The 2024 China Embodied Intelligence Conference (CEAI 2024) will be held on the west bank of Xuhui from March 30th to 31st. The theme of this conference is "embodied symbiosis, intelligent shaping of the future". On the first day of the opening ceremony, the Shanghai embodied intelligent industry and innovation alliance was officially launched, with initial members covering upstream and downstream enterprises in the industry chain, as well as government, universities, and research institutions.
According to the definition of the Chinese Computer Society, Embodied Artificial Intelligence involves disciplines such as artificial intelligence, robot learning, and computer vision. Essentially, it emphasizes the research paradigm of artificial intelligence where intelligent agents with physical bodies interact with the physical environment to obtain intelligence.
Currently, as a cutting-edge hot topic in the field of artificial intelligence, embodied intelligence is gradually moving from theory to practice, and from the laboratory to reality. But many people still find this huge potential "blue ocean" market unfamiliar. What can embodied intelligence do, change, and how does it relate to larger models? Behind these questions, it indicates that embodied intelligence, representing the symbiosis and mutual empowerment of software and hardware, is rapidly entering human production and life.
On site of the conference
Why is it the present
At the 2023 World Artificial Intelligence Conference, Generative Artificial Intelligence (AIGC) has become the absolute protagonist, and the attention to embodied intelligence has also skyrocketed. But non industry professionals are confused, "I've heard of it for the first time."
What kind of "intelligence" is embodied intelligence? Professor Lu Cewu from Shanghai Jiao Tong University provided a definition. "Simply put, it is intelligence with a body, and computers can be seen as representatives of 'intelligence without a body'. In his view, focusing on embodied intelligence is now the right time.". From a scientific perspective, the level of human understanding of oneself, vision, and motion control technology can be combined to form a scientific category around embodied intelligence. From an industrial perspective, the software and hardware technologies involved in embodied intelligence have initially matured, and the entire industry is poised for development.
Dai Qionghai, an academician of the CAE Member and chairman of the Chinese Association of Artificial Intelligence (CAAI), said that the government work report this year pointed out that it would deepen the research and development of big data, artificial intelligence and other applications, and carry out the "artificial intelligence+" action. Embodied intelligence is precisely the forefront direction of current artificial intelligence technology towards the physical world. It is not only an integrated application of machine learning, natural language processing, computer vision and other technologies, but also a challenge and exploration of humanoid abilities and intelligence.
The explosive point is right now. Jiang Changjun, an academician of the CAE Member, said that enterprises represented by OpenAI used a large amount of Internet text and image data to build big language models such as ChatGPT, which demonstrated AI's ability to solve multiple tasks for the first time, and also showed human beings the potential to overcome domain constraints and realize general artificial intelligence (AGI).
"Although big models such as ChatGPT have demonstrated the ability to handle multimodal data and meet human needs, this interaction is still limited to the digital world." Jiang Changjun said that to achieve truly universal artificial intelligence, AI needs to be endowed with the ability to interact in the real physical world. This means that AI needs to have entities that possess the ability to perceive, think, and act like humans.
Pine Spirit Robot
As a combination of two cutting-edge technologies, artificial intelligence and robotics, embodied intelligence is considered an important component of new productivity. Last May, Huang Renxun, CEO of chip manufacturer Nvidia, also stated that the next wave of artificial intelligence would be embodied intelligence, which is intelligent systems that can understand, reason, and interact with the physical world.
As for the current trend of "hidden explosion" of embodied intelligence, Jiang Shuqiang, a researcher at the Institute of Computing Technology of the Chinese Academy of Sciences, believes that ChatGPT, Sora and other models relying on Internet data training tend to be "detached intelligence", and embodied intelligence needs to enter the real physical world, empower various industries from civil life to aerospace, and give feedback to the real environment, which means that every industry will have a place to use embodied intelligence.
Association with large models
A week before the embodied intelligence conference, the Global Developer Pioneer Conference was also held at the same location on the west bank of Xuhui. The discussion on the big model a week ago has not yet dissipated, and the "new explosive points" of embodied intelligence have appeared one after another. What is the correlation between embodied intelligence and large models?
Lu Cewu stated that compared to viewing large models as the "brain" of embodied intelligence, it is more like a tool. Just like solving mechanical problems requires Newton's laws, embodied intelligence also has its own 'solving laws', that is, its own big model. However, unlike big language models and visual models, the training cost for real scenes is high and it is impossible to run at high speed 24 hours a day like in the digital world. Therefore, the embodied intelligence large model is bound to have a new training framework, which is exactly the direction that Lu Cewu and his team at Shanghai Jiao Tong University are researching.
Songling Robot Shushu Photography
On the same day of the conference, the attending guests mentioned that in the previous round, North America achieved a significant leading position in the field of generative artificial intelligence represented by ChatGPT, Sora, and others. But in the field of embodied intelligence, although China's technology started and emerged about three years later than North America, currently the world is basically at the same starting line of technology, and they are all in the "pre explosion period".
In Lu Cewu's view, China currently has two major advantages. One is the manufacturing of robot hardware, and the other is the number and richness of application scenarios. "The warehousing and logistics, food processing, catering, and medical service industries have all gradually adopted embodied intelligence. Perhaps in 20 years, people will have home care robots to take care of themselves and their families, all of which are embodied intelligence."
