In June 1984, robotics researcher Marc Raibert and two colleagues published a paper on a robot with only one leg.
The robot relied solely on a single elastic leg, repeatedly jumping on an open floor. Researchers attempted to control it from three aspects: direction, posture, and bounce height. The paper demonstrated that jumping could be systematically controlled, but limited by the technological level of that era, the robot always dragged hydraulic lines and data cables, connecting to an external power source and computer, unable to operate independently outside the lab.
Two years later, Raibert incorporated this research into "Legged Robots That Balance." He started with a single leg, then applied the same principles of motion to bipeds and quadrupeds. In 1992, he founded Boston Dynamics out of the MIT Leg Lab. Spot was not fully commercialized until 2020.
It took over thirty years from the first single-legged robot to jump to a quadruped robot being purchased by customers.
In August 2026, we witnessed another single-legged robot that could achieve stable continuous jumps without relying on external devices, and it took less than two years from paper publication to mass production. It is composed of a carbon fiber elastic leg and four rotors. Upon landing, it compresses a spring, and after takeoff, it uses the rotors to replenish energy, adjust posture, and even fly directly over impassable chasms. The research team at City University of Hong Kong named it Hopcopter, and the Hong Kong-based LeapTech gave it a more direct Chinese name, "Jumping Machine."
LeapTech originated from the City University of Hong Kong's Robotics and Intelligent Systems Laboratory. In 2024, Bai Songnan and others' paper on the Hopcopter was published in "Science Robotics." Wang Fangzheng then brought together senior brother Bai Songnan and mentor Professor Pakpong Chirarattananon to turn the research project into a company. The company has received a million-dollar seed investment exclusively from InnoFund and is advancing a new round of financing. The first-generation remote-controlled product is aimed at enthusiasts, with a future release of a version for developers for further development.
Our simple question of curiosity is, "What is this thing for?"
Wang Fangzheng admits that LeapTech is holding a big hammer and is still looking for the nail that suits it best. He also insists that jumping will become a distinct form of robotics. When investors ask about profits and PMF, he discusses the principle of motion and future visions. In conclusion, he says:
“More importantly, is to live a decade later.”
A Spring
Wang Fangzheng, born in 1999, obtained his master's degree in Human and Bio-robotics from Imperial College London and later pursued a Ph.D. at City University of Hong Kong. The initial technology of Stargazer Tech came from his senior fellow, Bai Songnan. Bai Songnan and the team abstracted the process of an animal running, landing, compressing, and taking off again into an inverted pendulum spring, then attached a passive telescopic leg underneath a quadcopter.
This path carried a strong laboratory atmosphere, and all the commercial issues that Stargazer Tech later faced started from this spring.
Dynamic Beating: First, please introduce yourself and how Stargazer Tech started.
Wang Fangzheng: My name is Wang Fangzheng, I graduated with a master's degree from Imperial College London, focusing on robotics. After graduating, I joined City University of Hong Kong to pursue a Ph.D. under Professor Pakpong Chirarattananon. Here, I met Dr. Bai Songnan, my senior fellow, and was introduced to his research on the Hopcopter, also known as the jumping robot.
I found this to be a very meaningful research outcome with a lot of future possibilities, perhaps becoming a new form of robot. Later on, my senior fellow, advisor, and I founded Stargazer Tech and have been working on it ever since.
Dynamic Beating: What problem did the lab initially want to solve, and why did it end up creating a jumping robot?
Wang Fangzheng: The lab we were in is a world-class drone lab that has explored many forms. Some drones can roll on the ground like wheels, some can spin and float like seeds, some are flown with a motor attached to a rope, and we've also worked on drones that can transform, combine, and disassemble.
Drones cannot escape gravity as long as they are in the air, making it challenging to balance payload and endurance simultaneously. We thought of using gravity to make the robot bounce upon landing. This way, both payload and endurance could be significantly improved.
A Machine 'Person' Doesn't Have to Resemble a Human
The biomimicry of the jumping robot lies in its motion. It does not replicate the shape of any specific animal but only captures the moment when a human or animal lands, compresses, and pushes off while running. Wang Fangzheng believes that the form a robot adopts should be determined by the task, cost, and efficiency.
This judgment also dictates how Stargazer Tech describes its technology. It is primarily a mode of transportation responsible for taking the machine to places that wheeled and legged robots find challenging to reach. Mechanical arms, sensors, and large models can be considered later.
Jumping Robot: Why do you think many tasks do not need to be assigned to humanoid robots?
