing 3D shape features, excelling in accurately reading shape features and simplifying feature quantities. [...] we found that NC allows for much faster forming predictions than traditional CAE, while also delivering sufficiently accurate results."
Future Integration and Expansion
Based on these successful demonstration tests, Subaru decided to fully integrate NC into their development workflow. Moving forward, Subaru and Cybernet plan to expand the scope of NC predictions to achieve CAE-level performance across the entire development process. This collaboration aims to continuously enhance the technology, ultimately contributing to a significant reduction in the overall development timeline for Subaru's vehicles.
Read more about the project in the Press Release published by Cybernet Systems Corporation, original version in Japanese or the English translation here.
Interested? Get in touch with the NC team to learn more.
