What Will the True Face of Robots Be?

Today, we are building humanoid robots without having yet decided what they should really look like or what they should represent. This is probably one of the clearest signals of the current state of the industry, and not merely an aesthetic detail.

Unlike other technologies such as cars, smartphones or computers, which have gradually developed relatively established and recognisable design languages, humanoid robots are still in a completely fluid and experimental phase.

There is no dominant style, nor a shared design grammar. Every company is testing its own idea of the “form of embodied intelligence”, as if we were looking at a collection of parallel industries that have not yet agreed on what they are actually building — or how to present it to the market, which, after all, is made up of people: humans.

This fragmentation becomes particularly evident when looking at the industry’s main players, also in light of what emerged in the previous V12 Observatory from CES 2026, where the distinction between demonstration robots and genuinely operational robots became increasingly clear.

(Missed it? Read the full article here 👇)

The industrial world: where efficiency and reliability come first

On one side, there is the industrial world, where design is subordinated to function and integration into existing processes. An interesting example is Renault’s Calvin, a humanoid robot designed for logistics that does away with the head in order to lower its centre of gravity and improve stability and operational safety. Here, design does not try to mimic the human form; it avoids it when it becomes a constraint.

Also in the industrial sector, Figure AI represents a different but equally important theme: scalability. Its robot has been tested over a year of real-world operations at BMW, performing production-line tasks such as handling metal components. The experiment demonstrated not only technical feasibility, but also the speed at which humanoid robots can be scaled industrially, moving from around one robot produced per day at the beginning of the test to a projected rate of almost one robot per hour by the end of the year. A shift in scale that transforms the robot from a prototype into a genuine production system.

The real challenge is at home

On the other side, we find an almost opposite approach: robots designed for shared and domestic environments, where the challenge is not only functional, but also deeply relational.

This is where 1X Technologies comes in. Its approach to domestic robotics addresses one of the most difficult questions of all: how do you introduce an autonomous machine into an intimate space without making it feel intrusive? The design inevitably has to become softer, more restrained, almost discreet, as if the robot first had to learn how not to disturb the people living in the home before learning how to impress them.

UBTECH Robotics is exploring a similar direction with robots designed for social interaction and education, where the appearance moves closer to the human form without fully embracing realism.

This brings us to one of the most interesting and still unresolved challenges in the entire sector: the face.

There is currently no consensus on how closely a robot should resemble a human being. Some approaches pursue an almost theatrical level of realism and expressiveness. Others favour stylised digital interfaces, while others eliminate the face altogether to avoid emotional ambiguity.

Between these two extremes lie a range of hybrid strategies. Tesla’s Optimus continues to maintain a deliberately neutral aesthetic, suspended somewhere between the industrial and domestic worlds, while Engineered Arts, with Ameca, explores a far more expressive and relational dimension, turning the robot into almost a communicative interface.

In all these cases, Figure AI, Tesla, 1X Technologies, Boston Dynamics, UBTECH Robotics and other players are not simply competing on performance and functionality. They are exploring radically different aesthetic directions, as if each company were designing a possible alternative species of humanoid without yet knowing which one will eventually become the standard.

Trust as the ultimate interface

None of these solutions has yet emerged as the clear standard, highlighting how the challenge is not purely technical, but cultural and social.

Robots may already have all the capabilities required to perform tasks, carry out activities and provide us with their technical skills. But if they do not have the right appearance — one that makes us comfortable enough to trust them and let them through our front door — those capabilities may never make it into our homes, or even into more human-centred working environments.

Among machines, robots already work extremely well. But when they have to work alongside a person of flesh and blood, everything changes.

This is also the direction being explored by companies such as Shift, which is approaching the problem from a different angle. The start-up collects data by having its robots perform domestic cleaning tasks free of charge in real homes, using cameras to record gestures, movements and the way people interact with their surroundings. The goal is not simply to teach robots how to clean, but how to move naturally through a home, avoid causing damage or discomfort, and learn the behaviours that increase the likelihood of being accepted in everyday life.

Everything is happening so quickly that perhaps one day — and perhaps not too far into the future — the question will simply be which “artificial presence” we ultimately decide is normal.

A concrete signal is already coming from UBTECH Robotics, which has opened pre-orders for its new U1 series, humanoid robots with silicone faces designed for domestic companionship. Around 13,000 reservations were reportedly made within just 48 hours, with a deposit of approximately €400 and prices ranging from €15,000 to €120,000 depending on the configuration. Numbers like these shift the conversation from prototypes to genuine market demand.

At this point, the question inevitably comes back to the design of the face — not simply as a technical or aesthetic problem, but as a cultural choice.

Because if trust is also shaped by appearance, perhaps the real question is not how much a robot should look like us, but whether the right face for a robot should be human at all.