On December 26, the annual meeting of China Robot Network was held in Shanghai. In the new product release link, GAC Group grandly launched the third generation of self-developed embodied intelligent humanoid robot - GoMate. At the same time, the new energy vehicle giant BYD also announced the establishment of an embodied intelligence-related laboratory, dedicated to the direction of embodied intelligence, and the main task is to carry out technical reserves.
According to public statistics, nearly 10 car companies in China, such as BYD, Changan, GAC, Chery, Xiaopeng, Xiaomi, Weilai, and Ideal, have made it clear that they will enter the embodied intelligence track in the future.
What is Embodied Intelligence?
Embodied intelligence refers to artificial intelligence systems that have a physical form and can understand and interact with the physical world through human-like perception (such as sight, hearing, touch, etc.). It is not only able to perceive its surroundings, but also act on this perceived information, and has the potential to learn and evolve. From the perspective of product form, embodied intelligence is not limited to humanoid robots, but also includes wheeled, quadruped and other forms.
At present, the humanoid robot referred to by many institutions and enterprises is actually a manifestation of embodied intelligence. This kind of robot is gradually becoming a new focus of technological development.
Market forecasts show that the market size of humanoid robots is expected to reach $154 billion by 2035, equivalent to one-third of the smartphone market in 2021. Citibank is more optimistic in estimating that the global humanoid robot market will swell to $7 trillion by 2050, and the number of humanoid robots in the world is expected to reach 648 million by then.
Thanks to the broad market prospects of embodied intelligence and its natural similarity to the automotive industry in terms of R&D, production and application, this field has already sparked a boom in the automotive industry. The development of embodied intelligence technology not only brings new opportunities to the automotive industry, but also heralds a major change in the way transportation and personal mobility are carried out in the future.
Domestic automakers are also actively deploying the field of embodied intelligence
Tesla was undoubtedly one of the first companies to set foot in the field of humanoid robots. Tesla CEO Elon Musk is convinced that Optimus will be the most valuable part of the company. At Artificial Intelligence Day 2021, Tesla announced its humanoid robot project for the first time; And at the second Artificial Intelligence Day in 2022, the Optimus prototype was unveiled for the first time without any support, demonstrating its advanced state of the art.
Today, Optimus has become one of Tesla's key businesses. Musk promised that Tesla will focus on mass production and reduce costs, and the final price is expected to be less than $20,000, and the production will reach millions of units. Recently released videos show that the Optimus hand has reached 22 degrees of freedom, which is close to the 27 degrees of freedom of the human hand, and is capable of completing complex and delicate movements. In addition, Optimus has been able to walk on complex terrain with sensors and neural networks in balance while blindfolded. According to the plan, Optimus will start performing a series of useful tasks this year and sell it to the public by the end of 2025.
Not only Tesla, but also domestic automakers are also actively deploying the field of embodied intelligence, especially humanoid robots. At the 2024 China Robot Network Annual Conference, GAC Group launched the self-developed third-generation embodied intelligent humanoid robot, GoMate. This robot can not only accurately control movements, but also integrate GAC's self-developed pure vision autonomous driving algorithm, with autonomous navigation and other functions.
Zhang Aimin, head of the robot R&D team of GAC Group, said that GAC plans to achieve mass production of self-developed parts in 2025, and take the lead in applying them to the workshop production lines and industrial parks of GAC Trumpchi, Aion and other OEMs; In 2026, the whole machine will be produced in small batches, and gradually expanded to large-scale mass production.
Other traditional car companies, such as Changan Automobile and Chery Automobile, have also released their own plans or technologies in the field of embodied intelligence. In terms of new car-making forces, Xpeng and Xiaomi have already launched related products. For example, Xiaomi's full-scale humanoid robot, CyberOne, was released in August 2022, and the phased implementation plan was launched at the Xiaomi Automobile Factory in Yizhuang, Beijing, in June this year. On November 6, 2024, Xpeng released the AI humanoid robot Iron, which has 62 full-body active freedoms and 15 hand movatory freedoms, uses self-developed Turing AI chips, has a computing power of up to 3000T, and has participated in the automotive production process at the Guangzhou factory.
NIO will also set up a dedicated humanoid robot team in 2023, focusing on the research and development of underlying technologies such as algorithms, dynamic perception, and large models, to promote the application of humanoid robots in automobile manufacturing. Li Xiang, CEO of Li Auto, made it clear that Li Auto will definitely get involved in the field of robots.
Robots made by car companies: natural advantages and broad prospects
Car companies have significant natural advantages in R&D, production, and application of embodied intelligent robots, which is reflected not only in technology reuse, but also in strong industrial capabilities and a wide range of application scenarios. Tesla CEO Elon Musk has commented that the robots of other companies lack artificial intelligence capabilities, or even the ability to mass-produce robots, while Tesla has both capabilities, so it has significant leading potential in this field. This advantage is also applicable to many domestic car companies, which has become the confidence for them to set foot in the field of intelligent robots.
Advantages of R&D
Car companies usually have a strong independent R&D system, and have accumulated rich R&D experience and professional technical personnel in the long-term development process. For example, Xpeng Motors' humanoid robot Iron is based on its end-to-end autonomous driving model, which has natural language processing, visual and tactile multimodal interaction capabilities, and generalization capabilities to adapt to multiple scenarios. The large model endows the humanoid robot with "universal cognition", improves its intelligence level, and enables it to perceive, cognitive, make decisions and perform complex tasks, providing the possibility of landing embodied intelligence.
The advantages of manufacturing
From the perspective of manufacturing, car companies have strong industrial capacity, especially the head enterprises have established a complete supply chain system and strong production capacity. These advantages enable automakers to accelerate the commercialization process while reducing the cost of R&D and production of humanoid robots. The relevant person in charge of GAC Group said that since GAC launched the research and development of embodied intelligent robots, it has made full use of the mature industrial chain of intelligent networked new energy vehicles, and shared R&D costs by sharing vehicle-end chips, lidars and other components, thereby effectively reducing the overall development costs.
App-side benefits
Globally, although many manufacturers have launched humanoid robot products, the application scenarios are still in a very early stage, resulting in slow progress in product upgrades and iterations. In contrast, car companies have a large number of ready-made robot application scenarios. Taking Tesla as an example, Optimus was designed for industrial manufacturing scenarios at the beginning, and plans to first apply it to Tesla's Gigafactory, with its own robot factory application scenarios. This means that Tesla can quickly test and optimize robot performance in a real-world operating environment, promoting iterative updates and technological advancements in products.
Reuse industrial and autonomous driving capabilities
It is generally believed in the industry that car companies can partially reuse the capabilities they have accumulated in the field of industrial and autonomous driving in the field of embodied intelligence. For example, functions such as sensor fusion, path planning, and environmental perception in autonomous driving technology can be directly applied to the development of humanoid robots, enabling robots to understand and interact with the physical world more intelligently. In addition, the experience of car companies in mass production and quality management will also help ensure the stability and reliability of humanoid robots.