Hong Kong Professors Building Startups in Embodied Intelligence
2026-08-23 19:20:13
Source: QbitAI
Written by Qiao Buchi; from Aofeisi Temple
As the term “embodied intelligence” has spread from laboratories to the capital markets, a particularly interesting group has emerged from Hong Kong’s universities.
They are not merely professors publishing papers at top conferences. They are also founders or chief scientists of companies valued at hundreds of millions or even billions of yuan.
With one foot in the ivory tower and the other at the upstream end of the industrial chain, they are managing both worlds without missing a beat.
According to QbitAI’s research and compilation, at least 16 professors at Hong Kong universities are personally entering the embodied-intelligence startup arena as founders, co-founders, chief scientists, and in other roles.
What makes this group particularly interesting is the breadth of their entrepreneurial focus. They are working on everything from the “thinking” brains and world models of embodied intelligence to 3D vision that enables robots to “see clearly”; from fingertip tactile sensors and dexterous hands to delicate operations on the surgical table, as well as autonomous vehicles on the road and drones in the sky. Their footprints can be found across the entire chain of perception, decision-making, manipulation, and real-world deployment.
Learning about their entrepreneurial journeys, you may see something quite different from the familiar story of a prestigious academic descending into business and effortlessly commanding the market. Instead, you see them leaving the ivory tower and “getting their hands dirty” in the real world.
Throughout this process, how did they set aside the respected identity of professor and “break themselves down and rebuild themselves” to turn laboratory technology into real products?
The journey may be arduous, but perhaps it is just as Professor Yu Wang, who has spent 40 years working on robot manipulation, put it:
This may be the best era ever for scientists to start companies. In the past, a supervisor and a student might have been merely two names on a paper. Today, they have every opportunity to become partners who solve industrial challenges together and change the world side by side.
Hong Kong University Professors Are Building Robots’ “Organs”
Let’s first take a look at what this group is actually working on.
When their companies are viewed together, an interesting pattern emerges: they have almost unanimously avoided building “a complete humanoid robot.” Instead, each is precisely developing a particular “organ” or function of the robot’s body.
Some are building the robot’s brain.
Yisheng Technology’s unified “brain + cerebellum” architecture gives robots highly coordinated eyes, brains, and hands
Yi Ma of the University of Hong Kong is the recipient of the Marr Prize, the highest award for papers in computer vision. He co-founded Yisheng Technology, which focuses on developing world models capable of autonomous learning.
Xiaogang Wang of CUHK is a co-founder and executive director of SenseTime. He founded Daxiao Robotics, which is currently providing robots with an “embodied brain,” with the long-term goal of developing general-purpose embodied robots.
Kui Jia of CUHK-Shenzhen is developing an embodied-intelligence engine—the “most powerful brain” for robots. The company’s valuation has already surpassed 10 billion yuan and it is currently preparing for an IPO.
In addition, YuanCe Future, which has just secured financing, is developing a “general-purpose whole-body humanoid foundation model.” The company was founded by Hongyang Li of the University of Hong Kong in April 2026.
This is the layer furthest upstream technologically and closest to artificial general intelligence, corresponding neatly to the group’s academic background in computer vision.
Some are building the robot’s “hands” and “nerves.”
Daimon Robotics, co-founded by Yu Wang of HKUST, develops the robot’s “nerve endings.” Each square centimeter of its robotic hands is densely packed with 40,000 sensing units, giving embodied robots a sense of touch.
Even without visual input, robots can rely on the tactile sensations at their fingertips to “feel” what is happening at the point of contact in real time and adjust their movements accordingly—roughly equivalent to being able to insert a key into a lock and open it with your eyes closed.
Ping Luo of the University of Hong Kong is a co-founder of Ultra Power. The company develops not only dexterous hands and embodied foundation models, but also an ultra-high-degree-of-freedom humanoid robot with 117 degrees of freedom.
A full-stack closed loop is critical to bringing embodied intelligence into the real world: data enters the model, the model drives the robot, and the results of the robot’s actions feed back into the training system. If any link is missing, the loop breaks.
Image generated by AI
This layer is widely recognized in the industry as a major technical bottleneck—and it also happens to be a traditional strength of Hong Kong universities in sensors and precision control.
