Shipyard Welding: A Viable First Job for Humanoid Robots?

Humanoid robots urgently need to find real employment instead of starring in YouTube videos, and Persona AI is one of the few companies focused exclusively on that goal. Persona's strategy has been to carefully select a job that is economically viable for robots today, and they found a role that also happened to be one of the first industrial robot jobs in history: welding. IEEE Spectrum first interviewed Persona two years ago, shortly after its founding by Nic Radford and Jerry Pratt. Radford previously led NASA JSC's Valkyrie program and founded Nauticus Robotics, while Pratt led IHMC's DARPA Robotics Challenge team before serving as CTO of Figure for a couple of years.

The Challenge of Humanoids

As of the initial conversation in 2024, Persona had committed to building an economically viable humanoid but had not yet determined where to focus. According to Radford, the company explored various environments, including warehousing and automotive, and even mentioned the home. These are the same settings that nearly every other humanoid robotics company is attempting to make economically viable, yet despite extensive demonstrations, none have achieved meaningful scale. The fundamental challenge for Persona and the entire industry is that it is insufficient for a robot to merely perform a task; it must do so efficiently, reliably, safely, and, crucially, generate profit for both the developer and the customer. Most humanoid companies target baseline unskilled labor to achieve this, but Persona did not see economic viability there.

Radford explained that Persona developed a thesis around skilled trades and tool usage, targeting much more expensive skilled labor. This approach is feasible because their entry point leverages skills that robots excel at. Radford referred to this as a 'last-mover advantage,' having observed the industry's efforts and choosing a different path. The first task for Persona's humanoid is welding, using a handheld tool to join metal pieces. According to Pratt, tool use is quite difficult, and the decision to use human tools adds to the complexity. However, the difficulty is mitigated by initially focusing on long, linear welds that are relatively straightforward. Radford noted that shipbuilding requires hundreds of kilometers of such welds per ship, making it a high-demand application.

Currently, Persona has two public partnerships: one with HD Hyundai, the world's largest shipbuilder, and another with POSCO, the world's third-largest steel producer, both in Korea. While Persona declined to provide details, Radford indicated that they seek customers who can support hundreds of robots per location. Shipyard welding is attractive due to a shortage of skilled, highly paid workers, the physically taxing nature of the work, and the relative ease for a humanoid to learn the skill. The welding process is robot-friendly because the maximum speed is about one centimeter per second, and high-quality welds require millimeter-scale repetitive motions, which robots perform well over long periods without fatigue or boredom. Pratt expects that for these simple welds, robots will soon match or surpass human performance.

Shipyards also make a compelling case for legged robots over stable wheeled bases. Radford described open-air shipyards as several hundred meters long, with horizontal and vertical spars that require stepping over, and work occurring on the ground, overhead, and through portholes. Persona considered other form factors but decided that a two-legged robot would be the least disruptive to existing shipyard operations. Deploying in shipyards brings additional advantages for Persona. Safety concerns with bipedal robots are reduced because shipyard workers are trained to work around potentially dangerous industrial equipment. The company also faces less pricing pressure because no other company is pursuing this use case. Pratt noted that the high value added by their robot allows them to maintain quality and features without cutting corners on price, unlike competitive markets such as home robotics.

The economic value for shipbuilders, according to Radford, does not come from replacing humans but from addressing labor constraints. Persona's partnerships are running at a significant backlog, and they want to help customers' top line by enabling more ships to be built, even if the robot costs more than a human worker. Since additional skilled labor is unavailable, Persona's robots offer an alternative for shipbuilders looking to scale. In the current commercial humanoid climate, where many companies focus on general-purpose robots with undefined future applications, Persona stands out with a specific, near-term business case. Radford acknowledged that they have been told they are not thinking big enough, but the tool-using heavy industrial humanoid has many potential future applications. Pratt views shipbuilding as a beachhead, with adjacent markets like grinding, painting, and other fabrication.

Persona's ambition, as stated by Radford, is to become the largest repository of industrial skills. Achieving this will require collecting tens of thousands of hours of expert demonstration data, creating high-fidelity simulations, and conducting real-world testing. Despite the promising approach, Persona must still prove that its idea is financially viable, a challenge faced by all humanoid companies. Pratt noted that many technical problems are similar whether in a home or a shipyard, and the industry is collectively working to solve them.