Professor Evgeny Burnaev, the director of the Skoltech AI and Applied AI centers, delivered a presentation at the Demo Day of the “Rocket and Space Industry” Industrial Competence Center. His talk focused on contemporary trends and methods in generative design, including the use of multi-agent systems and generative models to address complex engineering challenges.
Evgeny highlighted the growing application of generative models, including GAN-based algorithms (Generative Adversarial Networks), across various industries. He specifically mentioned projects, where his team developed generative design systems for the aerodynamic contours of aircraft and ships. These models significantly accelerated the design process and improved the accuracy of product characteristic predictions. Special emphasis was placed on the advantages of multi-agent systems, which enable the automation of designing complex engineering objects such as deposits and engineering structures. Evgeny provided an example of automating reservoir model construction, where the use of a multi-agent system reduced model setup time from several months to a few weeks, while simultaneously increasing the accuracy of production forecasts.
The potential application of generative models in processing engineering drawings was also discussed, particularly for vectorizing old paper documents. Generative models assist in cleaning drawings, restoring missing elements, and approximating lines, significantly enhancing the quality and accuracy of the final digital document. In conclusion, Evgeny Burnaev emphasized that large language models and knowledge graphs could play a crucial role in automating the processing of engineering documentation and integrating data from various sources. He expressed confidence that the continued development of these technologies would unlock new opportunities for automating and optimizing design processes in high-tech industries.