Weijia Wen of HKUST is the chief scientist of Mogan Technology. He is equipping robots with a “nervous system” resembling a layer of human skin, giving them human-like perceptual capabilities.
Some are developing the robot’s “fuel.”
There is a saying in the embodied-intelligence community: “The model is only the engine; data is the oil.”
Peng Yin of CityU has worked at NASA and NVIDIA. He founded Saiyuan, which specializes in providing training data for robotics companies.
Others are building complete machines for specific scenarios.
Guowei Ou of CUHK founded Cornerstone Robotics, which develops laparoscopic surgical robots. The company has already received approval from the National Medical Products Administration and has raised more than 3 billion yuan in total financing. Yunhui Liu of CUHK founded ForwardX Robotics, which develops autonomous forklifts for logistics and warehousing. Fu Zhang of the University of Hong Kong founded Silu Technology, which develops drones capable of autonomous flight through tunnels and narrow alleys.
Another group has extended the definition of “physical AI” into autonomous driving. Zexiang Li’s CIDI X develops commercial vehicles; Ming Liu of HKUST founded Yee Fung Innovation, which develops autonomous vehicles for industrial parks; and Shaojie Shen of HKUST—the author of the VINS navigation algorithm—founded卓驭科技, which develops intelligent-driving systems for heavy trucks.
From data to brains, from dexterous hands to complete machines, and then to autonomous driving, this group of Hong Kong professors has assembled nearly the entire technological backbone of embodied intelligence.
Get your hands dirty
Professors founding companies is nothing new.
Professor Zexiang Li of HKUST is widely regarded as a leading representative of entrepreneurship among Hong Kong university professors and as the embodiment of the “professor + student” startup model.
His story goes back to his connection with DJI.
In 1992, Li resigned from his faculty position in the Computer Science Department at New York University and joined the Hong Kong University of Science and Technology, which had been established only a year earlier.
Before long, he recognized a painful problem in traditional engineering education: graduates were often focused single-mindedly on heading to Silicon Valley or Wall Street and had little interest in industrial innovation.
Image from the HKUST (Shenzhen Research Institute) website
In Li’s view, university research must have an impact on industrial development in order to fulfill the purpose of a university.
He therefore began searching widely for “new engineering” talent with both innovative and hands-on abilities—“people who can use technology to create new things.”
That was when Wang Tao came into his view. Wang was Li’s student. During their time together, Wang’s passion and persistence in researching technology left a deep impression on Li.
Li made an exception to admit Wang, whose graduation project had received only a C, as his graduate student.
It was a case of truly recognizing talent. Under Li’s guidance, Wang founded DJI in Shenzhen in 2006 while still a graduate student.
In its early days, DJI’s cash flow once broke down, and nearly all of its technical staff left. With nowhere else to turn, Wang went back to Li. Li not only invested his own money but also introduced DJI to the first graduate student he had trained at the Harbin Institute of Technology Shenzhen Graduate School.
DJI eventually made a dramatic comeback and became a drone giant with annual revenue approaching 80 billion yuan.
DJI’s success marked a turning point.
Before that, students’ first choice after graduation was usually to pursue further study in the United States or join Silicon Valley or Wall Street. After DJI’s rise, 90% of the students went on to start companies. By then, the development opportunities in the Greater Bay Area had far surpassed those overseas.
Professors at CUHK were busy too. In 2014, Xiaou Tang, a professor in CUHK’s Department of Information Engineering, founded SenseTime with several academics and graduate students. Within just a few years, SenseTime became the world’s largest AI company by both total funding raised and valuation.
The entrepreneurial stories of these two Hong Kong professors proved something to the entire academic community: university professors can indeed build world-class companies through hard-tech entrepreneurship.
Fu Zhang, a tenured associate professor in the Department of Mechanical Engineering at the University of Hong Kong, also has a strong connection with Li and DJI.
Around 2015, Zhang joined HKUST at Li’s invitation and began systematically researching robots and drones.
Looking back on that experience from a decade ago, he still remembers it vividly: “Professor Li’s influence on me continues to this day.”
While working in the HKUST laboratory, Zhang was taught by Li that once you had identified a goal, you had to pursue it personally. You should not remain merely a scholar or bystander; you had to personally take on the dirtiest and hardest work.
“Get your hands dirty,” was one of Li’s favorite expressions. It meant: “Don’t be choosy or picky. Don’t spend every day sitting in the lab directing other people. Get down to the front line.”
