[Edaily Reporter Shin Young-bin] #Hanwha Ocean is launching a pilot project to deploy humanoid robots at its shipyards. This is expected to be an early example of testing the feasibility of using physical AI-based humanoid robots at domestic shipbuilding sites.

According to industry sources on the 8th, Hanwha Ocean is collaborating with companies such as N.DotLight and Airobot to conduct a pilot project involving physical AI-based humanoid robots for application in shipbuilding sites. This project is part of the government’s AX Sprint initiative, with the goal of completing a full-scale field demonstration using humanoid robots within one year.

AX Sprint is a government research project designed to rapidly validate the adoption of AI in industrial settings. It involves collaboration between companies and AI/robotics technology firms to verify the feasibility of real-world industrial applications within a short timeframe. Through this project, Hanwha Ocean will examine the potential for applying humanoid robots to dangerous and physically demanding tasks within shipyards.

Companies participating in the NVIDIA Inception ecosystem, a startup incubation program, are collaborating on this demonstration. 3D AI technology firm N.DotLight is responsible for building a simulation environment based on NVIDIA Omniverse and Isaac SIM, as well as SIM-ready data, digital twins, and synthetic data. Its role is to create a virtual environment similar to an actual shipyard site and support the repetitive learning and verification of movement, avoidance, task execution, and operation scenarios before the humanoid robots are deployed on-site.

Airobot, a humanoid robot company, will verify the feasibility of performing shipyard tasks using its proprietary humanoid robot, “Alice.” Airobot plans to train the humanoid using motion data from industrial sites and confirm whether it can perform physical tasks required in shipyards, such as autonomous navigation, walking on rough terrain, obstacle avoidance, carrying goods, and tool manipulation.

The shipbuilding industry is considered a highly symbolic sector for verifying the industrial applicability of humanoid robots. Shipyards consist of large blocks, complex work routes, confined spaces inside ships, and high-risk inspection zones, creating environments where it is difficult to apply existing fixed automation equipment or industrial robots. A significant portion of on-site tasks—such as welding, transport, and inspection—rely on skilled workers, and the burden of high-intensity, high-risk work is substantial.

Humanoid robots can expand into areas that were difficult to address with existing automation equipment, as they can utilize the same workspaces, equipment, and movement paths as humans. Particularly in spaces with complex structures and constantly changing work environments, such as inside ships, the key challenge is whether humanoid robots can secure mobility and operational flexibility.

This project is in the initial demonstration phase to verify on-site applicability. For humanoids to be reliably utilized in shipyards, capabilities such as rough-terrain walking, balance control, heavy-load handling, tool use, safety verification, and collaboration systems with workers must all be secured. This is why simulation-based learning and digital twin verification are crucial before actual deployment on-site.

The competition in physical AI is shifting toward manufacturing sites. While generative AI has focused on processing digital information such as text and images, physical AI must understand the real-world environment and translate that understanding into the physical actions of robots. Industrial sites with many variables, such as shipyards, serve as prime testing grounds for verifying the practicality of physical AI and humanoids.

For Hanwha Ocean, this demonstration could serve as an opportunity to confirm the feasibility of applying robots to realize an “AI shipyard.” If humanoids can replace or assist with certain tasks—such as inspecting hazardous areas, performing repetitive on-site movements, transporting equipment and materials, and assisting workers—it is expected to enhance safety, productivity, and workforce efficiency in the shipbuilding industry.

Park Jin-young, CEO of N.DotLight, stated, “To apply physical AI to actual industrial sites, a Sim-Ready data infrastructure that enables robots to learn and validate is absolutely essential.” He added, “We will support the shipbuilding industry’s demonstration efforts in advancing to the actual field application stage through NVIDIA Omniverse and Isaac-based simulation environments, Sim-Ready 3D assets, and synthetic data generation technology.”

Eum Yun-seol, CEO of Airobot, “Airobot’s humanoid robot, Alice, was introduced as a prime example of physical AI during NVIDIA CEO Jensen Huang’s CES keynote speech last January, accompanied by footage of it performing welding tasks at a shipyard,” he said. “By combining Alice’s robotic technology with EnDotLight’s Sim-Ready data infrastructure, we will demonstrate the potential for robots to take over physical tasks that are difficult or dangerous for humans to perform.”