POSCO DX will jointly develop industrial site robot intelligence technology with the AI corporations NC AI. The goal is to apply physical AI technology to robots that perform high-risk, high-intensity tasks.
POSCO DX said on the 31st that it signed a memorandum of understanding (MOU) on the 29th at NC AI's Pangyo headquarters to implement an AI-based robot autonomous work system. The signing ceremony was attended by Yun Seok-jun, head of the POSCO DX Robot Automation Center, Kim Min-jae, NC AI chief technology officer (CTO), and officials from both companies.
The core of this collaboration is to implement a robot foundation model by combining POSCO DX's robot simulation and control technology with NC AI's AI model. A robot foundation model refers to a basic AI model that allows robots to understand and make judgments in various work situations.
Under the agreement, POSCO DX will be responsible for motion planning for robots to be applied at industrial sites and for validating control simulations. To ensure robots can operate stably, it also plans to provide a virtual test environment based on a Digital Twin (a technology that tests in advance by implementing actual equipment or work environments in a virtual space). Motion planning refers to technology that calculates a movement path for a robot to perform a target task.
NC AI will handle development of the AI model for the robot foundation model. It plans to collect and analyze industrial site data so that robots can understand various work situations, and based on that, develop and apply a vision-language-action (VLA, technology that enables robots to understand visual information and language instructions together and translate them into actions) model.
The two companies expect that applying the collaboratively developed robot foundation model to industrial site robots will enable handling of unstructured tasks and more. They aim to develop technology that allows industrial site robots to understand visual information and language instructions together to interpret and judge work situations, and to respond even when the workpiece position changes or the method of connection between equipment differs. With rule-based control of conventional industrial robots, there were limits to handling such unstructured tasks.
POSCO DX is developing industrial site robot automation technology centered on its Robot Automation Center. Through simulation-based validation and the establishment of standardized models, it is improving the stability and deployment speed of industrial robots, and by developing a heterogeneous robot operating platform, it is strengthening operational efficiency by managing diverse robots in an integrated manner.
Yun Seok-jun, head of the POSCO DX Robot Automation Center, said, "Through ongoing collaboration with corporations that have specialized technologies, we are internalizing key solutions such as robot control and operating platforms and raising the level of automation technology for high-risk, high-intensity sites," adding, "We expect this collaboration to confirm the potential for using general-purpose robots at industrial sites."
NC AI CTO Kim Min-jae said, "General-purpose robot technology is rapidly evolving as next-generation AI technology applicable across diverse industrial environments," and added, "Through collaboration with POSCO DX, we will strengthen competitiveness in robot AI technology and jointly lead a global general-purpose physical AI ecosystem targeting the global market."