Humanoid Robots are Coming

Humanoid robots are projected as the next major automation technology. Videos show robots walking, carrying boxes, folding clothes, or working inside factories, and it is easy to put them all into the same category. But this falls short of the reality. There won’t be one standard humanoid robot. The design will be determined by the application, and industrial, service, and home robots will consequently develop in very different directions.

This is also what makes humanoid robots interesting for the foundry industry. We are not looking at one standardized component that a handful of suppliers will eventually produce. We are looking at a new product category whose architecture is still being developed and which requires very different structural solutions depending on where and how the robot is used.

What actually makes a Robot Humanoid?

A humanoid robot is defined primarily by its human form factor. It normally has a torso, arms, a head, and some form of hands or grippers because it is supposed to operate in environments that were originally designed for humans. From the torso down, however, there is considerably more freedom. Some concepts use two legs, while others use a wheelbase connected to the torso through an adjustable or articulated structure.

That difference is not primarily about appearance. Walking on two legs is significantly more complicated from a control and safety perspective, and there is little benefit in solving that problem when the robot operates on a flat factory floor. If forklifts and carts can already drive through the production environment, a wheelbase can do the same job considerably more easily. Legs become attractive when the robot has to deal with stairs, uneven terrain, or obstacles it can’t avoid.

The same logic applies to the hands. A robust gripper for handling boxes or heavy parts requires completely different characteristics from a hand intended for precise manipulation. Trying to use the same gripper to take a hot casting from a machine and later clean wine glasses would obviously make little sense. The final robot will therefore be shaped by the task it has to perform rather than by the ambition to copy the human body as closely as possible.

Industrial Applications will Come First

At the moment, industrial applications are the most attractive starting point because the environment is comparatively controlled and the tasks are much easier to define. A factory needs a robot to move parts between stations, feed a machine, or perform repetitive handling tasks. The environment can be mapped, the process can be trained, and the number of unexpected situations is limited.

This does not mean that humanoids will replace the industrial robots already installed in highly automated production lines. A conventional industrial robot performing the same movement every few seconds is extremely efficient because everything around that robot has been optimized for exactly this task. Replacing such a system with a more complex humanoid often makes little sense.

The opportunity appears where conventional automation becomes too inflexible or too expensive. There are still countless operations in factories and warehouses that are performed manually because the volumes, variations or changing conditions do not justify dedicated automation line. A general-purpose robot that can learn several of these tasks creates a completely different economic proposition. Instead of engineering a new automation cell for every process, the same type of robot can be deployed for different jobs.

Home Robots have a much Harder Job

The home is almost the opposite of a factory. Every house looks different, furniture changes, there are stairs, rugs, objects on the floor, and an almost unlimited number of situations that the robot has never encountered before. Even seemingly simple tasks such as unloading a dishwasher or folding clothes become complicated when the robot has to perform them in thousands of different environments.

This is why the development is likely to happen step by step. Instead of immediately creating the universal household robot that can clean the kitchen, do the laundry, and prepare dinner, companies are starting with individual tasks that can be trained and gradually expanded.

Over the next five to ten years, Wilberth Solano from Humanoid expects these robots to become increasingly common in manufacturing, assembly and logistics. Service applications could develop alongside them, while general home applications are likely to take longer because of environmental complexity.

A New Product Category for Foundries

For foundries, the important point is that all of these different robot concepts need physical structures. Arms, legs, joints, torsos, bases, and other load-bearing components have to connect the electronics and actuators into a machine that can repeatedly interact with its environment.

At the moment, much of this hardware is still produced with flexible manufacturing technologies because robot companies are continuously changing their designs. That won’t be economically viable if humanoid robots eventually reach the volumes many manufacturers are targeting. Higher production volumes will force the industry toward scalable processes, and this is where casting becomes increasingly relevant.

The interesting part is that the product architecture has not yet been frozen. Foundries are therefore not entering another mature market in which a customer sends out a finished drawing and asks ten suppliers for a quotation. Humanoid robotics is still developing the parts, manufacturing concepts, and even the basic product architecture.

For an industry that has spent years watching automotive volumes decline, this is exactly the kind of market worth understanding early.

The question is not whether every humanoid robot will contain the same twenty aluminium castings. It almost certainly will not. The opportunity comes from the variety of applications and from helping robot companies determine which structures should become castings in the first place.