In today’s era where AI technology is accelerating its penetration across industries, the question of how frontier technologies can truly serve the real economy has become a focal point of attention. Dr. Yintang Dai, Chief Scientist of the CASMT AI team, leveraging two decades of academic expertise and industrial practice in the field of artificial intelligence, has led his team to successfully develop the "CASMT AI Flexible Automation Design Large Model," providing breakthrough solutions for intelligent manufacturing and industrial design. This scientist, who integrates top-tier academic background with practical industry experience, is advancing an “integration of research, education, and application” innovation model, driving AI technologies from the laboratory to large-scale application.

Technological Breakthrough: Industrial Practice of the Flexible Automation Design Large Model

As one of the core achievements of the CASMT AI team, the "CASMT AI Flexible Automation Design Large Model" overcomes the limitations of traditional industrial design software characterized by “single functions and fixed processes.” By integrating natural language processing, deep learning, and knowledge graph technologies, the model realizes intelligent generation from requirement texts to design solutions, supporting multi-scenario and cross-domain automation design needs.

“Traditional industrial design requires engineers to manually write parameters and debug models, whereas our system can understand natural language descriptions of design requirements and automatically generate 3D models and process solutions that meet engineering standards,” explained Dr. Dai. This innovation stems from his deep insight into industry pain points: during intelligent manufacturing upgrades, enterprises often face challenges such as long design cycles, high customization costs, and shortages of specialized talent.

The technical team has built a large-scale industry knowledge base and, by combining self-supervised learning with reinforcement learning algorithms, has enabled the model to possess a closed-loop ability of “learning–reasoning–optimization.”

Integration of Research and Application: From the Laboratory to Industry

Dr. Dai’s career has always been driven by a persistent pursuit of “technology implementation.” After joining CASMT, he promoted the establishment of an innovation cycle of “demand insight – technology R&D – scenario verification – iterative optimization.” During the development of the CASMT AI large model, the team co-established joint laboratories with leading enterprises in the electronic information and other sectors. Leveraging CASMT’s strong design resource base, they collected millions of high-quality training data points from over 100 production lines and tens of thousands of design requirement data sets, ensuring that the model output aligns with engineering realities.

Ecosystem Building: Cultivating Innovative Talents in the AI Era

Dr. Dai has always emphasized knowledge transfer and team development. At CASMT, he introduced a “mentorship system,” requiring technical experts to undertake both R&D and training responsibilities, thereby cultivating interdisciplinary talents proficient in both algorithms and processes.

“The development of artificial intelligence requires a two-way collaboration between academia and industry,” Dr. Dai noted. The team is collaborating with multiple universities to establish “intelligent design joint laboratories,” opening part of the desensitized industry data for teaching and research while offering outstanding students opportunities to participate in real-world projects. This open collaboration model has already attracted more than 20 members, including 5 PhDs and 6 Masters graduates, to join the team.

Future Outlook: Making AI a Universal Tool for Industrial Upgrading

In the face of global competition driven by large model technologies, our goal is not to replace human labor with AI, but rather to free human resources from tedious and repetitive design tasks, enabling them to truly focus on higher-level research and development in manufacturing technologies.

The team’s next step is to optimize the lightweight deployment capability of the model to reduce the application threshold for small and medium-sized enterprises, while also exploring integration paths with technologies such as digital twins and the industrial metaverse. Currently, the team has also launched the research and development of the CAX AI unified platform, which integrates design, simulation, and manufacturing.

The ultimate value of artificial intelligence does not lie in the size of its parameters, but in whether it can truly solve real problems in human society. Leveraging its multimodal, domain-specific industrial large model technology, the CASMT AI Flexible Automation Design Large Model deeply empowers intelligent manufacturing, driving efficiency improvements and model innovation across the industry. At the same time, through data-driven decision-making and flexible automation design, it reshapes the industrial ecosystem and becomes a core engine for upgrading intelligent manufacturing.