Wang Fangzheng: I have always believed that robots do not need to look like humans. Of course, there are advantages to humanoid robots. They can dance, and many real-world environments are designed around the human body, making it easier for humanoids to adapt to these environments.
If a factory needs to tighten screws, you can either have a humanoid robot reach out to tighten them, or you can have a screwdriver come down from above to do the job, and then the assembly line can move the product to the next station. I prefer the latter approach because cost and efficiency need to be considered.
Many people say that the jumping robot is not biomimetic, and I must correct this. It is highly biomimetic; it just doesn't look like familiar animals to us. It mimics the moment when a person is running and either landing or taking off. We call this motion model an inverted pendulum with a spring. Biomimicry only needs to capture the core features. Humans create radars without needing to shape them like bat ears.
Jumping Robot: Compared to robotic dogs and wheeled robots, which work scenarios are most suitable for jumping robots?
Wang Fangzheng: The two core advantages of jumping robots are all-terrain capability and long endurance. Any place that requires these two abilities could become a potential advantage scene for them in the future.
It is imaginable that there will be many robots in the future. Humanoid robots may be more suitable for homes, where they can do household chores. Wheeled robots may be used in cities to perform basic obstacle avoidance tasks. In outdoor and more complex spaces where no existing robot form can efficiently move, jumping robots have the opportunity to appear in these places.
Jumping Robot: Could jumping itself become a hindrance?
Wang Fangzheng: Since jumping robots have not yet entered a specific scenario, it is difficult for us to fully describe their shortcomings. One limitation that can be thought of is transporting liquids, as jumping can cause liquid agitation.
Human running itself is essentially continuous jumping, also experiencing free fall. Whatever a person can carry while running, a jumping robot can probably carry something similar.
It may appear intense during movement, but it can stop at any time and work in one place. It does not need to hover all the time; it can use landing gear to support itself on the ground without consuming extra power. There will also be plans for legs that can retract in the future.
Jumping Robot: Jumping robots are still in the early stages; what metrics should be used to measure their progress?
Wang Fangzheng: Stability is definitely one. It needs to remain stable in stronger disturbances and more complex environments.
Looking ahead, the jumping robot will have more forms. After we understand jumping itself, it may not necessarily rely on one leg to jump on the ground. The specific form cannot be revealed at the moment, so everyone can look forward to it.
Scalability and intelligence are also inevitable directions. In the future, more functions need to be done with developers to move forward with more people's efforts.
Dynamic Outlook: Will a large model be added to the jumping robot in the future?
Wang Fangzheng: The endgame of the jumping robot will definitely have a brain and eyes, be able to move autonomously, and also be able to self-cognize. We are currently mainly working on the lower body, which is to make it jump stably. What you mentioned, perception, decision-making, and execution tasks, all belong to the upper body.
The jumping robot has a good load-carrying capacity, and in the future, these things can be added to it. We will first launch a more open version and then move towards autonomous movement and self-cognition.
The First Generation Is Not a Toy
A new robot form needs to step out of the laboratory, and first, it needs to solve the problems of mass production, price, and durability. Toya Tech chose to start with a remote-controlled jumping robot. It does not have autonomous navigation, and users control the direction, height, and jump switching through a joystick.
A more practical issue is whether users will continue to use it after the novelty wears off. Wang Fangzheng has focused the product on performance, durability, and price. The team even abandoned the sci-fi shell they had made and switched to a more exposed carbon fiber structure.
Dynamic Outlook: Such cutting-edge technology, but the first step is a remote-controlled product aimed at enthusiasts. Does the research team feel it has become a remote-controlled toy?
Wang Fangzheng: Let me correct one point first—it is not considered a small toy. The first generation is indeed a remote-controlled product, but looking back at Boston Dynamics and DJI, the things they initially created also relied on remote control.
Boston Dynamics' earliest Laikago did not have as many capabilities as today, mainly controlled remotely. At that time, it sold for tens of thousands of yuan, and because of the high price, people thought it was very professional and difficult to enter the mass market. But that does not mean Boston Dynamics does not want more people to see and use it.
Our advantage is that the jumping robot can rely on Shenzhen's mature drone supply chain for production to reduce costs. We hope to meet everyone at a price that users are willing to accept, allowing more people to see and use the jumping robot.
动察 Beating: Can the current pricing be disclosed? Some of our colleagues in the office are already interested in buying.
Wang Fangzheng: Currently, it is not convenient to disclose the pricing. We have conducted a survey, and those who are indeed interested in buying can basically accept this price, so there is no need to worry too much.
动察 Beating: Jumping is a very intense movement. Will the spring, fuselage, and rotor be easily worn, requiring users to frequently replace parts?
