Ruoli Dai
Founder & CEO, Noitom Robotics
Dr. Ruoli Dai received his bachelor’s degree from the University of Science and Technology of China and his Ph.D. from The Chinese University of Hong Kong. He has been recognized as a Distinguished Alumnus by both the Department of Mechanical and Automation Engineering and the Faculty of Engineering at CUHK. Dr. Dai is the founder and CEO of Noitom Robotics and co-founder of Noitom Ltd. He has long been dedicated to human motion digitization, human–computer interaction, and robot learning. Holding more than 50 international patents, he has led teams in building a comprehensive motion-capture technology stack spanning inertial sensing, optical systems, and multi-sensor fusion. Under his leadership, the company has delivered tens of thousands of professional systems to customers in more than 50 countries and regions, at one point capturing approximately 70% of the global professional motion-capture market. In 2025, Dr. Dai founded Noitom Robotics, extending more than a decade of expertise in human digitization into data infrastructure for embodied intelligence. The company focuses on multimodal human behavior data, robot teleoperation, and scalable data production, with the goal of transforming human interaction and behavior into machine-learnable data. Since its establishment, the company has completed multiple financing rounds, raising more than RMB 1 billion in aggregate. Since 2024, Dr. Dai has introduced the concept of building “large-scale embodied intelligence data factories” in a number of public talks and has actively pursued its implementation. In 2026, he led his team in launching ModalityNet, an embodied intelligence data platform, and subsequently advanced the open-sourcing of related datasets. His work aims to make the real world learnable and establish human interaction data as a key training resource for robotics foundation models.
Topic
From Data Factory to World Compiler
Embodied intelligence is facing a challenge far more complex than that of language models: how can robots understand the real world and continuously learn from human behavior and real-world interactions? This talk will explore the data challenges of embodied intelligence, sharing Noitom Robotics' practical experience and insights—from building a large-scale “Data Factory” to exploring the concept of a “World Compiler.” Drawing on real-world practices in data collection, human motion and human-object interaction, multimodal data production, data standardization, and robot learning, the talk will examine what kinds of data embodied intelligence requires, and why data infrastructure is evolving beyond simply pursuing scale toward structured representation and compilation of the real world. Agenda From Motion Capture to Data Infrastructure for Embodied Intelligence Building a Data Pyramid for Embodied Intelligence from the First Principles of Biomimicry Data Factory: Scaling Machine-Learnable Real-World Data Production From Data Factory to World Compiler Key Takeaways Understand the fundamental differences between Physical AI and traditional AI at the data level, and why real-world data is becoming critical infrastructure for the development of robotic intelligence. Develop a comprehensive understanding of embodied intelligence data systems, covering the full pipeline from data collection, synchronization, structuring, and quality control to model training and robot deployment. Learn about the latest evolution in embodied intelligence data production, including how Data Factories are moving beyond large-scale collection toward higher-quality data, richer modalities, and broader scenario coverage. Understand the concept of the “World Compiler”: how continuous and complex physical reality can be transformed into data representations that machines can understand and learn from, and how this direction may shape future paradigms of robot learning. Gain first-hand insights from real-world industry practice, including the challenges currently being addressed across the journey from data to models and from laboratory environments to real-world robot deployment, as well as emerging technical directions to watch.
