Music of the future

The age of machine people

For some years now, we have been seeing technical beings that look and behave like their human creators. However, apart from the futuristic encounter between man and machine, dealing with android and humanoid robots takes some getting used to. On the one hand, there are the androids. Robots that are developed with the aim of being as human-like as possible - both in appearance and behavior. They often have artificial skin, human facial features, wear hair and move like humans. With such technical imitations of humans, their developers want to explore social interaction in particular - or simply entertain. In Japan, Hiroshi Ishiguro, a professor at Osaka University, has created a robot that looks just like him and is intended to represent him beyond death.

Humanoid robots, on the other hand, are only based on the human physique in terms of their stature - i.e. they have a head, two arms and two legs. Compared to androids, their design is more mechanical and technical, and their functionality is geared towards practical, now often industrial applications. Humanoids do not want to imitate humans. Their developers are more interested in using machines with the dimensions and movement possibilities of the human body to become active in the three-dimensional working and living environment designed by us and for us. Humanoid robots are supposed to be useful helpers.

The uncanny valley of disturbance

The idea of artificial humans entering our lives and taking over tasks is not new. In 1818, for example, Mary Shelley published her novel "Frankenstein - The Modern Prometheus", in which she tells the story of a science student who, after nights of work, succeeded in creating a human being from dead matter. He finally succeeds, but the budding scientist is horrified by the sight of his monster and flees the laboratory.

In 1970, in the early days of industrial robot development, Masahiro Mori wrote an essay exploring the question of how people would react to robots that looked and behaved almost like humans. Mori was then a professor of robotics at the Tokyo Institute of Technology, "Recently, there are many industrial robots, and as we know, these robots have no face or legs, but can only turn or extend and contract their arms," Mori described the state of the art at the time. "They bear no resemblance to humans", their shape simply follows their functionality. However, Mori was convinced that humans themselves were the ultimate goal of robotics, "which is why we are striving to build human-like robots". And this, he suspected, had unintended consequences. In his essay, he postulated that a person's reaction to a human-like robot would abruptly change from empathy to revulsion as its appearance approached, but did not quite reach, a lifelike appearance: "I have noticed that the more human-like robots appear, the more our sense of their familiarity increases, until we enter a valley. I call this relationship the Uncanny Valley".

In recent times, the concept of the "Uncanny Valley" has also attracted interest because the film industry has become increasingly successful in creating and staging computer-animated, human-like characters. In order to avoid slipping into the Uncanny Valley, James Cameron, for example, did not make the inhabitants of the alien world look exactly like humans in his film "Avatar - Departure to Pandora", but instead altered their appearance and, among other things, lengthened their arms and legs slightly.

Interest in the psychological Uncanny Valley effect is likely to increase now and in the future, as the technical possibilities for producing android robots are constantly improving and it is now possible to equip the machines with artificial intelligence. Angela Schoellig, Humboldt Professor at the Technical University of Munich (TUM) and coordinator of the Robotics Institute Germany, refers to this "organic" interaction between mechanics and AI as "embodied intelligence".

Encounters with androids can make us humans shiver. Before Ishiguro built his robot doppelganger, he created a lifelike robot copy of his then four-year-old daughter in his laboratory. The child is said to have been so frightened at the sight of this copy that she did not enter her father's laboratory again for fear of encountering her doppelganger again.

Unlike androids, humanoid robots will not (yet) cause us such worries. The models currently leaving the laboratories in the major economic areas of North America, Europe and East Asia have an explicitly machine-like, high-tech appearance. They fascinate us with their human-like appearance, but do not frighten us. Even though I have never come face to face with one of these humanoid robots, I still like to marvel at their sometimes acrobatic skills, which are well documented in numerous videos on the internet.

Yuchuang Tong, Haotian Liu and Zhengtao Zhang provide an up-to-date overview of the technical status of development in their study "Advancements in humanoid robots: A comprehensive review and future prospects" [1]. The functionality of humanoid robots has increased enormously in recent years due to advances in mechanics, electronics, artificial intelligence and sensor technology. In their work, the authors analyze the structure of the machines, control mechanisms, perception of the environment and interaction with humans as well as possible areas of application. Their findings can be summarized as follows:

• Technological progress: advances in bionic mechanics, materials and control technologies enable humanoid robots to have human-like mobility and interaction. However, there are still challenges such as energy consumption and balance.
• Research and development: Historically, the development of humanoid robots has ranged from basic walking skills to highly dynamic movement sequences, such as the "Atlas" system from Boston Dynamics. Current models integrate machine learning and neural networks to make autonomous decisions.
• Areas of application: Potential applications range from healthcare to disaster relief, education, industry and space travel. The ability of humanoids to take on dangerous and monotonous tasks significantly expands the possible applications.
• Future prospects: Research is still needed to better understand biological structures and processes so that humanoid robots can be further optimized. The integration of biological control, efficient energy conversion and new materials is seen as crucial.
• Challenges: In addition to technical hurdles, the focus is on ethical issues and social acceptance. Robot safety and fault tolerance are crucial if humanoid robots are to be used safely in human environments.

