POSCO DX has revealed it is developing a practical hybrid robot for factory floors.
This robot is cleverly designed by combining readily available, off-the-shelf parts: a four-legged robot from Unitree (B2) and a collaborative robotic arm from Universal Robots. This modular approach makes the robot significantly more affordable than custom-built systems from Western competitors. The goal is to create a cost-effective solution that can handle tasks like sample handling and transfer, with a total hardware cost that is likely under ₩200 million. This makes it a financially viable option for factory automation.
So, why is this happening now? Three key factors have aligned. First, the cost is no longer a major barrier. The public pricing of components like the Unitree B2 (around $100,000) makes the project economically feasible. Second, there is a clear technological upgrade path. NVIDIA's advancements in robotics AI, from Project GR00T to the latest GR00T N2 previews, mean this robot won't be a dead end. It can start with simple, rule-based controls and later be upgraded with sophisticated VLA (Vision-Language-Action) models. Third, POSCO has the internal experience. The company has already tested four-legged robots for inspections and established a dedicated Robot Automation Center, which reduces the risks associated with implementing new technology.
This development also fits into a larger strategic context. Korean factories are facing intense price competition from Chinese and Japanese robot manufacturers. At the same time, there are geopolitical risks, such as potential anti-dumping tariffs on Chinese parts or U.S. sanctions against companies like Unitree. POSCO DX's strategy of using a modular, cost-effective design is a smart response to these pressures. It allows them to deploy capable robots quickly while keeping the option to swap components if supply chains are disrupted.
In essence, POSCO DX's project is a pragmatic and forward-thinking move. It's not just about building a fancy robot; it's about deploying a useful, affordable automation tool today that is ready for the AI-powered future of tomorrow. The calculated payback period of roughly 1.2 to 2.2 years, depending on the tasks it replaces, makes it a compelling business case for improving safety and efficiency in heavy industry.
- VLA (Vision-Language-Action): An advanced AI model that allows a robot to understand tasks from visual cues and natural language commands (e.g., "pick up the red box") and then perform the corresponding physical action.
- Proof of Concept (PoC): A small-scale test or trial to demonstrate that a business idea or new technology is feasible and works in practice.
- Bill of Materials (BOM): A comprehensive list of all the raw materials, components, and assemblies required to manufacture a product.