Prof. Sooyoung Lee delivers an invited talk titled "Physical AI for Manufacturing: A Call for Clarity and Rigor" at the International Conference on Precision Engineering and Sustainable Manufacturing (PRESM 2026).
Video Link
Abstract: Artificial intelligence (AI) is rapidly evolving beyond purely digital domains into the physical world, giving rise to what is now referred to as Physical AI. While the concept is gaining increasing attention, its definition and scope remain not yet fully established. Physical AI encompasses intelligent systems capable of perceiving, reasoning, and acting within real-world environments through integration with machines and engineering systems. Moving beyond conventional automation, these systems are advancing toward embodied intelligence, where they can learn from interaction, adapt to dynamic conditions, and continuously improve their performance in complex and unstructured environments. This talk provides a systematic overview of the current state of Physical AI, aiming to bring greater clarity and rigor to the field. The seminar also addresses critical challenges facing Physical AI in manufacturing, along with strategies to overcome these limitations. Ultimately, this talk aims to provide a clearer and more rigorous perspective on how Physical AI can be effectively defined and deployed to enable future manufacturing systems.
- Hyeongbae: Diffusion Schrödinger Bridge Framework for Path-aware Stochastic Transport in Manufacturing
- Jeongwoo: A Differentiable Physics-based Framework for Hybrid Discrete-Continuous Process Optimization
Prof. Sooyoung Lee delivers an invited talk titled "Physical AI for Manufacturing: A Call for Clarity and Rigor" at the International Conference on Precision Engineering and Sustainable Manufacturing (PRESM 2026).
Video Link
Abstract: Artificial intelligence (AI) is rapidly evolving beyond purely digital domains into the physical world, giving rise to what is now referred to as Physical AI. While the concept is gaining increasing attention, its definition and scope remain not yet fully established. Physical AI encompasses intelligent systems capable of perceiving, reasoning, and acting within real-world environments through integration with machines and engineering systems. Moving beyond conventional automation, these systems are advancing toward embodied intelligence, where they can learn from interaction, adapt to dynamic conditions, and continuously improve their performance in complex and unstructured environments. This talk provides a systematic overview of the current state of Physical AI, aiming to bring greater clarity and rigor to the field. The seminar also addresses critical challenges facing Physical AI in manufacturing, along with strategies to overcome these limitations. Ultimately, this talk aims to provide a clearer and more rigorous perspective on how Physical AI can be effectively defined and deployed to enable future manufacturing systems.
- Hyeongbae: Diffusion Schrödinger Bridge Framework for Path-aware Stochastic Transport in Manufacturing
- Jeongwoo: A Differentiable Physics-based Framework for Hybrid Discrete-Continuous Process Optimization