Wang Fangzheng: We have considered these issues. During internal testing, we played very aggressively. The latest footage was shot in South Africa, where we brought it to jump on seaside rocks and in the wilderness. The machine could operate normally, and it could jump continuously for a long time.
There was a major shift in the product concept along the way. Initially, we wanted to create a sci-fi-looking product, similar to the concept machine in the promotional video. That machine could already jump and demonstrate. Later, when we played for a long time ourselves, we found that everyone always wanted to perform some intense, exciting maneuvers, and falls and control errors were difficult to avoid completely. The casing was easily damaged in these situations, and the dispersed weight also affected control stability.
We rethought what the first batch of users cared about the most. They were primarily concerned about the jumping itself and cared about being able to continue playing even after a fall. The current version mainly uses carbon fiber, prioritizing performance.
动察 Beating: What is the current level of control accuracy?
Wang Fangzheng: There is an indicator that demonstrates its stability. When no directional control is given, it can basically jump in place within a small range, without jumping further each time.
This generation does not have additional sensors, so it cannot accurately determine how many centimeters to jump or land on a specific point. The operation logic is very simple, similar to a remote-controlled car. Push the joystick forward to move forward, pull back to move backward, following everyone's intuition in using remote-controlled products.
动察 Beating: Many robotics companies often start with B2B customers. Why did you start with the consumer market?
Wang Fangzheng: There is a saying, "If you can do B2C, why do B2B." B2B often requires a lot of resources, and the reasons behind it are more complicated. B2C can directly deliver the product to the market for validation, giving everyone who wants to buy the opportunity to do so.
We carefully considered starting directly with B2B, and many B2B clients approached us. However, our team is very small, and the technology is still in its early stages. If we were to start with inspections, the team would have to spend a lot of effort studying the inspection equipment carried on top. Our advantage lies in the jumping movement itself, not in inspection equipment.
Starting with inspection robots, we will reduce the time spent perfecting jumping technology, making it easy to get stuck in a dead end. By first creating a consumer product and then releasing a more open version to allow more people to develop together, we believe this path is more suitable.
Moment Beating: The first-generation product mainly serves enthusiasts. How do you envision forming a community around it?
Wang Fangzheng: The community can be divided into two parts. The first part is enthusiasts who will share creative gameplay. The second part is developers who will add functionality to the robot, such as follow-up, obstacle avoidance, or integration of large models.
The first-generation product does not yet have secondary development capabilities and mainly serves enthusiasts. A semi-open-source version will follow, mainly targeting developers.
Looking for Nails with a Big Hammer
First acquire the technology, then look for problems — this sequence runs through the company's products and financing. Investors in consumer electronics often ask who it will sell to, while deep-tech investors are more interested in where the technology's capabilities can reach.
Wang Fangzheng summarizes this divergence as "seeing up close" versus "seeing far." This statement carries the perspective of an entrepreneur and also reflects the practical funding needs of an early-stage company. The company aims to continue research, prepare for mass production, and get the product ready for market entry, all of which require funding.
Moment Beating: You have the technology first and then look for the problems it can solve. This is different from the startup sequence of many companies.
Wang Fangzheng: This is exactly what sets us apart from most companies in the market. Many companies first identify a pain point and then look for technology to solve it. We already have a powerful hammer in hand and then think about which nails we can hammer in the future.
In 2024, humanoid robots were just beginning to gain popularity. At that time, many humanoid robots were still unstable in walking, requiring assistance or only able to walk on flat ground. We have already encountered some unpublished results in the lab. The jumping robot can run quickly on campus, leap over obstacles, and adapt to different terrains. I believe this is a significant technological breakthrough in the field of robotics, and its future possibilities may be broader than humanoid robots.
From day one, the company has been discussing where it can be used. There are many possible directions to consider, but it is still too early to directly enter a specific scenario at this stage. One thing we are very clear about is that this technology has the potential to change the world in the future, and we want to drive it forward.
Moment Beating: Why were investors willing to invest in your company in the first round when the scenario was not fully clear yet?
Wang Fangzheng: Willing to invest in our institution, can be considered as willing to invest early, invest small, and invest in deep tech. Many deep tech projects in the very early stages often cannot clearly define a specific use case, and investors first look at the technology itself.
Now, compared to the seed round, the product has made great progress, and our thinking about the path has become clearer. We start with enthusiasts, then build a community, allowing everyone to develop this technology together and explore use cases collectively. We have made our roadmap and key milestones clearer.
Dynamic Beating: Do you consider Tianyue Technology to be a commercial company or a new lab?
