In 2004, Lei Jun invested in Xiaohuang to develop robots.

In 2023, Huang Yi, a freshman at the time, won the first prize in the National College Students' Science and Technology Competition with his land - air amphibious drone project. When asked about his future career plan, he said, "I hope to found a robotics company and use technology to change lives and influence the world."

At the end of the same year, Huang Yi spent only dozens of days in his dormitory to build the bipedal robot AlexBot with a cost of less than 20,000 yuan, and simultaneously completed full - stack open - sourcing on platforms such as Zhihu, GitHub, and Feishu. At that time, he was 19 years old and could independently achieve full - stack development of the robot's walking ability from scratch.

In 2025, Huang Yi founded RoboParty in Shanghai, focusing on the research and development of fully open - source bipedal humanoid robots. Currently, its users cover university scientific research institutions, robotics startups, industrial inspection research institutions, etc. More than a dozen enterprises and universities have replicated its technical solutions.

On November 12th, RoboParty announced the completion of a seed - round financing of tens of millions of dollars (the largest in the domestic bipedal humanoid robot body field), jointly led by Matrix Partners China and Xiaomi Strategic Investment, with participation from L2F Light Source Entrepreneurs Fund and Galaxy General. The funds will be mainly used for the independent research and development of core components and motion control technologies.

As of now, although RoboParty has not officially mass - produced and delivered products, it has already rejected hundreds of intention orders for robots, hoping to prioritize the construction of the technical foundation and community ecosystem.

01

RoboParty was founded in 2025. Its founder, Huang Yi, was born in 2004 and entered Harbin Institute of Technology in 2023. He created the AlexBot series of bipedal robots during his undergraduate studies.

Instead of keeping these robots private, he chose to open - source them, making the entire engineering process, codes, and documents public. He has accumulated more than 4,000 stars on GitHub, and the document views have exceeded 200,000. In his early years, Huang Yi also won awards in college science and technology competitions with his land - air amphibious drone project. He once said that he was willing to "use technology to change lives and influence the world."

Traditional humanoid robot projects usually have extremely high technical barriers: closed - source hardware, proprietary protocols, non - standard designs, and high prices, making it difficult for developers to enter. On the other hand, many companies are eager to mass - produce and pursue orders but neglect the construction of the underlying architecture and ecosystem. Huang Yi believes that open - source is the key to solving these pain points. Open - source can lower the threshold, allow more people to participate, and promote underlying standardization through community - accumulated experience, reducing redundant development.

In February this year, sponsored by Fourier, Huang Yi completed the iterative version AlexBotMini. In March, Huang Yi graduated early from his junior year and immediately founded RoboParty, focusing on the research and development of fully open - source bipedal humanoid robots. It is reported that Huang Yi is also the youngest CEO of a humanoid robot enterprise in China.

After its establishment, the team invested millions of yuan in underlying research and development. It is worth mentioning that in the first week of starting the financing, RoboParty received an intention order for hundreds of robots from a listed company. But Huang Yi chose to reject it. He is not in a hurry to expand production but prioritizes building the technical foundation and community ecosystem. He once said that before the technology matures, it is more important to attract high - quality developers than to blindly pursue quantity.

In terms of technology layout, RoboParty has developed the ATOM01 and ATOM02 - M series of prototype robots and plans to officially release ATOM02 in the second quarter of 2026 and open - source the entire stack.

As of now, although RoboParty is still in its early stage, its influence is rapidly accumulating. Its founding team is young, has full - stack technology, and the open - source community is active, all of which are its core competitiveness.

02

The biggest difference of RoboParty lies in its fully open - source positioning. Most humanoid robot companies prefer a closed - source commercial path, emphasizing their hardware and algorithm advantages. In contrast, RoboParty is like building a developer ecosystem, opening up all modules such as machinery, electronics control, algorithms, and training.

This strategy may not generate high profits in the early stage, but in the long run, it is expected to accelerate iteration through community co - construction and build platform - level barriers in the future. As Huang Yi said, they are building a "next - generation computing platform." They are not just selling robots but also creating an ecological foundation for developers and users to participate in.

This industry is being driven by several major technological trends.

Firstly, it is the combination of large models and multi - modal AI. As the combination of language models, vision models, and reinforcement learning matures, humanoid robots can understand more complex instructions, make plans, and interact more naturally.

Secondly, it is modular hardware and standard protocols. Open - source and standardization can lower the development threshold and accelerate industrialization.

Thirdly, it is the sim - to - real technology. The closed - loop between simulation training and real - world deployment is getting closer, enabling robots to adapt to real environments faster.

Finally, with the construction of training bases and data centers (such as training centers in Shanghai), the industry will accumulate more physical interaction data, providing nutrients for intelligence.

Through open - source, RoboParty can attract researchers and developers to participate, feeding back simulation and real - world data into the system. Through modular design, they can quickly upgrade hardware in the future. Through the community, they can gradually build their own user base and ecosystem, which will contribute to sustainable expansion in the future commercialization stage.

03

All along, bipedal robots have been the most "hardcore" direction in the robotics field. Although two legs make the robot look closer to human form, it means exponentially increasing engineering difficulty. The robot needs to maintain balance continuously in a series of uncertain environments. With each step, it has to handle complex variables such as center - of - gravity shift, ground - contact feedback, and inertia change.

Another layer of difficulty lies in intelligence. It is often said in the industry that current robots "can walk but cannot do things." Motion control can be achieved through traditional algorithms, but for robots to understand the world, handle tasks, and explore paths independently in unfamiliar tasks, they must rely on stronger decision - making intelligence, which has been difficult to break through in the past decade.

Although various companies are working on motion libraries and task modeling, when faced with real environments, irregular objects, and sudden changes, these capabilities often perform far below expectations. They are precise but mechanical, powerful but inflexible. Moreover, the cost issue has long kept the industry in a stage where robots "can be made but cannot be used."

Against this background, the industry has instead welcomed an optimistic mood like "before dawn": Artificial intelligence has finally filled the long - missing piece of the puzzle for robots. The rapid maturity of technologies such as large models has enabled robots to have learning and understanding abilities. Robots are not programmed but trained, and they begin to have generalization ability, that is, they can try to solve tasks that have never appeared before.

The maturity of supply - chain capabilities has also opened a historical window for this field. The prices of key components such as high - torque actuators, lightweight materials, low - cost sensors, and compact motors have significantly decreased in the past three years, making bipedal robots transform from "expensive toys" into "mass - producible products" for the first time.

Although today's bipedal robots still seem far from large - scale commercial use, some believe that the industry already has the early characteristics similar to those of smartphones in 2007: the technology is feasible, the cost is starting to decline, the ecosystem is taking shape, the number of developers is increasing, and software frameworks and hardware standards are gradually being unified.

At this time, new companies like RoboParty, which choose an open - source - centered path, are actually seizing a longer - term track. Instead of starting from selling robots, they start from building an ecosystem, gathering developers, and lowering the industry threshold. For an industry still in its early stage, this may be the most leveraged approach - first build the entire industry's infrastructure and let later - comers continue to develop on its "foundation."

This article does not constitute any investment advice.

This article is from the WeChat official account "Pencil News" (ID: pencilnews), written by Song Ge and published by 36Kr with authorization.