The International Federation of Robotics (IFR) lists the further development of humanoid robots as one of the "Top 5 Robotics Trends 2025". The vision: robots will become all-purpose tools that can independently load a dishwasher or work elsewhere on an assembly line. According to the IFR, robotics start-ups are working on humanoid all-rounders. Large manufacturers, on the other hand, are concentrating on humanoids that can perform individual tasks. Most of these pilot projects are running in the automotive industry. This sector has always played a pioneering role in the development of robot applications, the IFR recalls. This applies to industrial robotics as well as logistics and warehousing. From today's perspective, however, it remains to be seen whether humanoid robots represent an economically viable and scalable business case for broad industrial application.

(Photo: Unitree)

Although the potential of humanoid robots is often praised, concrete applications are still rare, the Fraunhofer Institute for Manufacturing Engineering and Automation (IPA) also expressed its scepticism in a study published in January of this year [2] as part of the AI Progress Center and analyzed the trend towards humanoid robots under the heading "Game changer or aberration?" The team of authors led by Simon Schmidt, Joshua Beck, Lasse Höltge, Alexandra Huber and Ramez Awad reports that in the more than 100 responses from industry experts, material handling, machine loading and gripping complex objects were most frequently mentioned as possible and sensible application scenarios for humanoids. They describe the hurdle to success as follows: "Compared to previous solutions, humanoids should stand out in particular due to their flexibility in performing various tasks." Werner Kraus, Head of the Automation and Robotics research department and co-editor of the study, specifies this requirement as follows: "In my opinion, it is precisely the combination of possible changes of location and flexible gripping technology that is game-changing. This is because it can also be used to automate tasks in existing systems, the brownfield, with little integration effort."

The interviewees in the study are currently still cautious about the technical possibilities of humanoids and assume that such machines will initially carry out tasks where accuracy, system stability or process speed are less relevant as robot capabilities. In many cases, the transportation of crates was seen as a possible task. According to Sudie, 60 percent of respondents doubt whether humanoids need to have two legs for the desired tasks. They find a wheel-driven platform or even a stationary application with a two-armed robot more practical.

Safety and cost-effectiveness as key criteria

According to the Fraunhofer study, the biggest challenge in practical use is functional safety, which is currently still largely unclear and poses special requirements due to the robot design, for example in terms of stability. "That's why I don't see mixed operation with humans as a likely scenario for initial deployments for the time being," Werner Kraus is convinced.

Excessive expectations and uncertainty about the economic viability are further hurdles that companies currently see. Around half of all respondents would be prepared to pay up to 100,000 euros for a humanoid. "In my opinion, the successful use of humanoids will not only require use cases that are technically possible, but also business cases that are economically interesting," says Simon Schmidt, Division Director at Fraunhofer IPA and co-author of the study, summarizing the results.

The recommendations for action derived from the study are as follows:

• Development of software-based safety functions and features
• Development and optimization of precise handling and sensitivity of effectors
• Simple programming and optimization of humanoid robots for their use
• Technologies for perceiving the environment and reacting accordingly
• Creation of a standard/technical specification comparable to that for human-robot collaboration
• Creation of legal evaluation options for companies
• Promotion of critical and differentiating technologies, in particular the growth financing of start-ups

Everything mentioned in these seven points sounds logical, plausible and necessary - and yet a little "typically German", is my impression. Have you ever heard of "China Speed", you might ask? One cultural difference between our proverbial "German thoroughness" and the players in the Far East in their approach to new technical challenges and their solutions is, in my observation, that we in Germany only act when all the milestones on the way to the goal have been precisely formulated and their implementation seems 100 percent certain. In China, on the other hand, people are ready to plunge straight into an adventure - and get started. Those involved are fully prepared for things not going as planned at the beginning. But then they just take a detour and carry on differently. What may seem haphazard to our eyes is not really. The players never lose sight of their goal. In short: systematic planning competes with systematic trial and error. Admittedly, my characterization of the culturally different action strategies is simplistic and striking - and is based on nothing more than conclusions from subjective observation. However, in response to the question of why China is currently bubbling over with approaches to humanoid robots, I would consider that there could also be something like a playful approach to the subject. Who would think of organizing a half marathon for humanoid robots in this country? In the first half of this year, such a robot race is due to start in the Beijing Economic-Technological Development Area (Beijing E-Town).