Wang Fangzheng: We are a technology company. The company side is responsible for transforming research results into products, making the technology ready for daily life as soon as possible.
The lab is also a very important part. My senior brother and mentor have always maintained collaboration, and they are now teaching at universities. New research results from the lab will be transferred to the company for commercialization, allowing Tianyue Technology to continuously engage with cutting-edge breakthrough technologies.
If we only look at the company side, the investment in research and commercialization is probably evenly split. The lab side focuses solely on research.
Dynamic Beating: You have recently joined the Sweet Potato Robotics ecosystem and also participated in the "Core Gravity DemoDay 2026." What kind of support do you currently need most from them?
Wang Fangzheng: We have just joined the Sweet Potato Robotics ecosystem. By participating in that DemoDay, our main goal was to connect with more investors. The company is currently undergoing the second round of financing, and at this stage, what we need most is funding.
With funding, we can hire more people to accelerate technology development. Research needs to continue, product preparation for mass production is essential, and we need to start entering the market.
Dynamic Beating: How was your experience at DemoDay?
Wang Fangzheng: I am very grateful to them. When we applied, most slots seemed to have been filled, and only the last two spots were left on the poster. Initially, I thought I might just be there to make up the numbers.
Later on, it turned out that our product was indeed impressive, and my presentation went well. In the end, we won third place and took home a prize of ten thousand yuan. I was quite happy about it.
Dynamic Beating: How is the progress of the new round of financing? What are the most common questions investors ask you?
Wang Fangzheng: The current progress is still relatively smooth. Many people's first question is, what can your thing do and what is the use case. Those who ask this question are unlikely to be truly interested by the end of the conversation. They may have long been focused on consumer electronics and have not yet viewed this matter from a hard technology perspective. According to the research data on our first-generation product, it can sell very well and enhance the company's ability to generate profits. However, what we hope for more is that investors can support our future development strategy from the perspective of technological progress.
Everyone is interested in the Jumping Machine, and there is a lot of debate. I've heard about some heated debates within investment firms, similar to the comments section on Douyin. One group is more concerned about how long it will take to become profitable and when they will find product-market fit. There are also people who genuinely love technological development, believe that this technology will definitely be useful in the future, and are willing to promote it early.
Dynamic Insight Beating: How do you view the conflict between these two judgments?
Wang Fangzheng: I think it's about short-term vs. long-term perspectives.
Cyberspace is a good example. Many investors I've spoken to have invested in Cyberspace before or have thought about it but missed the opportunity to invest. Cyberspace had a hard time fundraising before appearing on the Spring Festival Gala. Before and after its sudden rise, it was very difficult for people to clearly define its most certain use case, apart from university labs and the developer market. The most significant change was in how the market viewed it.
Stargaze Technology is probably even earlier in its development than Cyberspace was initially. When Cyberspace started, Boston Dynamics was already there. We are a bit like when Boston Dynamics first released the video of the dancing robot, and everyone had just seen this kind of thing. The road to making the world understand how amazing the Jumping Machine is and what it can do in the future is still long.
Reality Within the Bubble
Robotics is one of the hottest entrepreneurial directions in 2026. Wang Fangzheng acknowledges that there is a bubble in the industry and some have advised him to ride the trend and tell a bigger story. Get an initial batch of B2B intentions, then scale up the financing, and slowly find the direction after the money is in hand. However, this approach did not convince him.
He admires Xiaomi and Cyberspace, with the reasons all coming down to the word "reality." His assessment of Musk also follows the same line, admiring someone's ability to bring technology into existence while also maintaining judgment on how money is made and entrepreneur responsibilities.
Dynamic Insight Beating: Which tech company do you like the most?
Wang Fangzheng: Xiaomi. First of all, I greatly admire Lei Jun. From the early setbacks in entrepreneurship to Kingsoft, Xiaomi, and now Xiaomi's automotive venture, he has always been striving to achieve greater personal value.
Xiaomi is a very down-to-earth company. Once it enters an industry, it often reshuffles the market, making a lot of information and prices more transparent. A tech company should bring tangible benefits to ordinary people, and I think Xiaomi has achieved that.
动察 Beating: Is there a Bubble in the Robotics Industry Now?
Wang Fangzheng: I think there is. Other FAs have approached us, saying that the story we are telling now is wrong. They feel that we should be like some fast-financing robotics companies, securing many B2B cooperation intentions and then telling a big story. Holding fifty billion yuan in hand, they can develop however they want in the future.
For now, we don't want to do that. We adhere to the path we recognize.