How to turn these advantages into a prosperous era requires joint efforts from all sectors of industry, academia, and research.
The reporter noticed that in the display area of the embodied intelligence conference, many companies have moved out their "house keeping treasures", including customer service robots that can communicate smoothly, quadruped robots that adapt to complex terrain environments, and industrial robots that match various assembly line operations. Robots equipped with AI large models, known as "brains," can engage in various tasks such as household services, elderly care, education and medical care, facility inspections, and disaster relief, showcasing the latest application achievements of embodied intelligence in general scenarios, professional fields, vertical industries, and other fields.
At the booth of Shanghai Fourier Intelligent Technology Co., Ltd., a dark gray robot with a handsome metallic luster is waving its arms. This robot is called GR-1, which stands for General Robotics. The on-site staff introduced that currently, the enterprise focuses on the robot body, and has core technological advantages in hardware structure and motion control algorithms.
The intelligence of this embodied intelligent product is mainly reflected in two aspects: athletic ability and brain interaction ability. In terms of sports, it can complete movements such as jumping and running, as well as waving, walking, and dancing, which can meet the needs of different customers in different scenarios. "We open interfaces, and customers only need to complete the upper level design based on our platform, so that robots can complete the corresponding work. It has been applied in industrial production and education fields."
Fourier universal humanoid robot
In the more vertical field, coffee robots and ice cream robots have been active in multiple scenarios such as hotels and social dining. Lian Wenjie, Sales Director of Shanghai Lebai Robotics Co., Ltd., introduced that the company's seven axis retail robots and six axis collaborative robots have core technological advantages in the field of grabbing items weighing less than 3 kilograms. In hotels and restaurants such as Marriott and Quanji, Lebai's noodle cooking robot has served for more than 8800 days, penetrating people's daily lives.
How Shanghai takes action
Shanghai is the earliest pioneer in the field of embodied intelligence in China. The Action Plan for Promoting High Quality Innovation and Development of the Intelligent Robot Industry in Shanghai (2023-2025) issued last year proposed to focus on tackling advanced technologies such as embodied intelligence. Luo Dajin, Director of the Shanghai Municipal Commission of Science and Technology, stated at the meeting that Shanghai will accelerate the breakthrough of key technologies and underlying technologies such as embodied intelligence and humanoid robots, and accelerate its move towards a technology innovation center with global influence.
As an important carrier area for Shanghai's construction of a globally influential science and technology innovation center, Xuhui District has recently launched a "combination punch" to support the development of embodied intelligence, including the release of embodied intelligence industry policies and the joint creation of Shanghai's embodied intelligence innovation center with the 21st Research Institute of China Electronics Technology. The total output value of the artificial intelligence industry in the district has exceeded 80 billion yuan.
The Supporting Opinions of Xuhui District on Accelerating the Development of embodied intelligence Industry (Trial) released in February this year is the first district level support policy for embodied intelligence industry in Shanghai. Yu Linwei, Deputy District Mayor of Xuhui District, stated that last year Xuhui seized the opportunity of large-scale model development and jointly built the first and only large-scale model innovation ecological community in Shanghai, the "Model Speed Space". Within six months, more than 60 enterprises and institutions settled in the first 20000 square meter carrier.
Modal velocity space relaxation
On this basis, Xuhui seized the opportunity to pursue and closely follow the strategic opportunity of upgrading and leapfrog development of intelligent robots this year. It took the lead in launching embodied intelligence industry support policies in the city, focusing on key technology breakthroughs, open collaborative innovation, high-quality incubator construction, expanding "embodied intelligence+" applications, and promoting industrial development in 10 aspects. It supports major original technological innovation in intelligent terminals such as humanoid robots and their application fields, serving the incubation and growth of hard technology enterprises.
"We need to tackle three dimensions of challenges now, namely lowering supply chain prices, increasing market share, and increasing revenue and performance." The "three small goals" mentioned by Lian Wenjie represent the true idea of many small and medium-sized artificial intelligence enterprises: the government provides abundant sunshine and rain, and enterprises are responsible for working hard to grow and thrive.
The reporter noticed that the support opinions for the embodied intelligence industry released by Xuhui have multiple highlights in terms of financial support and support areas, catering to enterprises of different scales. For example, projects that lead the development of embodied intelligence or achieve disruptive breakthroughs will be supported up to 10 million yuan, and the same amount of support will also apply to embodied intelligence products that achieve certain commercial results. Xuhui will also provide up to 5 million yuan in subsidies to enterprises for the application of industry-leading and demonstrative technologies in scene construction, or for digital and intelligent transformation projects that meet the "industrial upgrading" standards. Certified high-quality projects in smart spaces can enjoy a maximum of 30 million yuan in municipal special subsidies.
Lu Cewu stated that the development cycle of large models and robot products related to embodied intelligence is relatively long, and government departments need to provide more support and attention to technology research and development investment and development teams in the initial stage. "'Everywhere is a robot 'is a big investment, but as long as we can overcome the hurdle of the initial stage and start scaling up, the snowball will roll bigger and bigger."