Humanoid robots from Tesla, NEURA & Co.: Thoughts

(The post was created before Tesla's AI Day presentation and added to afterwards).

More than forty years ago, Kraftwerksang"Am Heimcomputer sitz ich hier, programmier die Zukunft mir" (music below). Developed with far greater effort than on the home computer, three companies at once have recently presented prototypes of humanoid robots. Tesla started and was first caught up by Xiaomi and on Thursday by NEURA Robotics. Apparently cars are not robots after all, so know how in cars would be transferable to robotics, according to Elon Musk. But: Tesla's presentation on Friday showed the potential. Price under 20,000 dollars, market entry by 2027, was heard. Samsung may also be planning a humanoid robot.

Discuss with us! The comment field is open below for anonymous comments.

Market maturity may still take years

Nikolai Ensslen, head of Synapticon, an important robotics supplier with headquarters in Stuttgart and a branch in Silicon Valley, among other places, told the Handelsblatt before the AI Day presentation on series production readiness at Tesla: " "That will take at least five, if not ten years." That's because while 7 lines of freedom are already something special in robots, a humanoid robot is more likely to consist of 50 joints. David Reger also sees his model presented yesterday not tomorrow on the market, but rather in a few years. In a blog entry, the fantasy was directed to the year 2030. No matter, one way or another, humanoid robots will come. Not until later, even if "drumming" is already taking place.

The market and society still have time to answer important questions

David Reger had already asked three important questions yesterday when he introduced his humanoid robot:

1. Does our society need humanoid robots?
2. If so, should they show emotion or act like a machine?
3. Society should become aware of how far robotics has already come. If a humanoid robot can be built, how can it still doubt the usefulness of cobots and the rest of robotics?

I would like to share my thoughts here. To 1.) comes from my side a very clear yes. The question for me is rather, are there limits or how hard are these? There are existential areas for all of us that can possibly only be maintained by humanoid robots. One example: garbage disposal. What would we do without garbage collectors? Here, humanoid robots could maintain the system. For me, it would also make perfect sense to use them as caregivers in nursing homes. David Reger pointed out to me that a robot could protect the self-esteem of the person in need of care. For many elderly people, it would be downright humiliating to be washed by strangers. I, at least, would feel guilty. A robot would be perfect for that.

For our German society, humanoid robots would be difficult in the area of security. The G20 riots with hundreds of injured police officers have shown that we consider "man-on-man fights" to be quite normal. In Switzerland, which today already uses mobile barriers and rubber bullets to keep larger gatherings under control with just a few police officers, humanoid robots should be much more likely to be used in the security sector. In dictatorships, after all, anything seems possible.

To 2.) With the emotions must be probably differentiated. On the one hand, it is about showing emotions. NEURA Robotics has given its robot a typical machine name "4NE-1". Emotions would not fit to this name. In the end, however, there is little to be said against emotions in my opinion. Xiaomi has already announced that its humanoid robot will recognize human emotions, e.g., by their voice, and act accordingly. This would increase the added value. At most, emotions toward children would be difficult. In my opinion, these enjoy protection of trust. A robot that feigns emotions would be problematic.

3) Our society is a difficult subject. We have to be careful not to overtax them and not to raise expectations too high. I think the mass of people can slowly be introduced to robotics via service robots in the catering trade. In addition, cobots are becoming increasingly widespread in companies. In any case, it should be clearly communicated that it will still take years before widespread use takes place. First come the household robots.

Other aspects

It's all a question of cost. As soon as a humoid robot costs as much as the integration of a robotic arm, it can translate it. Today, adapting a manufacturing process to the cobot often costs as much again as the cobot or more. But if a humoid robot could simply work without any adaptation, cobots are likely to come under pressure. For this reason, I have long considered them to be a transitional technology, but one that still has its best years ahead of it. I.e. buy a cobot today or tomorrow. It is worth it!

But what makes the humanoid robot so special for manufacturing? In a considerable number of cases, probably only its two arms. Perhaps more robotic arms should cooperate than before and also be provided with hands. At automatica, cooperating cobot arms were already on display at Agile Robots and NEURA Robots. The possibility of walking is probably not quite as important for the industry - unless mobile robots including the arm are to be replaced.

As is well known, I am an opponent of the robot tax (reason: robots tend to replace low-skilled workers. However, they cost the social system money, i.e. their loss as contributors hardly harms the system. And: If a company makes more profit by using robots, it also pays more taxes).

But: Today, robots reduce the shortage of skilled workers. But what happens when humanoid robots can replace a human at lower cost? Then we would have unemployment again, which would have to be financed. If this danger is imminent, a robot tax would certainly make sense. Perhaps already before because the state simply needs the money.

Overall concept: If I consider that NEURA Robotics has already presented three robots with Maira (cognitive robot with strong AI), MAV (mobile robot) and Mipa (household robot), which together are most likely to provide a convincing overall construct, I consider the Neura approach to be the most promising. Xiaomi has a high competence in AI, but not in mechatronics. In the area of robotics & AI, the Chinese robot manufacturers also have little to offer. The situation is similar for Tesla. Here, the question also arises as to the feasibility of the above-mentioned ratio of dead weight to payload of less than 1.5. Even robot arms today quickly weigh three times as much as they can lift. And they don't even have to keep their balance, because they are firmly mounted.

The Tesla demonstration was good. It can be assumed that CEOs of other corporations will now panic ("are we oversleeping again?") and take up the issue.

Youth: In Germany, it neglects STEM subjects. Perhaps they will now become more interesting for students/school leavers.

P.S: The comment field is open below - without registration. It would be interesting to hear what representatives of robotics companies who are not even interested in AI so far think.

Are we networking? LinkedIn
-> To the Cobot group on LinkedIn (Link)

In my own account/advertisement
The author of this blog is significantly involved in the AI/robotics project Opdra. He consults around robotics. More about him can be found here.