Physical AI drives shift toward universal robot intelligence and edge computing

The manufacturing and robotics sectors are undergoing a significant shift toward Physical AI, focusing on the digitization of human expertise and the development of universal robot intelligence. Experts emphasize the need to transform the ‘tacit knowledge’ of master craftsmen—years of hands-on experience—into structured digital data that AI can utilize to maintain national manufacturing competitiveness.

Technological advancements are moving toward Robot Foundation Models (RFM), which allow different types of robots, such as humanoids and robotic arms, to share learned experiences. Companies like NVIDIA and Google DeepMind are leading efforts in creating these shared intelligence frameworks. Simultaneously, there is an intense global race to secure ‘experience data.’ While China utilizes large-scale human labor for teleoperation, companies like NC AI are using digital twins and virtual environments to simulate complex scenarios for robot training.

Furthermore, the industry is seeing a transition from cloud-based AI to Edge AI. This shift allows AI models to run directly on-site in factories and warehouses, reducing latency and enhancing data security. Software providers are developing specialized platforms to optimize, deploy, and manage these models across diverse industrial hardware, ensuring that generative AI and vision-language models can function effectively in real-world manufacturing environments.