Who will win the race to develop a humanoid robot?

On a bright spring morning in Hanover, Germany, a humanoid robot named G1 was making waves at the Hannover Messe, one of the largest industrial trade shows in the world. Created by the Chinese company Unitree, the G1 stands at about 4 feet 3 inches tall, making it smaller and more affordable than many other humanoid robots available today. The G1 is known for its impressive range of motion and dexterity, which has captivated audiences at various trade shows. As I arrived at the event, I saw the G1 being controlled remotely by Pedro Zheng, the sales manager at Unitree. He explained that customers need to program the G1 for it to perform tasks on its own. Many attendees were eager to interact with the G1, reaching out to shake its hand and laughing when it waved back or bent backward. The human-like appearance of the G1 seemed to put people at ease, making it stand out among the other machines at the conference. Unitree is just one of many companies around the globe working on humanoid robots. The potential for these robots is enormous, as they could provide a workforce that does not require breaks or salary increases. Additionally, they could serve as the ultimate household helper, capable of doing chores like laundry and dishwashing. However, the technology is still in its early stages. While robotic arms and mobile robots have been utilized in factories and warehouses for years, introducing a humanoid robot into unpredictable environments, such as homes or restaurants, presents significant challenges. For humanoid robots to be effective, they must be strong, but this strength can also pose risks. A simple fall could lead to dangerous situations. There is still much work to be done on the artificial intelligence that will control these machines. A spokesperson from Unitree mentioned, 'The AI simply has not yet reached a breakthrough moment. ' They explained that current robot AI struggles with basic logic and reasoning, making it difficult for robots to understand and complete complex tasks. At present, the G1 is primarily marketed to research institutions and tech companies, which can utilize Unitree's open-source software for development. Many entrepreneurs are currently focusing on humanoid robots for warehouses and factories. One of the most notable figures in this field is Elon Musk, who is developing a humanoid robot called Optimus through his car company, Tesla. Musk has stated that 'several thousand' of these robots will be built this year, and he expects them to perform 'useful things' in Tesla factories. Other car manufacturers are also exploring similar paths, with BMW working on robots for their US factory and South Korean company Hyundai acquiring robots from Boston Dynamics, a firm it purchased in 2021. Thomas Andersson, the founder of a research firm called STIQ, is tracking 49 companies that are developing humanoid robots, which are defined as robots with two arms and two legs. If the definition is broadened to include robots with two arms that move on wheels, the number of companies rises to over 100. Andersson believes that Chinese companies are likely to dominate the humanoid robot market due to their strong supply chain and ecosystem for robotics. He stated, 'The supply chain and the entire ecosystem for robotics is huge in China, and it's really easy to iterate developments and do R&D. ' Unitree exemplifies this advantage, as its G1 is priced at $16,000, which is relatively affordable for a robot. Furthermore, Andersson pointed out that nearly 60% of all funding for humanoid robots has been raised in Asia, with the US attracting most of the remaining investment. Chinese companies also benefit from support from their national and local governments. For instance, in Shanghai, there is a facility dedicated to robots, where numerous humanoid robots are learning to complete various tasks. This positions Chinese firms well to lead the humanoid robot market. But how can US and European robot manufacturers compete with this? Bren Pierce, a Bristol-based entrepreneur who has founded three robotics companies, has recently launched the KR1 robot. Although the KR1 was designed and developed in the UK, it will be manufactured in Asia. Pierce explained, 'The problem you get as a European or American company is that you have to buy all these sub-components from China in the first place. ' He noted that it would be inefficient to transport all the parts halfway around the world when they could be assembled closer to the source in Asia. To keep costs down, Pierce has opted not to create a fully humanoid robot. The KR1 is designed for warehouses and factories and does not have legs, as most of these environments have flat floors. He believes that adding legs would only increase costs and complexity. Instead, the KR1 is built with mass-produced components, such as wheels similar to those found on electric scooters. Pierce's philosophy is to use as many off-the-shelf parts as possible, ensuring that all motors, batteries, computers, and cameras are commercially available and mass-produced. Like his competitors at Unitree, Pierce emphasizes that the real 'secret sauce' lies in the software that enables the robot to work alongside humans. He remarked, 'A lot of companies come out with very high-tech robots, but then you start needing a PhD in robotics to be able to actually install it and use it. ' His goal is to design a robot that is easy to use, allowing an average warehouse or factory worker to learn how to operate it within a few hours. The KR1 can learn to perform a task after being guided through it by a human 20 or 30 times. This year, the KR1 will be given to pilot customers for testing. But will robots ever make their way into our homes? Even the optimistic Pierce acknowledges that it is still a long way off. He shared, 'My long-term dream for the last 20 years has been building the everything robot. This is what I was doing my PhD work in. I do think that is the end goal, but it's a very complicated task. ' He believes that eventually, robots will be in our homes, but that could take at least 10 to 15 years.