My latest trip to Japan opened my eyes to something I have been thinking about for years: culture is not a side topic in robotics. It is infrastructure.

This text is not a research paper. It is an observation, mixed with years of working around robotics, designing robots, meeting robotics companies, and watching this strange industry try to grow up.

You do not see robots everywhere in Japan. That is not the point. In fact, if you go there expecting a Blade Runner city full of machines, you will probably be disappointed. The interesting part is more subtle. You feel that people have a sense of proximity to robots. They do not necessarily see them as magic. They also do not immediately see them as monsters.

I have been thinking a lot about the recent wave of small humanoid robots.

Not childlike robots designed for children. Not entertainment robots. Not toys. I mean the growing category of 120 to 140 cm humanoids that look like compressed versions of the future we are being promised.

And honestly, I think their best use case today is research.

In this essay, I look at Moravec’s paradox at industry scale. My hypothesis is that robotics struggles most in industries where time has optimized the work around human improvisation instead of machine legibility.

Construction, mining, and agriculture may look simple from the outside. Move this. Drill that. Carry this. Inspect that. But the real work often lives in invisible judgment: reading the environment, adapting to bad information, sensing material behavior, recovering from small mistakes, and constantly repairing the task.

That is where robots struggle.

And it is also why robotics often succeeds faster in structured environments like factories and warehouses, where the world has already been simplified before the robot arrives.

Robotics is moving from specialized machines to converged systems.

Classical robotics worked by narrowing the problem: industrial arms, AMRs, and grippers each solved specific tasks inside structured environments. The complexity did not disappear, it was pushed into fixtures, workcells, conveyors, safety systems, and process design.

Modern robotics is moving in the opposite direction. As robots are expected to operate in less structured environments, mobility, manipulation, grasping, perception, and autonomy are starting to merge into single integrated systems.

That convergence creates versatility, but it also brings a serious complexity tax.

The real question is not whether a robot is complex. It is whether that complexity turns into useful, reliable, and economically meaningful versatility.

What is GenAI actually good at, and why it fails in areas such as robotics. Here I mostly focus on CAD GenAI and expand it to robotics shortly.

GulaMannen: A Modular Platform for Humanoid Robot Design and Simulation

You see things as a consultant that you simply don’t see otherwise. One fun one, maybe not: I often know when someone is about to quit or get fired before others do. Patterns are patterns.

Are robots going to replace people’s jobs?

It is a question that comes up again and again, usually framed as a near-term threat or a sudden breakthrough. From inside robotics, the picture looks very different.

Robots do not replace jobs. They replace tasks, and only under very specific conditions. Jobs are made of structured and unstructured work, responsibility, regulation, and accountability. Replacing execution is one thing. Replacing accountability is something else entirely.

In this piece, I break down why environment structure, task value, and system complexity matter more than intelligence headlines, and why humanoids are not a shortcut to job replacement.

Part 1 focuses on tasks, structure, and limitations. Part 2 will look at human-robot collaboration and what actually scales in the real world.

humanoid robots in 2025

The year is coming to its end, and it truly was an exciting one for robotics. Below are some of my observations, very much prone to be challenged. Anecdotal and non-structured, try to make sense of it yourself :)

We saw faster growth in humanoids than the previous year. The number of companies almost doubled, and the investment going into them increased a lot. As expected…

Humanoid robot design hasn’t reached the standardization we see in other industries like mobile phones or cars. While industries mature and products become more uniform, robotics remains in a unique space. This blog explores why humanoid robots aren’t yet standardized, the factors behind this, and what this means for the future of robot design.

CMF (Color, Material, and Finish) is a critical factor in the design and branding of humanoid robots, shaping their emotional connection with users and enhancing their functionality. The choice of color can convey the robot's personality and influence how it is perceived in different contexts—whether as friendly, authoritative, or professional. By understanding how different colors impact user perception and robot identity, we can design robots that are both functional and emotionally engaging. This post delves into the significance of color in humanoid robot design, analyzing how colors influence the robot's purpose, user interaction, and brand identity.

As an industrial designer focused on the design and interaction of robots, I've spent the last 9 years designing various types of robots. By far, the most enjoyable and challenging projects have been humanoid robot designs. In this post, I’ll share some key tips for designers and roboticists navigating this exciting field.

Explore the concept of the "Uncanny Valley" in humanoid robot design, a challenge where robots that resemble humans too closely but imperfectly cause discomfort. Learn about the factors contributing to this eerie feeling and discover design strategies to avoid it, ensuring robots are both functional and emotionally acceptable in various real-world applications.