The US investment bank Morgan Stanley's "Humanoid 100" provides investors with an overview of the current composition of the humanoid value chain [3]. The study shows that 73% of companies with a proven involvement in humanoids and 77% of integrators are based in Asia (56%, 45% in China). There is no doubt that China is becoming a hotspot for humanoid robots. And it should be added: a powerful hotspot. These days, in mid-March, the Chinese manufacturer Unitree Robotics, based in Hangzhou, distributed a video showing the humanoid "Unitree G1" performing a side somersault [4].

After their appearance at the Spring Festival Gala at the end of January on Chinese television, which was watched by 2.8 billion viewers, humanoid robots suddenly became popular rental items on e-commerce platforms in China. The G1 model from Unitree Robotics, for example, priced at 99,000 yuan (13,700 US dollars) and the more expensive H1 model (650,000 yuan) are particularly popular. Customers mainly use the machines for exhibitions or events, but also for brand advertising or company receptions.

The Chinese robot manufacturer Ubtech has equipped the e-car manufacturer Zeekr with robot employees that collaborate with each other (Photo: Ubtech)

For a long time, it seemed as if the American company Boston Dynamics, based in Waltham, Massachusetts, had an unassailable lead in humanoid robots with its "Atlas" model. However, rapid advances in artificial intelligence mean that the cards can now be reshuffled. German companies also want to prove themselves in the competition. At the Consumer Electronics Show (CES) 2025 in Las Vegas at the beginning of January, for example, Schaeffler AG from Herzogenaurach showcased innovations in the field of motion technology. The company has announced its intention to break new technological ground and intends to expand its global presence with new production facilities and strengthen its research and development capacities through the merger with Vitesco Technologies, which will be completed in 2024. "From humanoid robots to automotive and industrial innovations, we are demonstrating how Schaeffler is at the forefront of motion technology," commented Marc McGrath, Regional CEO Americas.

A Humanoid Technology Exhibit took center stage at Schaeffler in Las Vegas. The exhibit offered interactive insights into robotics development by simulating human movements. With its production technology expertise incorporating artificial intelligence and based on its core competence in vertical integration, Schaeffler aims to enable a higher degree of automation and greater efficiency in the use of humanoid robots.

The German start-up Neura Robotics, founded in Metzingen in 2019, is also in the game. In the age of cognitive robots, the Metzingen-based company wants to drive forward the development of humanoid robots using Nvidia's Isaac platform, for example. Among other things, CEO David Reger wants to use cognitive robotics solutions to help make the production of the ailing automotive industry in Germany internationally competitive again: "If Germany makes bold decisions now and turns cognitive robotics into a new economic driver, we can once again shape global markets," he is convinced.

Neura now has more than 300 team members working on environmental perception, drive and control technology, materials technology, mechanical design and artificial intelligence. In line with the company's philosophy, all key components such as AI, control software, sensors and mechanical components for the humanoid robot "4NE-1" are also developed in-house.

In an international comparison, the humanoid robot "Optimus" from Tesla certainly deserves the accolade of being the top dog. At a company event in mid-March, Elon Musk commented on its prospects. Tesla is currently the only company in the world that can scale up the production of humanoid robots, the CEO emphasized in his speech to employees, which is available as a video on the YouTube platform [5]. One basis for this is that Tesla has all the necessary components in-house to be able to produce intelligent humanoid robots in large quantities. Tesla plans to start pilot production on a large scale this year. Optimus will be "by far the greatest product ever" - nothing will even come close, Musk enthused.

And he quotes figures: "We hope to produce around 5,000 Optimus robots this year. Our goal is to have enough parts for 10,000, maybe even 12,000. However, as this is a completely new product - everything about it really is new - it would be a success if we could manage half of the 10,000. But even 5,000 robots - which, by the way, is the size of a Roman legion - is an impressive number." According to Musk, ten legions of Optimus robots should be populating factories by 2026. Musk likes his bon mot: "I think that's cool - legions as a benchmark. So it will probably be around 50,000."

The announcement to "roll out" Optimus is also to be understood as a statement for the further automation of production, which will now also include assembly work that previously required human hands. Musk had already announced the expected prices for the Optimus months ago. The basic version is expected to cost around 30,000 dollars. We can look forward to the imminent arrival of the iron journeymen in the factories.

SOURCES:

[1] https://ieeexplore.ieee.org/document/10415857
[2] https://www.ki-fortschrittszentrum.de/de/studien.html
[3] https://advisor.morganstanley.com/john.howard/documents/field/j/jo/john-howard/The_Humanoid_100_-_Mapping_the_Humanoid_Robot_Value_Chain.pdf
[4] https://www.youtube.com/watch?v=9Wz4XIhODPM
[5] https://www.youtube.com/watch?v=QGJysv_Qzkw