Industry background and investment logic

Humanoid robots: high degree of generalization, high integration and intelligence

There is currently no universal definition of a humanoid robot , but according to Humanoid Robots" concludes that humanoid robots should be able to "work in the environment where people work and live, operate tools and equipment designed for people, and communicate with people".

Under this premise, humanoid robots should eventually have a human-like body structure, including a head, torso, and limbs, walk bipedally, perform various actions with multi-fingered hands, and have a certain degree of cognitive and decision-making intelligence. The technical difficulty lies in imitating the process of "perception-cognition-decision-execution" in various scenarios of human beings as much as possible, involving cutting-edge technologies in various fields such as bionic perception and cognition technology, biomechanical and electrical integration technology, artificial intelligence technology, big data cloud computing technology, and depth of vision navigation technology.

Tesla's Optimus iterates rapidly, leading the mass production of humanoid robots

In August 2021, Musk was in AI After two years of technical iteration, TeslaBot has been able to classify objects, yoga, and grasp objects smoothly. In January 2024, Musk disclosed at the performance exchange meeting that Tesla's humanoid robot will be delivered in 2025. Analogous to the automotive industry, Tesla's humanoid robot is expected to issue actuator fixed-point to Tier1 in 2024.

In 2023, the humanoid robot industry is still in the early stage of R&D tests and customer sample delivery, and in January 2024, Sanhua Intelligent Control and Tuopu Group have successively increased investment in robot projects to verify that the industry has initially met the industrial conditions for the transition from R&D test to mass production.

General intelligent robots are one of the core hardware of the next round of change

Humanoid robots are not false propositions: integration and generalization are the inevitable trend of technological development

Humanoid robot prospect prediction

Musk has said that his vision for designing Tesla's robots is to make them serve thousands of households, such as cooking, mowing lawns, and caring for the elderly. It is expected that in 2025, humanoid robots will be expected to achieve breakthroughs in the application of manufacturing scenarios, and small batches will be used in electronics, automobiles and other manufacturing environments.

In January 2024, Musk agreed with the prediction of David ·Holtz, founder of the artificial intelligence research lab Midjourney, on social media. Holtz has predicted that there will be 1 billion humanoid robots on Earth by the 2040s, and by the 2060s, there will be 100 billion robots (most of them extraterrestrial) robots in the entire solar system. In November 2023, Brett was the founder of Figure Adcock has said in an interview that in the next few decades, the real challenge will be how to scale up the production of humanoid robots, and how to supply a sufficient number of humanoid robots to the market.

In the long run, every person and every family will want a humanoid robot, just like a car or a mobile phone is now. And there will also be billions of humanoid robots in the labor market to do jobs that are dangerous, monotonous and tedious for humans.

Humanoid robots will also be used in aviation, such as building human settlements in space and caring for the elderly. If we were given enough time to do mass manufacturing, we could build billions of humanoid robots.

|Logical attribution of the dominance of the domestic industry

Representative industries: photovoltaics, batteries, energy storage, new energy vehicles, smart phones, real estate, infrastructure

Negative industries: chip manufacturing (technology/industrial chain), high-end software (technology/scale effect)

The industrial policy is positioned as a "disruptive product", and Beijing and Shanghai are the first to explore

The positioning is high enough, positioning the humanoid robot as a subversive product after computers, smart phones, and new energy vehicles, and raising the strategic height of the humanoid robot, highlighting the country's attention to the humanoid robot industry.

In the early stage of policy layout, it mainly played a leading role. At present, it is mainly Beijing, Shanghai, Shenzhen and other core cities that have made plans and successively introduced relevant policies on the construction of humanoid robots, and the industrial policy is also in the exploratory period.

Sparks, AI algorithm models accelerate the transformation of the industry 0-1

There are three unsolved problems in humanoid robots: the brain, the cerebellum, and the proprioception.

The so-called brain, refers to the robot's comprehension ability, that is, the robot's understanding of human instructions and environmental perception, the emergence of AI large models mainly improves the ability of the robot's brain, but it is still independent of the existence of the ontology, how to really "access" the body to play a generalized function still needs to be explored.

The cerebellum refers to the robot's refined motion control ability, which is currently more at the imitation level of fixed motion mode, and there is still great room for improvement in unstructured long scenes and intelligence.

The ontology is whether the various parts that make up the prototype of the humanoid robot, such as joints, limbs, heads, etc., can fully undertake the action instructions transmitted by the software algorithm. It is only a matter of time before the hardware part matures, and the core pain point is cost reduction.

Migration of advantages of mature automotive industry chain

Elon Musk: Tesla is the world's largest robot/semi-intelligent robot – a robot on wheels. New energy vehicle enterprises and humanoid robot products are interconnected at some technical levels, and both have the feasibility of technology migration from hardware to software. The development of humanoid robots by car companies can make full use of existing resources to reduce manufacturing costs and shorten the development process.

For example, at the hardware level, the computing power chips, control boards, vision sensors, high-performance batteries and energy management systems used in new energy vehicles can be transplanted to humanoid robot products, and the lightweight design and production experience of auto parts can also help the production of humanoid robots; At the software level, humanoid robots can "stand on the shoulders of giants" and be customized and optimized based on relatively mature algorithms such as environmental perception, target recognition, path planning, and safety protection for new energy vehicles.

Upstream hardware, midstream integration, and downstream applications all have scale effects

Upstream hardware: The upstream is mainly suppliers of components and software systems, including motors, joints, sensors, controllers, operating systems, etc., and hardware products have scale effects, which are expected to significantly reduce costs by expanding scale; The software is relatively at a disadvantage, focusing on the breakthrough of the AI general robot model, and the country is in a backward position.

Midstream integration: A strong manufacturing foundation provides integration advantages, and at the same time, it is easy to give full play to the scale effect, form bargaining power over the upstream and downstream, and it is easy to continue to lead once it runs out of commercial advantages.

Downstream applications: Abundant and large-scale B-end industrial application scenarios, such as automobile manufacturing, high-risk inspection, etc., provide enough opportunities for product upgrades and iterations, and provide a large enough market to attract capital for continuous investment.