A car is a good type of robot. It travels on basically fixed roads every day, and buildings and roads do not suddenly change. With four wheels below, its dynamics are relatively simple. Despite so many cars collecting data every day, fully achieving autonomous driving has not yet been realized.
A humanoid robot's structure is much more complex. To put it into production, enter people's homes, there is still a long way to go. The dynamics of a leaping robot can be fully analyzed, with lower costs and a simpler structure. We have the opportunity to let leaping robots enter practical work scenarios earlier than humanoid robots.
动察 Beating: What do you think of Musk?
Wang Fangzheng: In the capitalist environment of the United States, he understands the entire set of business rules very well. He has technological ideals, knows how to make money, and can also use the cryptocurrency world.
Just looking at his help in the tech field and entrepreneurial achievements, I admire him a lot; he is indeed very impressive. But to evaluate a person as a whole, I hope someone in the future can do better than him.
Living Ten Years Later
On the morning of the interview, Yushu Technology was listed on the Shanghai Stock Exchange's Sci-Tech Innovation Board. In Wang Fangzheng's circle of friends, investors have sent their congratulations, and some have also made money in this investment.
LunarTech is far from that day. The company is still collecting user feedback on the first-generation product, planning KOL beta tests, attending exhibitions in the United States, having just won third place in the Sweet Potato Robot DemoDay, and preparing to meet investors for the second round of financing. Wang Fangzheng talks about DJI and Yushu ten years later, also clear about the most challenging conditions ahead.
动察 Beating: What is the Entrepreneurial Environment in Hong Kong Like? What support have you received from the government and schools?
Wang Fangzheng: Hong Kong, China, provides strong support for tech startups. Our first investment came from InnoHK in Hong Kong, China. The funding behind it is connected to the startup support system in Hong Kong, China.
We were also selected for the HK-SZ Tech Innovation Hub's incubation program. The hub is located at the border of Shenzhen and Hong Kong, China, and will provide companies with space and subsidies. The policy strongly supports tech entrepreneurship.
Educational institutions are also crucial. City University of Hong Kong has the HK Tech 300 program, which has supported many startup teams. For students, taking the first step in entrepreneurship is very challenging, and the school's assistance is key.
BeatX Insights: What are the next steps for BeatX?
Wang Fangzheng: Recently, we have been mainly conducting user research to gather everyone's expectations for the product. We will continue to refine the productized features to make the first generation more user-friendly.
Next, we may engage KOLs for internal testing. The product is expected to be released after participating in CES in early 2027.
BeatX Insights: Ten years from now, what kind of company do you hope LunarTech will be?
Wang Fangzheng: If we were to look for a reference, I would choose Tesla and DJI.
If in the future a very clear scenario emerges in which we have a distinct advantage, similar to aerial photography for drones, we will independently develop relevant technology for that scenario and strive to achieve an 80% market share. That would be a DJI-style path.
Ten years later, if LunarTech still does not have a very large and clear scenario, requiring many people to explore together, we may be more like Tesla.
More importantly, we hope to still be around in ten years.
BeatX Insights: What is your biggest takeaway from Tesla's IPO?
Wang Fangzheng: I saw many investors in my social circle congratulating each other, as they may have also made money. Tesla has taken a very steady path. Among these robotics companies, my favorite is still Tesla, as it has indeed driven the development of quadruped robots.
Many people invested in Tesla, especially the investors in the later rounds, were attracted by market hype and market share. They may not necessarily appreciate how cutting-edge the technology itself is, nor do they necessarily recognize the grounded approach taken. I find this a bit amusing for a tech company.
I have contacted some of these people. If they had the chance to chat with Wang Xing a few years earlier, they most likely wouldn't have invested.
Beat Spotter: What do you think is the market share of such investors?
Wang Fangzheng: I didn't say they are not good. Investors have a responsibility to their LPs, so of course, they will invest in projects within their cognitive range, ones they can understand, and ones where they can more easily see clear returns. I completely understand.
Such investors still have a high market share. In RMB funds, the proportion may be even higher.
In "Legged Robots That Balance," Raibert wrote about how a single leg maintains balance. At that time, no one could see Spot, which would appear over thirty years later, directly from that continuously rising and falling machine. What the research left behind is a set of principles about dynamic balance. It later sprouted two legs, four legs, and also birthed a company.
Wang Fangzheng is taking a more direct path. He already has a machine that can jump, and next, he will turn it into a product that people will buy and continue to use. This process will not be completed with just one beautiful jump.
Every time a machine lands, the spring compresses. To jump again, it first has to catch its own weight.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
