“Investors should remain disciplined and back companies that have realistic goals based on economics, not hype,” warns Daiva Rakauskaitė of Aneli Capital, as investment dollars flow into humanoid robotics that have expectations unrealistic and quite impracticable.
It would lead a significant rise in humanoid robot investments, leading to the development of artificial intelligence that would eventually help move automation to spaces designed for human presence. Development of intelligent systems in robots has so far made it possible for humanoid robots to move like humans. Currently, research institutions are working on improving capabilities of robots and making them as nimble as humans. At ETH Zurich’s Soft Robotics Lab, engineers developed humanoid robot hands that use artificial tendons instead of motors. The hands of the humanoid robot have 21 degrees of freedom and were trained by reinforcement and imitation learning. This will help the robots perform in unstructured spaces, meaning that the humanoid robots will equally be able to perform outside the developed factory floors.
However, there is still a huge gap between these research models and what can be called fully functional humanoid robots. This is because, as robotics pioneer and Roomba inventor Rodney Brooks argues, contemporary models of biped robots are even dangerous when they co-exist with humans. This is because full-scale humanoids’ dynamics mean that even if they fall, they can unleash forces that could be harmful to people. In addition, as Brooks argues persuasively, because today’s “vision-only” model driven by video recordings of human tasks being pursued by companies such as Tesla Motors and Figure has remained the norm even as far as robotics innovation is concerned, robots cannot be dexterous enough if they do not incorporate the use of the human hand’s touch. The human hand has 17,000 mechanoreceptors.
Voices such as those from China’s National Development & Reform Commission have reiterated warnings of over-duplication in the 150-plus companies making humanoid robots in the country. It would seem there is some trend where multiple companies have come out with similar offerings, driving out investments in basic tech development, something which has happened in every bubble market trend, say the bike-sharing economy. Even with estimates by Citigroup that the humanoid market would be worth a staggering $7 trillion by 2050, the government in Beijing seems to be urging their companies toward sustainable development.
From an engineering standpoint, the scope is well understood, where real-time inference, dexterity, robustness, and affordability continue to hold humanoid robots within expected settings like warehouses. Even the most sophisticated models developed within cloud simulations of cybernetic systems struggle with autonomy if there isn’t web accessibility, hence triggering hybrids that implement their computing capacity within the robots. Advances on musculoskeletal models and soft robots remain on promising tracks but are not market-ready.
The challenge is not trivial for investors either. Humanoid start-ups rarely had short-term monetization models but based their ideas on long-term replacements for millions of hazardous or undesirable jobs. According to Rakauskaitė, this approach to generating revenue in start-ups has been well-proven in industrial robotics, though so far not in humanoids, without proper allocation it may end in the same pattern as in previous tech bubbles, where valuations burst before technologies were fully developed.
The competitive scene is also complex in this arena, as although U.S.-based players such as Figure AI, Apptronik, and Agility Robotics have accumulated hundreds of millions of dollars in investments, China’s government-supported investment (estimated at $138 billion in the robotics boom) is massive compared to what is spent in America. Since humanoid robots boost not only economic growth but also act as an instrument of military and industrial might, AL algorithms are continuously being developed along with tendon-driven actuators and cloud-based reinforcement learning. However, the ongoing enhancement of humanoids will eventually reach maturity, but there are considerations based on industry that need pragmatic analysis.
In this respect, an astutely tech-conscious investor will find opportunity within the dichotomy which splits hype-driven investments seeking to speculate on frothy valuations versus investors focused on core technologies of sensing, security, or monetization to shape the future humanoid robots that will create real value.
