Humanoid robots have captured our imagination throughout history. Science fiction has transported us to idyllic worlds where we live in harmony with friendly robots and to dystopian futures with droids that rebel against their own creators. Beyond science fiction, the industry has been working on the development of humanoid robots for decades.

In recent years, we have seen prototypes, proofs of concept and short-lived commercial attempts. Now, advances in robotics and artificial intelligence open the door to a leap in the robotics revolution. In fact, the industry has embarked on a race to commercialize general-purpose autonomous bipedal humanoid robots that can interact with humans safely.

In a first phase, they will be able to work alongside us in various industries and, in the future, become home assistants. “For decades, we have longed for this type of robots, but there was not sufficiently advanced technology, they were not economically viable and there was no clear utility,” summarizes Pablo Medrano, CEO of Casual Robots, a Spanish robotics rental and consulting company.

When technology giants take positions, something is moving. Last week, Microsoft, OpenAI, Nvidia, Amazon and Jeff Bezos entered the $2.6 billion startup Figure AI to accelerate their plans to create humanoid robots. For his part, Elon Musk has shown his ambitions with Optimus, a model in the development phase that he wants to turn into a line of business for the car manufacturer Tesla.

Meanwhile, the Norwegian start-up 1X has recently raised $100 million in a round led by EQT Ventures, in which Samsung participates, to accelerate the commercialization plans of its NEO robot. For its part, the American company Apptronik presented Apollo last year, a droid designed to work alongside humans. The Canadian Sanctuary is developing Phoenix, equipped with an artificial intelligence control system, and the Spanish PAL Robotics has been working in this field for two decades.

China, meanwhile, hopes to mass produce these types of robots in 2025, as recently revealed by the Chinese Ministry of Industry and Information Technology.

This great activity is possible because in recent years significant advances have been achieved in areas such as interaction with humans, locomotion, artificial vision, movement control, object manipulation and artificial intelligence. “There has been a technological leap in the post-pandemic years,” certifies Carlos Balaguer, professor of Robotics at the Carlos III University of Madrid, who has developed his own humanoid, Teo, for research. “We will not see these types of robots on the streets in the short term, but we will see them in controlled environments,” he says.

Robotics companies envision their use in logistics centers, where they will be able to perform tasks such as unloading trucks, moving boxes and preparing orders. Although they may seem like simple tasks, getting a humanoid to open and close a box dexterously is a complex challenge, according to Balaguer.

Humanoid bipeds are also expected to work in industrial plants and perform retail jobs. “Bipedal humanoids adapt 100% to a plant. A robot with wheels can be more stable and robust, but its adoption requires enabling existing spaces,” explains Alexandre Saldes, Director of Innovation at PAL Robotics.

For example, Sanctuary AI has tested its robot in a store of the Canadian chain CTC. For a week, he carried out tasks of picking and packing merchandise, cleaning, labeling and folding. Amazon, for its part, has tested Digit from Agility Robotics – a company in which it has invested – to move empty containers. Meanwhile, BMW will test the first Figures at its South Carolina factory.

These companies argue that humanoids will free us from boring and repetitive tasks, as well as dangerous jobs. For example, they may work in mining, rescue work, nuclear reactor maintenance or chemical product manufacturing. Figure AI argues that these robots will fill the lack of labor in the United States in industries such as logistics.

The big question is when the leap will be made from the laboratory to mass production. Goldman Sachs predicts that by 2030, 250,000 humanoids will be sold, mainly for industrial use, a figure that would exceed one million units in just over a decade; predictions that Professor Balaguer considers unrealistic if the scientist’s concept of a humanoid robot is taken into account. The investment bank estimates that global business volume will reach $38 billion in 2035.

Price remains the biggest obstacle. Although the cost of components has decreased, it is still high. Elon Musk draws a hypothetical scenario with robots that can cost “like a utility vehicle.” “Not so long ago, they didn’t go below one million euros,” says Medrano. According to Goldman Sachs, the cost of manufacturing a humanoid robot is in a range of between $50,000 and $250,000 (for the most advanced versions), 40% less than a year ago.

From a technological point of view, researchers work in various fields. Goldman Sachs points out that, thanks to artificial intelligence and machine learning, robots can train themselves, which accelerates their development and allows them to perform more tasks and adapt to new situations.

Professor Carlos Balaguer indicates other areas of work in the robotics industry. For example, what is known as soft robotics; that is, creating humanoids with muscles and artificial skin for a more natural and safe interaction with humans. The Carlos III laboratory is developing robots with soft necks that can move in all directions. Work is also being done to reduce complexity and weight, for example by replacing electric motors with artificial muscles.

Finally, in the field of interaction with humans, a rising trend is the implementation of psychological perception so that the robot recognizes the emotional state of the human and can empathize with him.