What if the most dangerous, hazardous, and complex tasks in heavy industry could be bolstered by robots capable of crouching, welding, and lifting completely autonomously? During a demonstration conducted in collaboration with HD Hyundai—one of the world's largest shipbuilders—the "Persona Jane One" robot executed the entire procedure without any external structure supporting its body.

It walked to the work area, crouched to lower its center of gravity, correctly positioned the welding tool, and maintained a stable electric arc throughout the operation before returning to its starting position. To protect its vision, the system employs special optical filters capable of mitigating the impact of the intense light generated by welding, allowing the cameras to continue identifying the environment despite constant flashes.

But there is an even more impressive detail: despite appearing completely autonomous, the robot's most delicate movements were actually controlled by a human operator wearing virtual reality goggles. Every movement the robot makes—every minor course correction and force adjustment—is recorded in a massive database. In practice, the robot is learning by watching human specialists perform the job under real-world conditions; this strategy allows the company to accelerate the training of future artificial intelligence systems. Instead of relying solely on digital simulations, the algorithms analyze thousands of real-world welding examples, learning how experienced professionals react to various situations.

The goal is for the system to eventually perform these tasks with progressively less human intervention, ultimately achieving high levels of autonomy.