"ROI is key for AI to take hold in manufacturing."

At the seminar 'Physical AI, a Game-Changer that Transcends the Limits of Robots' held on the 11th and hosted by Safetics, Seo Hyeong-ju, CTO of Carbon Six, announced the 'world's first industrial Physical AI Kit'.

CTO Seo Hyung-joo stated, "Manufacturing has long been evolving through innovative technologies like optical inspection, specialized welding, and suction pad transport. However, the recent trend is moving away from 'humanoid robots' that simply mimic humans, and toward solving industrial problems through the combination of creative technologies and AI." He added, "I am skeptical of the argument that we must unconditionally move toward humanoid robots. He emphasized, “True innovation is possible only when existing engineering achievements and AI intelligence meet.”

He continued, “For physical AI to be fully adopted into the manufacturing industry, three ROI conditions must be met: cost, tact time, and yield.” He added, “Mere technical feasibility is not enough; actual investment value must be proven.”

To this end, companies are developing standardized robot software and customizable systems that can be applied to various industrial sites, it said.

In this regard, CTO Seo Hyeong-ju unveiled a demo, focusing on solving practical problems in the manufacturing industry.

The vinyl stripping task demonstrated that the industrial robot arm delicately removed the vinyl and was able to adapt in real time to irregular tasks, and the ring hanging task in the painting process demonstrated that the robot could accurately hang the ring in an irregular position after only 100 learning sessions.

Additionally, the film attachment process in the smartphone assembly process was completed reliably even after repeated abuse using a suction gripper.

These examples demonstrate the potential for robots to perform human-like tasks through intelligent adaptation and real-time judgment, rather than simple automation.

Physical AI solutions are not limited to specific robots. They can be applied to a variety of equipment, including existing gantry and SCARA robots, and can be combined with rule-based systems and AI intelligence to achieve optimal efficiency.

It is also designed to provide a UI/UX that is easy to use even for non-experts, so that field engineers can utilize it without any advanced knowledge.

In particular, it provides a practical effect of reducing construction work by reducing the burden of data collection according to model changes. This is a significant step forward in addressing one of the biggest barriers to AI adoption in manufacturing.

Ultimately, the goal of this technology is to transform manufacturing from a traditional rules-based language to a data-driven one.

This enables a new concept, not just simple automation, but the High-Variance Ready Factory.

In other words, the goal is to create a factory where AI can flexibly respond even in industrial settings where various variables exist.

CTO Seo Hyung-joo asserted, “Physical AI is driving fundamental innovation in the manufacturing industry, going beyond simply ‘robots replacing people.’” He added, “Meeting ROI, diverse application cases, scalability, and transitioning to a data-driven language are the key to solving the challenges facing the manufacturing industry.”