The Intelligent Manufacturing and Machine Vision research group has long focused on the study of industrial embodied intelligence, especially deep learning methods for intelligent manufacturing, industrial vertical domain based large models, embodied intelligent robotics, and hand-drawn sketch CAD modeling.
In the field of embodied intelligent robots, the main research focuses on human like perception and decision-making technology for mobile robots in complex dynamic indoor environments, including mobile robot learning platforms, semantic SLAM, embodied intelligence, etc. It is mainly applied in food delivery robots, cleaning robots, etc. At present, we have established a Joint Research Center for Intelligent Service Robot Technology with Purdue Technology, a unicorn enterprise in the field of distribution robots in Shenzhen. We have invested tens of millions of yuan to carry out key research and engineering application research on core technologies related to intelligent service robots.
In the field of intelligent manufacturing, in response to the significant demand and industry pain points of multi variety and small batch product assembly in the era of product personalization, the research team has proposed a cross category reconfigurable flexible assembly production line deformation design theory and technical system, widely using knowledge graphs, deep learning, virtual debugging, digital twins and other methods to achieve the flexibility, numerical control and intelligence of product assembly production lines. At present, the technology has been successfully industrialized for Guangzhou Fuwei Intelligent Technology Co., Ltd. and has been successfully applied in large state-owned enterprises.
In the field of CAD industrial software, hand drawn sketch engineering product assembly modeling technology has been proposed to address the gap between conceptual design and detailed design. This includes research tasks such as hand drawn sketch understanding, part creation, and complex product assembly based on deep learning, data mining, and other methods. Our team has independently developed the 3D modeling core iDesignCAD with the continuous support of National Natural Science Youth and general projects. At present, with the support of the national key research and development program Industrial Software, we plan to develop a large-scale multimodal dataset of hand drawn sketches, laying the data foundation for deep learning methods for hand drawn sketches, and developing hand drawn sketch CAD industrial software, laying the foundation for the industrial application of hand drawn sketch CAD industrial software.