Zhang took his mentor’s advice to heart. Once a laboratory project entered the manufacturing stage, some mold factories were unwilling to produce the small quantities the lab needed. Zhang personally traveled to Dongguan to negotiate prices and bargain with mold manufacturers.
During drone tests in the mountains, he would also go ahead of the team, climbing for an entire afternoon until his legs were trembling when he returned. “Looking back, it was indeed extremely hard, but it forces you to settle down and communicate with frontline teams, users, and the supply chain.”
Beginning in 2016, Zhang served as a senior consulting scientist at DJI for eight years, becoming deeply involved in the company’s core technology development and product deployment. The experience benefited him greatly.
Previously, in academia, the process ended once research findings had been published as papers.
At DJI, he participated in the entire process of transforming technology into products and taking those products to market. He communicated with R&D personnel, visited factories to observe workers’ operations, and even operated the equipment himself several times. Only then did he understand how many difficulties and challenges a product must face between development and commercialization.
With these two experiences behind him, Zhang joined the Department of Mechanical Engineering at the University of Hong Kong in August 2018 as an assistant professor—he is now a tenured associate professor—and established the MaRS Lab, the HKU Mechatronics and Robotic Systems Laboratory.
“When I established the lab, the goal I set was to solve real problems in industry, not to publish papers.”
If you solve real problems in the industry, recognition will come naturally, and academia will acknowledge you. But if you conduct research solely to publish papers from the beginning, nine times out of ten the research will end when the paper is published. It will simply become another item on your annual evaluation checklist at the university.
In February 2026, Zhang commercialized technology developed at the MaRS Lab by founding Silu Technology. The company aims to build a brain for drones and turn them into aerial intelligent agents that can truly perceive, think, and act.
Zhang said the startup experience also showed him “two completely different worlds.”
“Universities are like ivory towers. Professors are protected too well; everyone flatters and respects you. After starting a company, as your identity changes, you inevitably have to endure a great deal of frustration. For example, the people you want to hire may not want to join, the subsidies you apply for may not come through, or people at the company may not understand the decisions you make.”
Through this process, Zhang realized that he had to set aside his status and face the far more complex environment of society.
He had to learn to handle all kinds of complicated relationships: between people, between individuals and society, between people and government, and between people and the law.
Silu Technology’s aerial intelligent agent operating in an indoor warehouse
In his view, being a professor can often become a burden. Entrepreneurship means you must be willing to break yourself apart and put yourself back together.
Li has always been Zhang’s role model. “Professor Li is arguably one of the wealthiest professors in the robotics industry. He is also highly successful academically, with no shortage of honors or money. He could easily have chosen a comfortable path. But he still chose to do difficult work, incubate a group of outstanding companies, and influence a group of people, including me. That in itself is remarkable.”
After DJI, Li went on to support his HKUST doctoral student Jinbo Shi in founding Li Qun Automation, his master’s student Shizheng Tao in founding Avid Technology, and his doctoral student Di Zhang in founding Benmo Technology.
But Li’s real strength lies not simply in helping his students. It lies in transforming “a professor helping students start companies” from an isolated practice into a replicable, systematic “investment + incubation” model.
In 2014, Li decided to open this model to the outside world. Rather than focusing only on students from HKUST laboratories, he began working with promising entrepreneurs from universities across China.
He invited Bingqiang Gao, former dean of the School of Engineering at HKUST, and Jie Gan, associate dean of Cheung Kong Graduate School of Business. Together, the three founded the International Robotics Industry Base—XbotPark Robotics Base—on a site in Songshan Lake, Dongguan.
In 2015, they established a supporting incubator fund, the Hong Kong Clear Water Bay Entrepreneurship Fund, later renamed the XBOTPARK Fund. It focuses on robotics and intelligent hardware, primarily investing at the seed, angel, and Series A stages.
How popular is the base? A saying has become common in the venture capital community: “If you want to invest in robotics, go to Songshan Lake.”
The base provides young entrepreneurs with abundant resources and has incubated innovative companies including Whale Robotics, EcoFlow, Hai Robotics, and WOAN Robotics. Li has also become a true “godfather of entrepreneurship.”
