Key points for this issue: 1. What is embodied AI? 2. How does technology empower? 3. Demand-side driving forces 4. How big will the future market be?

01 What is embodied AI?

Embodied AI is formed by the integration of Ontology and intelligent agents. Ontology gives it a physical form, while the intelligent agent uses algorithms to enable it to perceive, make decisions, and control movement, executing tasks through interaction with the environment, much like a sentient life form. Early robots had single functions and simple forms, only capable of performing basic movements, but now possess a certain level of intelligence; this transformation has undergone multiple technological breakthroughs and innovations.

Currently, many governments have positioned the development of humanoid robots as a national strategy. For example, China's policy focuses on breakthroughs in core component technologies and the construction of an industrial ecosystem, with the 2023 "Guiding Opinions on the Innovative Development of Humanoid Robots" clearly setting phased goals for 2025 and 2027; the USA has prioritized robotics research and development as an AI priority, with NASA and others providing funding support for research projects; Japan has launched a "moon-landing type" research and development plan to promote the integration of AI and robotics technology.

How does technology empower embodied AI?

02 How does technology empower?

First, speaking of algorithms, traditional algorithms have limitations and are quite difficult to execute skills beyond the training data, whereas multimodal large models are different. For example, the PaLM - E model, which is based on the Transformer neural network, allows robots to perform long-sequence tasks like "taking rice crackers out of the drawer."

Computing power is also very important. Currently, the multi-level computing system of Cloud-Edge-Terminal can work together. The Cloud Computing platform acts like a super brain, capable of processing a large amount of complex data; Edge Computing nodes, being close to the data source, can quickly process real-time tasks, reducing data transmission time; terminal computing power ensures that robots can make rapid local decisions, but the stability of Cloud-Edge-Terminal Communications and optimization of resource allocation still need to be resolved to ensure the system runs efficiently.

In terms of data, embodied intelligence data features multi-modal, spatiotemporal correlation, and high dimensions. The industry uses simulation to synthesize data, such as a certain chip manufacturer using simulation technology to expand data, as well as jointly building open source datasets to address data shortages. However, the differences between virtual data and real data and the enhancement of open source dataset quality remain challenges, and obtaining real diverse data is key to driving development.

What about the demand side?

03 Demand Side Drivers

First, humanoid robots are expected to enhance social efficiency. In areas such as strength, dexterity, and spatial navigation, the efficiency of humanoid robots may exceed that of humans; in public service, from commercial performances to reception, humanoid robots are already being used, and their applications are expected to expand into more complex interaction scenarios in the future; in special scenarios, a certain company's Pioneer series can already conduct inspections in hazardous environments, with great potential for applications in rescue scenarios in the future. Overall, the development trend of humanoid robots is to penetrate from specialized scenarios into general ones.

Secondly, they are expected to fill the labor gap.

The Global population growth rate is slowing down. According to the UN Population Division, the global total population growth rate in 2023 is 0.9%, a significant slowdown from 1.8% in 1980. Additionally, the proportion of the aging population is expanding, and the proportion of the labor force is decreasing. The UN Population Division predicts that in the next decade, the proportion of aging populations in major economies will continue to expand, potentially reaching over 20%. This has led to increased labor costs, and humanoid robots are expected to bridge the labor gap, improve work efficiency, and help enterprises reduce costs and increase efficiency to cope with the new labor structure in the future.

Furthermore, humanoid robots, as tools for enhancing production efficiency, are more flexible than industrial robots and can handle a wider variety of tasks, driving production flexibility and increasing production efficiency. Their development can stimulate upgrades in the Industry Chain technology; core component technology is relatively mature, but domestic capability still needs to improve. China should focus on a full industry chain layout, using humanoid robots as a lever to drive upstream component technology breakthroughs and industry upgrades.

So, how big is this market?

04 How big is the future market?

According to our predictions, with the reduction in costs, the future humanoid robot market in China will grow rapidly. By 2030, shipments are expected to reach 0.35 million units, with a compound annual growth rate of 317% from 2024 to 2030; the market size will also expand rapidly, expected to reach 58.1 billion yuan by 2030, with a compound growth rate of 259%, and is likely to become an important driver of economic growth. As the market expands, the demand for key components is expected to increase significantly, with higher requirements for supply quality and stability; coordinated development of the Industry Chain is the trend.

Embodied AI has begun to unfold, where will humanoid robots go? What surprises and changes will they bring to humanity?

Is the robotics industry expected to bounce back? Make good use of it."Investment Theme"Function, capture investment opportunities!

Editor/Somer