Large models have become the core engine propelling the rapid advancement of smart manufacturing, reshaping every corner of industrial production. According to IDC's report titled "Progress and Prospects of Industrial Large Model Applications," industrial large models have been widely applied in several typical industrial production scenarios, giving rise to a series of powerful and highly specialised single-point application functions. However, the common challenge faced by industries worldwide is how to integrate these single-point functions into the entire production workflow, forming an ecosystem with ecological effects in large model systems.
As a continuous innovator in the field of smart manufacturing, SmartMore is committed to becoming a globally-serving, all-scenario intelligent company, focusing on innovative technology research and development and practical application. In 2023, SmartMore independently developed the world's first industrial multi-modal large model, IndustryGPT, introducing artificial intelligence technology into traditional tracks and becoming the first enterprise to implement large models in industrial applications.
Over the past year, SmartMore has not only accelerated the commercialisation process of large models but has also continuously led innovative explorations, consistently developing a large model product matrix based on IndustryGPT. This matrix introduces the "super brain" into every industrial production scenario, significantly enhancing the dynamism of smart manufacturing from all angles, leading technological innovation and propelling the development of new quality productivity.
Core Large Model Engine
Multi-modal and Universal Industrial "Super Brain"
The core of smart manufacturing is "intelligence," and large model technology is the key to continuous evolution of intelligent attributes. SmartMore's independently developed industrial large model, IndustryGPT, positioned as a generative AI engine for industrial application scenarios, serves as the "super brain" for industrial intelligence and digitisation. It deeply integrates industrial proprietary data, deep learning algorithms, and advanced large language model technology, empowering research and development design, production manufacturing, management services, and various other processes.
Addressing industry pain points such as complex manufacturing processes, high specialisation, and high learning thresholds, IndustryGPT, with its robust computing and learning capabilities, has integrated key functions such as industrial knowledge question answering, AIGC defect sample generation, and language reasoning. These innovations have been applied in knowledge management, product assistants, process parameter analysis, defect detection, task delegation, and various other scenarios.
Based on an advanced model design architecture, IndustryGPT can be efficiently integrated with existing systems, simplifying deployment processes and quickly merging digitisation, automation, and other technologies with real production scenarios. This integration aids traditional manufacturing enterprises in revamping production methods, continuously reducing costs and increasing efficiency. This year, IndustryGPT has been upgraded to version 2.0, expanding its application scope to the field of embodied intelligence, enabling real-time analysis of voice commands, visual images, and video data to more precisely control robotic devices such as robotic arms, promoting flexible production and human-machine collaboration.
Integration of Software and Hardware
Empowering Product Intelligence Across the Spectrum
SmartMore not only pursues disruptive technologies but also focuses on the generalisation and popularisation of AI achievements. Through the innovative linkage mechanism of large models, SmartMore has successfully embedded AI Agents in software and hardware carriers, enhancing the level of intelligence of the entire product line.
IndustryGPT drives ViMo for model training
In practical production, IndustryGPT and ViMo work closely together, significantly simplifying complex processes such as model training and tuning. Users only need to provide natural language and instructions, and IndustryGPT can autonomously learn technical documents, understand the development process of machine learning models, and drive ViMo to complete data labelling, tuning, and training. Leveraging the capabilities of large models, ViMo has achieved innovative functions such as AIGC defect generation, intelligent labelling, and exclusive AutoML automatic tuning training, further reducing the operational complexity and learning costs for non-professional users.
Smart Sensors empowered by large models
Similarly, the integration of large models has brought comprehensive innovation to hardware products. Taking the SmartMore VS1000P intelligent code reader, which features a large model core, as an example, its decoding algorithm employs cutting-edge deep learning technology. By learning from a massive number of barcode images generated by AIGC, it can produce barcode images with various styles, high controllability, and high simulation fidelity. With the powerful learning capabilities derived from large model training, intelligent sensors refine efficient on-device basic inference models, swiftly identifying subtle features in complex scenarios. With the aid of small sample visual cues, these sensors continuously learn new scenes and targets, elevating the level of intelligence in industrial applications.
Highly Advanced Integrated Equipment
Meeting the High-End Manufacturing 'Necessities'
Moreover, SmartMore organically combines large models with mechanical products. As a star product embodying SmartMore's full-stack technical capabilities, the AI-AOI equipment continues to achieve innovation breakthroughs, offering efficient one-stop solutions for smart manufacturing across various industries.
5-Axis AI-AOI Cosmetic Inspection Machine
AI-AOI equipment deeply integrates applications from multiple industries such as 3C, new energy, automotive, and precision industries. Based on an understanding and adaptation to different scenarios, its powerful intelligent core achieves more intelligent, efficient, and cost-effective quality inspection and dimensional measurement to meet diverse industry requirements. Simultaneously, the large model comprehensively empowers the entire delivery process of AI-AOI equipment, from intelligent labelling and defect data generation to industry detection model applications, and to the collaborative control of generic embodied intelligent agents. Each step undergoes intelligent upgrade and transformation, significantly enhancing production efficiency, significantly reducing manual intervention costs, and making the entire manufacturing process smoother and more efficient.
It is worth mentioning that AI-AOI smart manufacturing equipment optimises hardware from an artificial intelligence perspective. Through software-hardware coordination, the equipment has significantly improved in terms of cost, speed, universality, and intelligence. This innovative design gives the AI-AOI equipment a more competitive edge in high-end manufacturing, better meeting diverse customer needs.
As a continuous innovator in the field of smartmanufacturing, SmartMore, with large models as its technological core, empowers industry rejuvenation, creating industrial large models and a series of scene-based applications, driving the digital transformation of global manufacturing industries. Currently, products and solutions incorporating SmartMore's large models have been mass-delivered, assisting global customers in inspecting billions of product batches, helping them improve quality, reduce costs, and increase efficiency. In the future, SmartMore will continue to intensify technological innovation and expand product deployment scenarios, continually delving deeper to create greater value for global industrial innovation.