Two Role Models Are Connecting the Dots in the “Mentor + Student” Startup Model
Companies founded by professors often adopt a “mentor + student” model.
This model has a long history. In 1986 and 1988, Peking University professor Wang Xuan and Northeastern University professor Liu Jiren founded the predecessors of Founder Group and Neusoft Group, respectively. Their teams included students.
Some of those students later became the core personnel driving the companies’ continued growth. These teams were pioneers and practitioners of the professor-student entrepreneurship model.
Li’s incubation of student-founded companies is also “highly representative.”
A professor’s or laboratory’s research direction has been developed over a long period, achieved breakthroughs in the field, and those breakthroughs align closely with industrial needs. A team then specializes in technology transfer, scaling up the research results. This is probably the most common model people see.
Li Zhongyu, co-director of the Hong Kong Embodied Intelligence Laboratory, explained: “This shows that both students and professors have confidence in their technology and can make the business model work. It also reflects their confidence in the team. The team has greater cohesion and is highly familiar with the technology, making it well suited to starting a company together.”
In Hong Kong, there are many others who have taken the “mentor + student” path.
ForwardX Robotics, co-founded by Yunhui Liu of CUHK, has a pure mentor-student lineup: CEO Luyang Li and CTO Mu Fang are both doctoral students who came out of his laboratory.
Yu Wang of HKUST co-founded Daimon Robotics with his doctoral student Jianghua Duan. The company’s core R&D team came from Wang’s HKUST laboratory.
Wang clarified to QbitAI that Daimon is not a traditional story of a mentor leading a student into entrepreneurship.
Instead, Duan pulled him into the venture. “Jianghua Duan is Daimon’s founder and CEO, and the company’s overall commander. I chose to support him wholeheartedly.”
Yu Wang, co-founder of Daimon Robotics
Wang believes that the most practical advantage of the mentor-student startup model is trust.
“Jianghua and I had a foundation of many years of academic collaboration. We clearly understood each other’s standards and bottom lines, so we could work back-to-back from day one, without spending a great deal of time on coordination. Second, the division of labor was clear: he was on the front line making decisions, while I provided scientific oversight and assessed the frontier. In addition, the academic networks behind the mentor-student relationship can continuously provide talent for the company and help earn the trust of peers and partners more quickly.”
At Hong Kong Polytechnic University, Pei Zheng brought the RAIDS research team he led into the technology-transfer process, commercializing the laboratory’s research in human-robot collaboration and embodied intelligence by founding CobotAI. A number of laboratory members joined the company.
Weijia Wen of HKUST brought a doctoral team from HKUST (Guangzhou) into entrepreneurship, incubating Mogan Technology directly from a project that won first prize in a bionic robotics competition.
Chaoxu Yan, who holds a PhD in microelectronics, serves as CEO, while Hang Zhou, a doctoral student in robotics, serves as CTO. Wen is the company’s chief scientist and also holds equity.
Li once made a statement that precisely captures how this group of entrepreneurial professors views their roles:
“Professors may not necessarily have the answers. A professor is someone who guides and accompanies entrepreneurs, helping them connect the dots.”
This positioning is also reflected in other “mentor + student” startups.
Most mentors follow the principle of “empowerment without interference”: they do not serve as CEOs or intervene in day-to-day management. Instead, they provide technical support and credibility, serving as chief scientists or founders while giving students the freedom to lead from the front as CEOs or CTOs.
Hongyang Li, founder and chief robotics officer of YuanCe Future, strongly identifies with this approach.
Li, an assistant professor in the Department of Artificial Intelligence and Data Science at the School of Computing and Data Science at the University of Hong Kong, co-founded YuanCe Future with Li Chen, the first author of the UniAD paper that he led, and Tianyu Li, a PhD from Fudan University.
Chen is now the chief AI officer, while Tianyu Li serves as CEO. Both are long-time collaborators who have worked with Hongyang Li for four or five years.
He emphasized to QbitAI that, put plainly, their startup is “a teacher leading his own troops,” but he is not the leader. The three are equal partners with clearly defined roles, and their responsibilities are not assigned according to academic seniority. They have very different personalities, but their collaboration and rapport are strong.
YuanCe Future primarily develops a general-purpose whole-body humanoid foundation model. The goal is to enable robots to do more than simply perform tasks with their upper bodies; they should also possess whole-body coordination and generalization capabilities. For example, when retrieving clothing from beneath a cabinet, a robot needs to squat while reaching out to pick up the item. This human-like ability to move and manipulate with the entire body is precisely what YuanCe Future is developing.
In Li Zhongyu’s view, professor entrepreneurship is a widespread phenomenon worldwide, whether in Hong Kong, mainland China, or the United States. When he was pursuing his PhD at Berkeley, he saw many professors conducting research and starting companies at the same time.
When a professor’s research direction closely overlaps with industrial needs, cases emerge in which technology is transferred from university research into commercial applications. It is simply that embodied intelligence and AI have become extremely popular in recent years, with substantial capital participation, so people have only now begun to notice that many professors in Hong Kong are starting companies.
Li Zhongyu elaborated further: the large number of professors launching companies in Hong Kong is not simply the result of one or two successful role models.
In his view, it is inseparable from the ecosystem and culture of Hong Kong universities, which place strong emphasis on research investment and industrial commercialization.
First, entrepreneurship is concentrated in applied disciplines where technology and industry are closely integrated.
In applied engineering fields, universities encourage people to think about how technology can be applied and commercialized, and how it can generate greater social and economic value.
When professors write research proposals, they are routinely asked: “Can what you are studying be applied? Can it be commercialized?”
Entrepreneurship provides a way to apply technology in real-world settings and complete the loop in a business environment. This process is a touchstone for testing the technology itself and a crucial part of research in applied disciplines.
Second, the development of embodied intelligence will create a wide range of application scenarios.
Li Zhongyu believes that embodied intelligence will eventually empower industries across the economy, using cutting-edge technology to solve practical problems with social and economic value. This makes it an excellent direction for research.
He offered a very realistic hypothetical question: if embodied intelligence attracted no attention and had no practical use in the real world, who would start a company? And who would support you?
Long Startup Cycles, Yet Capital Keeps Chasing Them
Home-use scenario dataset collected by Daxiao Robotics
The robotics startup cycle is long, and many technical paths have yet to fully mature.
The journey from R&D to real-world deployment and eventually revenue generation is a long one, and investment logic is changing accordingly.
Yet startups founded by professors at Hong Kong universities seem to be consistently picked out by investors: most companies raise hundreds of millions of yuan in their angel or seed rounds.
For example, Yisheng Technology secured hundreds of millions of yuan in angel financing, YuanCe Future raised hundreds of millions of yuan in seed financing, and Daimon Robotics has secured multiple rounds of financing worth hundreds of millions of yuan.
People often say that Hong Kong university professors do not have to worry about money when they start companies. But after looking at this list of financing figures, it becomes clear that they are not merely well funded—capital is actively chasing them.
First, consider the speed. For this generation of professors entering entrepreneurship, financing is measured in days.
Mogan Technology, founded by Weijia Wen, was only a month old when Sequoia China, Hillhouse, and Xpeng Robotics lined up to invest;
Silu Technology, founded by Fu Zhang, completed four financing rounds within six months, with investors competing for allocations; Daxiao Robotics, founded by Xiaogang Wang, progressed from its angel round to an “angel+” round within four months.
By the standards of general hard-tech sectors, this pace is extraordinarily intense.
Then there is the size of the rounds—not small change either. Cross-Dimension Intelligence raised 1 billion yuan in a single Series B round, pushing its post-money valuation above 10 billion yuan.
The surgical-robotics company Cornerstone Robotics has raised 3.2 billion yuan in total. Daxiao Robotics raised hundreds of millions of US dollars in the first half of the year. Even Daimon Robotics’ angel round reached hundreds of millions of yuan, setting a new record for angel financing in the tactile-sensing sector.
Yisheng Technology and YuanCe Future are even more striking: both started with angel rounds worth hundreds of millions of yuan.
What has attracted the most attention is the lineup of “guests” participating in each financing round.
Leading venture capital firms such as Sequoia, ZhenFund, and IDG are merely standard participants. Industrial giants such as BYD, FAW, and Lenovo are the regulars.卓驭科技 is backed by FAW, which made a single strategic investment of 3.6 billion yuan, as well as BYD and SAIC; capital has directly become a customer. Mogan Technology’s angel round also included Xpeng Robotics, an embodied-robot manufacturer, creating a “brain + body” partnership.