“Scaling Physical AI for 99.9% Reliable Warehouse Automation”
Recent breakthroughs in foundation models and Physical AI have unlocked unprecedented robotic capabilities. While viral demos of humanoids hint at general-purpose machines entering our homes in the near future, the true frontier of production-grade, customer-ready intelligent robotics lies today in pick-and-place robots for warehouse automation.
In this talk, I will share experiences from Nomagic’s journey in deploying fast, robust Physical AI at scale. We will look beyond the hype to discuss the reality of efficiently handling not only typical situations but also the long tail of edge cases—the critical factor distinguishing a working demo from production reality. I will share insights into how recent advances complement established approaches to enable the 99.9% reliability required to drive real-world operations and concrete ROI.
Bio
Senior Machine Learning Manager at Nomagic, where he leads development of physical AI for pick-and-place robots in warehouses. He specializes in building and deploying solutions at the intersection of machine learning, traditional software engineering, and hardware, enabling machines to interact with the physical world in real time. He holds a degree in Computer Science from TU Delft.