NC AI and POSCO DX partner on general-purpose robot AI models
NC AI and POSCO DX signed a memorandum of understanding on May 29 to jointly develop robot foundation models and advance physical AI capabilities for industrial environments.
The partnership targets a specific technical gap: existing industrial robots handle only predefined tasks in fixed settings. The companies aim to build AI systems that let robots adapt to variable conditions, make autonomous decisions, and complete complex tasks without explicit programming for each scenario.
What the collaboration covers
NC AI will focus on optimizing vision-language-action (VLA) models-systems that process visual input, language instructions, and robot actions together-and building digital twin simulation environments. These digital replicas of real-world conditions let developers test AI models before deploying them on physical hardware.
POSCO DX will advance industrial robot control systems and operation platforms. The company is working on simulation-based verification methods and standardized model development to improve robot stability and deployment speed.
Both companies plan to validate their work in actual industrial settings, where robots encounter misaligned workpieces, equipment interface differences, and other variables that rule-based control systems struggle to handle.
Why this matters for development teams
General-purpose robot models represent a shift from task-specific automation toward flexible systems. Rather than programming each robot behavior separately, developers could deploy foundation models that adapt across different industrial sites and tasks.
The partnership addresses unstructured tasks-those without clear rules or predictable conditions-that current industrial automation cannot process reliably. Manufacturing sites, logistics facilities, and other high-risk environments frequently encounter these situations.
Kim Min-jae, NC AI's chief technology officer, said the collaboration aims to "strengthen competitiveness in robot AI technologies and jointly lead the global general-purpose physical AI ecosystem." Yoon Seok-jun, head of POSCO DX's Robot Automation Center, said the partnership will "help verify the practical applicability of general-purpose robots in industrial environments."
The companies plan to expand their research scope and establish phased verification systems for robot AI technologies based on real industrial conditions.
For development professionals, this signals where industrial automation is heading: from rigid, scenario-specific systems toward AI agents and automation platforms built on foundation models and LLMs that can generalize across tasks and environments.
Your membership also unlocks:
