# 李泽湘 · Li Zexiang (Zexiang Li) — 封面传记 ACF-00-00119

> 亚洲封面人物 Asia Cover Figure · 机器可读档案（LLM-ready）。本文件由官网结构化档案数据自动生成，供 AI 检索与引用。中文全文与英文全文对照编排。

## 档案元数据 Metadata

- 封面编码 ACF Code：**ACF-00-00119**
- 姓名 Name：李泽湘 / Li Zexiang (Zexiang Li)
- 职务 Title：教授、科技投资人 / Professor, Tech Investor
- 公司 Company：香港科技大学 / HKUST / XbotPark / Shenzhen InnoX Academy
- 出生 Born：1961，湖南省永州市蓝山县 (Lanshan County, Yongzhou, Hunan Province, China)
- 篇别 Category：格局（格局篇 / Cover Biography (Geju)）
- 入档日期 Accessioned：2026-05-26
- 标签 Tags：硬科技, 创业孵化, 机器人, 运动控制, 大疆, XbotPark, 新工科教育, 产学研, 工程教育, 粤港澳大湾区
- 永久档案链接 Archive URL：https://coverfigure.com/acf/ACF-00-00119/geju
- English archive：https://coverfigure.com/acf/ACF-00-00119/geju?lang=en
- 官网原文报道 Feature story：https://coverfigure.com/acf/figure/lizexiang

## 信息口径声明 Editorial Sourcing Policy

- 工商登记信息可通过国家企业信用信息公示系统核验；经营数据来自企业公开披露、官方发布与权威媒体报道，引用处标注来源。
- 收购、IPO、估值、融资等事项以官方公告为准；人物观点与言论为公开演讲、采访的事实记录，不代表编辑部立场。
- 本档案仅供行业研究参考，不构成投资建议。事实更正联系：editorial@coverfigure.com。

## 卷首 Editorial Intro

2003年，香港科技大学的一堂机器人课上，一个学生的毕业设计演示失败，成绩只拿了一个C。老师没有放弃这个C——后来，那个学生叫汪滔，他做出了大疆。

这位老师叫李泽湘。1979年，18岁的他作为中国改革开放后首批公派本科生赴美，在卡内基梅隆大学拿双学位，在伯克利拿下数学硕士和电机工程博士，在MIT人工智能实验室做过研究，1992年加入创校第二年的香港科技大学。他创立了机器人在非完整约束下运动规划这一学术领域，2007年当选IEEE会士，论文被引过万次。

他本可以只做一个安静的学者。卡内基梅隆的同窗多数留在了硅谷，伯克利的导师们沿着经典学术轨迹一路攀升。但他选择了另一条路：1999年，他与高秉强、吴宏创办固高科技，一家运动控制公司。一个大学教授放下论文去做工厂，这在当时几乎不可理解。他后来解释这个抉择：中国的工业控制系统长期依赖进口，如果没有自主的底层技术，再好的学术论文也改变不了制造业受制于人的局面。固高走了二十四年，2023年在深交所上市。

此后他加速将教室延伸到产业一线：2014年联合创办松山湖XbotPark机器人基地，把创业孵化做成从课程、实验室到供应链、基金的全链条工程；2021年又创办深圳科创学院，探索新工科教育。据2026年8月中国经济网报道，他的体系累计孵化硬科技企业超过280家，总估值超5000亿元。

他说过，大学里最重要的成果应该是学生，而不是论文。一个C的故事，就是这句话的注脚；而一个学者转身走向产业的选择，则是一整套方法论的起点。亚洲《封面人物》独立编委会以终身档案编码的形式，将这份从学者到创业铺路人的转型路径纳入永久存证。

In 2003, in a robotics course at the Hong Kong University of Science and Technology, a student's graduation-project demonstration failed and he received a grade of C. His professor did not give up on that C—and that student, Frank Wang, went on to build DJI.

The professor was Li Zexiang. In 1979, at 18, he was among the first Chinese undergraduates sent to study abroad after the reform and opening era. He earned dual degrees at Carnegie Mellon, a master's in mathematics and a PhD in electrical engineering at Berkeley, and researched at MIT's AI Lab before joining HKUST in 1992, the university's second year. He founded the academic field of motion planning under nonholonomic constraints, was elected an IEEE Fellow in 2007, and his papers have been cited more than 11,000 times.

But he cared more about what happened outside the classroom. In 1999 he co-founded Googol Technology, a motion-control firm that took twenty-four years to list on the Shenzhen Stock Exchange in 2023; in 2014 he co-founded the Songshan Lake XbotPark robotics base, turning incubation into a full-chain engineering process running from courses and labs to supply chains and funds; in 2021 he founded the Shenzhen InnoX Academy to explore a new model of engineering education. By August 2026, China Economic Net reported, his ecosystem had incubated more than 280 hard-tech companies with a combined valuation exceeding 500 billion yuan.

He has said the most important output of a university should be students, not papers. The story of a single C is the footnote to that sentence.

## 人物速览 Lead

李泽湘，1961年生于湖南蓝山，机器人与自动化领域学者，香港科技大学电子与计算机工程学系教授、林高演教授席教授[s1]。他是中国首批公派赴美本科生之一，在卡内基梅隆大学获电机工程与经济学双学士学位，在加州大学伯克利分校获数学硕士及电机工程与计算机博士学位，后在麻省理工学院人工智能实验室和纽约大学工作[s2][s3]。1992年他加入创校翌年的香港科技大学，创办3126实验室[s1][s6]；1999年与高秉强、吴宏创办固高科技[s2]；2006年起作为导师支持学生汪滔创办大疆创新并曾任大疆董事长[s4][s8]；2014年联合高秉强、甘洁创办松山湖XbotPark机器人基地，模式后复制至宁波、重庆、长沙等地[s2][s8]；2021年创办深圳科创学院，探索新工科教育[s10][s19]。据2026年8月中国经济网报道，其体系累计孵化超280家硬科技企业，总估值超5000亿元，其中4家已上市、12家为独角兽或准独角兽[s12]。2019年他与汪滔共获IEEE机器人与自动化大奖，为该奖首批中国获奖者[s14][s15]。

Zexiang Li, born in 1961 in Lanshan, Hunan, is a robotics and automation scholar and professor in the Department of Electronic and Computer Engineering at the Hong Kong University of Science and Technology (HKUST), where he holds the Lam Ko Yin Professorship [s1]. Among China's first government-sponsored undergraduates in the U.S., he earned dual B.S. degrees in electrical engineering and economics from Carnegie Mellon University, an M.A. in mathematics and a Ph.D. in EECS from UC Berkeley, followed by work at the MIT AI Laboratory and NYU [s2][s3]. In 1992 he joined HKUST, founded the year after the university's establishment, and set up Lab 3126 [s1][s6]. In 1999 he co-founded Googol Technology [s2]; from 2006 he mentored his student Frank Wang in founding DJI and served as its chairman [s4][s8]; in 2014 he co-founded the Songshan Lake XbotPark robotics base, later replicated in Ningbo, Chongqing, Changsha and beyond [s2][s8]; and in 2021 he founded the Shenzhen InnoX Academy for new-engineering education [s10][s19]. According to an August 2026 China Economic Net report, his ecosystem had incubated over 280 hard-tech companies with a combined valuation above 500 billion yuan, including 4 listed firms and 12 unicorns or quasi-unicorns [s12]. In 2019 he and Frank Wang jointly received the IEEE Robotics and Automation Award, its first Chinese recipients [s14][s15].

## 正文 Archive Chapters

### 1. 南岭少年与留美十年：从蓝山到伯克利 / A Boy from the Nanling Mountains and a Decade in America: From Lanshan to Berkeley

李泽湘1961年出生于湖南省南部边界的蓝山县，在教师家庭长大，母亲在小学任教，父亲是中学物理老师。蓝山地处南岭山脉腹地，他后来在深圳科创学院开学第一课以《从南岭到岭南的几个科创小故事》为题，回忆的正是这段从湘南山村走向世界的起点。[s2][s7][s10]

1978年，恢复高考后的第一年，李泽湘考入中南矿冶学院，即后来中南大学的前身。次年一个偶然的机会改变了他的路径：美国铝业公司（ALCOA）代表团访华，为答谢中方接待，提出愿意为中国培养两名本科生。18岁的大一新生李泽湘被选中，作为中国改革开放后首批公派赴美本科生之一，进入匹兹堡的卡内基梅隆大学（CMU）就读。据后来的报道回忆，临行前他几乎没有英语基础，飞往美国的航班上一位老华侨见他们没有手表，硬把自己的手表塞到他们手中；抵达美国后，两个中国年轻人用大约六个月时间过了语言关，开始跟班上课。[s7][s2]

在卡内基梅隆，李泽湘同时修读电机工程与经济学，1983年作为优秀毕业生获得双学士学位，并获得学校大学学者奖（University Scholar Award）。经济学训练后来被他视为理解市场与产业的重要底色。毕业后他转入加州大学伯克利分校，1986年获数学硕士学位，1989年获电机工程与计算机博士学位，其间于1989年获得伯克利E.I. Jury奖。机器人学在当时仍是新兴学科，李泽湘属于最早一批在美国系统接受机器人与自动化训练的中国内地学生。[s1][s3][s2]

博士毕业后，李泽湘没有立刻回到中国。他先后在麻省理工学院人工智能实验室担任研究科学家，在纽约大学库朗数学研究所体系下的机器人与制造实验室担任助理教授。在MIT期间，两位机器人学者给他留下深刻印象：创立波士顿动力（Boston Dynamics）的马克·拉博特（Marc Raibert）和创立iRobot的罗德尼·布鲁克斯（Rodney Brooks）。两人的实验室产品充满天才创意，却长期无法实现量产、难以进入市场。这种"技术惊艳但商业失语"的落差，让李泽湘形成了一个延续此后三十多年的判断：科技成果必须完成向产品和产业的转化，研究的价值要在市场中检验。[s4][s8]

1990年，他获得美国国家科学基金会研究创新奖（Research Initiation Award）。1992年，在香港科技大学创校第二年，李泽湘做出一个当时令不少人意外的决定：放弃美国的教职，加入这所全新的大学。他后来解释，选择香港是因为这里毗邻内地，能够做产学研结合的事；而港科大的创办为他提供了一个平台。同年，他在港科大创办数控研究实验室与自动化技术中心，办公室与实验室编号3126，这个实验室后来被称为3126实验室。[s6][s3][s5]

**English:** Li Zexiang was born in 1961 in Lanshan County, on the southern border of Hunan Province, and grew up in a teaching family: his mother taught at a primary school and his father was a middle-school physics teacher. Lanshan lies deep in the Nanling mountain range; years later, he titled his opening lecture at the Shenzhen InnoX Academy "A Few Innovation Stories from the Nanling to the Lingnan," recalling the starting point of a journey that took him from a mountain village in southern Hunan onto the world stage. [s2][s7][s10]

In 1978, the first year of the restored national college entrance examination, Li was admitted to the Central South Institute of Mining and Metallurgy, the predecessor of Central South University. A chance encounter the following year changed his path: when an ALCOA delegation visited China, the company offered to sponsor two Chinese undergraduates as a gesture of thanks. The 18-year-old freshman was selected and entered Carnegie Mellon University in Pittsburgh as one of the first mainland Chinese undergraduates sent to study in the United States after reform and opening. Reports recall that he had almost no English before departure; on the flight, an elderly overseas Chinese, noticing the students had no watches, pressed his own watch into their hands. After arriving, the two young Chinese spent about six months clearing the language barrier before following regular classes. [s7][s2]

At Carnegie Mellon, Li studied both electrical engineering and economics, graduating with distinction in 1983 with dual bachelor's degrees and receiving the University Scholar Award. He later regarded his economics training as essential background for understanding markets and industry. He moved on to UC Berkeley, earning an M.A. in mathematics in 1986 and a Ph.D. in electrical engineering and computer sciences in 1989, the same year he received Berkeley's E.I. Jury Award. Robotics was still an emerging discipline, and Li belonged to the first cohort of mainland Chinese students trained systematically in robotics and automation in the United States. [s1][s3][s2]

After his doctorate, Li did not return to China immediately. He served as a research scientist at the MIT Artificial Intelligence Laboratory and as an assistant professor in the Robotics and Manufacturing Laboratory associated with NYU's Courant Institute. At MIT, two robotics figures left a deep impression: Marc Raibert, founder of Boston Dynamics, and Rodney Brooks, founder of iRobot. Their labs produced dazzling inventions that long resisted mass production and market entry. That gap between technical brilliance and commercial silence shaped a conviction Li held for the next three decades: research must be converted into products and industry, and the value of technology is tested in the market. [s4][s8]

In 1990, he received the U.S. National Science Foundation's Research Initiation Award. In 1992, the second year after the Hong Kong University of Science and Technology was founded, Li made a decision that surprised many: he gave up a U.S. faculty position to join the brand-new university. He later explained that he chose Hong Kong for its proximity to mainland China, where industry-academia collaboration was possible, and that the founding of HKUST gave him a platform. The same year, he established the CNC Research Laboratory and the Automation Technology Center, housed in room 3126 — the lab later known simply as Lab 3126. [s6][s3][s5]

### 2. 机器人学的数学根基：非完整约束与统一建模 / The Mathematical Foundations of Robotics: Nonholonomic Constraints and Unified Modeling

在学术身份上，李泽湘的核心领域是机器人、运动控制、机构学和制造科学。他发表论文200多篇、专著4本，其中一本被称为机器人研究生课程的第一本教材。北京大学工学院在其客座教授简介中写道，他创立了机器人在非完整约束下运动规划这一重要学术领域，其工作被国际同行普遍采用或引用。[s3][s2]

这一领域的起点可以追溯到他在伯克利和纽约大学时期的研究。非完整约束（nonholonomic constraint）描述的是一类无法通过积分转化为几何约束的速度约束，典型例子包括滚动的轮子、漂浮的航天器和多指机械手的接触——系统能到达的构型空间远大于瞬时速度所能张成的方向，使得运动规划成为非平凡的数学问题。李泽湘发表了该领域的第一篇论文，组织了第一次国际学术研讨会，并出版了该领域的第一篇专著。[s2]

传播最广的成果是1994年由CRC Press出版的《A Mathematical Introduction to Robotic Manipulation》（《机器人操作的数学导论》），与加州理工学院的Richard M. Murray和加州大学伯克利分校的S. Shankar Sastry合著。这本书抛弃了传统机器人教材依赖Denavit-Hartenberg参数和局部坐标系拼凑的建模范式，改用微分几何、李群李代数和旋量理论作为统一语言：刚体运动被表述为特殊欧氏群SE(3)上的群作用，关节运动、末端位姿、速度、力与力矩都可以用旋量（twist）和余旋量（wrench）做坐标无关的内蕴描述，动力学方程被写成李群上的测地线运动。书中从刚体运动、开链机器人运动学、动力学与控制，一直写到多指手协调操作和非完整运动规划，1990至1993年间其讲义先后在伯克利、纽约大学库兰特研究所、加州理工和香港科技大学讲授。中译本1998年由机械工业出版社出版，成为国内机器人方向研究生长期使用的参考书。[s3]

李泽湘的学术贡献获得了体系化的认可。1997年他获得中国国家自然科学三等奖；2000年获国家自然科学基金委员会杰出青年科学基金；2007年当选IEEE会士（IEEE Fellow），这是国际电气与电子工程师学会的最高等级会员。据松山湖管委会2018年发布的资料，当时他的学术著作被引用次数已超过1.1万次，他参与建立的工业机械臂数学建模理论是工业机器人界最常用的标准模型之一。他还长期服务于学术共同体：2001至2004年担任IEEE Transactions on Robotics and Automation副编辑，2011年担任IEEE国际机器人与自动化大会（ICRA）总主席，并多次担任ICRA、IROS、CASE等会议的程序委员、分会主席和奖项委员。[s3][s15][s2]

在李泽湘自己的叙述中，数学工具始终与工程实践互为表里。他在港科大的教学工作包含四条线：成立自动化技术中心做科研、承担校企合作课题、组织机器人比赛和毕业设计、以及寻找机器人学中统一的数学工具。他认为机器人学研究里最重要的是数学模型，数十年来一直在寻找统一的数学工具和模型；这些工具最终体现在学院派创业公司的产品里，构成它们与珠三角传统代工企业的本质区别。[s4][s6]

**English:** As a scholar, Li Zexiang's core fields are robotics, motion control, mechanism science and manufacturing science. He has published more than 200 papers and four monographs, one of which is described as the first textbook for graduate robotics courses. Peking University's College of Engineering notes in his visiting-professor profile that he founded the important academic field of motion planning for robots under nonholonomic constraints, and that his work is widely adopted and cited internationally. [s3][s2]

The field traces back to his research years at Berkeley and NYU. Nonholonomic constraints describe velocity constraints that cannot be integrated into geometric constraints — classic examples include rolling wheels, floating spacecraft and multi-fingered hand contacts: the configuration space a system can reach is far larger than the directions its instantaneous velocities span, making motion planning a non-trivial mathematical problem. Li published the field's first paper, organized its first international workshop and produced its first monograph. [s2]

His most widely circulated work is A Mathematical Introduction to Robotic Manipulation, published by CRC Press in 1994 and co-authored with Richard M. Murray of Caltech and S. Shankar Sastry of UC Berkeley. The book set aside the traditional Denavit-Hartenberg parameter-and-local-frame patchwork, building instead on differential geometry, Lie groups and screw theory as a unified language: rigid-body motion is expressed as group action on the special Euclidean group SE(3), and joint motion, end-effector pose, velocity, force and torque are described intrinsically — independent of coordinate frames — through twists and wrenches, with dynamics written as geodesic motion on a Lie group. From rigid-body motion and open-chain kinematics through dynamics and control to coordinated multi-fingered manipulation and nonholonomic motion planning, the lecture notes were taught at Berkeley, NYU's Courant Institute, Caltech and HKUST between 1990 and 1993. The Chinese translation was published by China Machine Press in 1998 and became a long-standing reference for Chinese robotics graduate students. [s3]

His scholarly contributions received systematic recognition. In 1997 he received China's National Natural Science Award (Third Class); in 2000, the National Science Fund for Distinguished Young Scholars; and in 2007 he was elevated to IEEE Fellow, the society's highest membership grade. According to a 2018 Songshan Lake administration release, his works had by then been cited more than 11,000 times, and the mathematical modeling theory for industrial robotic arms he helped establish was among the standard models most commonly used in industrial robotics. He served the academic community for years as an associate editor of IEEE Transactions on Robotics and Automation (2001-2004), as General Chair of ICRA 2011, and repeatedly on program committees, as session chair and on award committees for ICRA, IROS and CASE. [s3][s15][s2]

In Li's own account, mathematical tools and engineering practice are two sides of the same coin. His teaching at HKUST ran along four lines: running the Automation Technology Center for research, undertaking industry-university projects, organizing robotics competitions and capstone projects, and searching for a unified mathematical instrument for robotics. He held that the mathematical model was the heart of robotics research; the tools he pursued for decades eventually appeared in the products of academically-rooted startups, constituting their essential difference from the Pearl River Delta's traditional contract manufacturers. [s4][s6]

### 3. 固高十五年：运动控制器与广东农民运动控制讲习所 / Fifteen Years at Googol: Motion Controllers and the Guangdong Farmers' Motion-Control Institute

李泽湘的第一次完整创业，是固高科技。1999年，深圳市政府、北京大学与香港科技大学三方共建深港产学研基地，推动两校科研成果在深圳产业化；项目启动仪式需要有公司签约入驻，有人找到李泽湘，固高科技便在这样的背景下"仓促问世"。他与港科大同事高秉强、吴宏共同创办公司，英文名GOOGOL意为10的100次方，这个名字的创意来自李泽湘的妻子，一位数学家。固高是亚太区第一家专注运动控制及控制系统的高科技公司，比美国那家著名的同名跨国科技公司（Google）成立还要早一个月。[s6][s1]

固高做的是运动控制器，也就是工业装备的"大脑"。但公司成立初期市场远未准备好，产品一度卖不出去。李泽湘和团队转而到各地办培训班，为珠三角制造业老板培训运动控制技术人才。他自嘲这些培训班是"广东农民运动控制讲习所"——当时珠三角制造业的很多老板原来都是种田的农民，企业模式基本是三来一补、来料加工，从国外拿订单、买设备，招廉价劳动力组织生产。李泽湘的判断是，中国企业不能总是买设备，未来必然要发展自己的设备产业，而设备制造最核心的技术就是机器控制系统。[s6][s4]

这是一场漫长的垦荒。从早期的PCB钻床、滴灌设备，到后来的半导体设备、LED设备，再到机器人和数控机床，固高团队见证了深圳乃至珠三角装备产业的整条演进路径。他们把珠三角产业链从终端产品制造向上延伸到设备制造、系统集成和运动控制环节，为区域产业升级提供基础平台。到2017年前后，固高已成为中国最有影响力的运动控制公司，占据国内市场约50%的份额，产品销往全球。[s6][s4]

固高的价值不止于一家公司。李泽湘后来总结，通过创办固高，他对产业需要什么样的人才、学生需要具备什么能力有了深刻认识："大学研究必须对产业的发展产生影响，才能显现出大学的意义。大学里最重要的成果应该是学生，而不是论文。"固高也是他自己第一次完整走过从技术到产品、从实验室到市场的闭环，期间碰壁的经验直接促成了他后来的战略转向——从自己创业，转向系统性地支持学生创业。他也公开讲过失败的案例：据《中国企业家》报道，其早期学生中有8人合伙创办了做LED焊线机的比锐公司，一度发展势头良好，却因创始人内讧而失败。[s12]

2023年8月，固高科技股份有限公司在深交所创业板上市，证券代码301510。从1999年成立到2023年上市，这家公司走过了二十四年，被媒体称为李泽湘"人生第一个IPO"。上市之后，固高成为李泽湘孵化体系中基础工业软件与控制环节国产替代路径的资本市场样本。[s2][s13]

**English:** Li Zexiang's first complete startup was Googol Technology. In 1999, the Shenzhen government, Peking University and HKUST jointly built the Shenzhen-Hong Kong Industry-Education-Research Base to commercialize the two universities' research in Shenzhen. The launch ceremony needed companies to sign on as tenants; Li was approached, and Googol was, in his words, hastily born. He co-founded it with HKUST colleagues Gao Bingqiang and Wu Hong. The English name GOOGOL means 10 to the power of 100 — an idea from Li's wife, a mathematician. Googol was Asia-Pacific's first high-tech company focused on motion control systems, incorporated a month earlier than the famous American multinational sharing a similar name. [s6][s1]

Googol made motion controllers — the brains of industrial equipment. But in the early years the market was far from ready and products barely sold. Li and his team instead ran training courses across the region, teaching motion-control skills to Pearl River Delta factory owners. He jokingly called them the "Guangdong Farmers' Motion-Control Institute": many delta manufacturers had been farmers, running compensation-trade and processing businesses that took foreign orders, bought foreign equipment and hired cheap labor. Li's judgment was that Chinese companies could not forever buy equipment; China would eventually need its own equipment industry, and the core technology of equipment making was the machine control system. [s6][s4]

It was a long reclamation. From early PCB drilling machines and drip-irrigation equipment to semiconductor and LED equipment, then robots and CNC machine tools, the Googol team witnessed the entire evolution of Shenzhen's and the delta's equipment industry. They extended the regional supply chain upward from end-product manufacturing into equipment making, system integration and motion control, providing a foundation platform for industrial upgrading. By around 2017, Googol had become China's most influential motion-control company, holding about half the domestic market and selling worldwide. [s6][s4]

Googol's value went beyond a single company. Li later said that founding Googol gave him a deep understanding of what talent industry needed and what capabilities students required: "University research must influence industrial development for the university's meaning to be realized. The most important output of a university should be students, not papers." Googol was also his first complete loop from technology to product and from lab to market; the bumps along the way directly drove his strategic shift — from starting companies himself to systematically supporting students in starting them. He spoke publicly about failures too: according to China Entrepreneur magazine, eight of his early students co-founded Birui, a maker of LED wire-bonding machines, which briefly showed strong momentum but collapsed amid infighting among the founders. [s12]

In August 2023, Googol Technology listed on the ChiNext board of the Shenzhen Stock Exchange under stock code 301510. Twenty-four years after its 1999 founding, the company was described by the media as Li's "first IPO." After listing, Googol became the capital-market sample in his ecosystem for the domestic-substitution path in basic industrial software and control. [s2][s13]

### 4. 大疆：一门课、一个C和一段全程护航 / DJI: One Course, One C, and End-to-End Escort

李泽湘被外界提及最多的身份，是大疆创新的早期引路人。这段关系始于一门课。2003年前后，香港有电视台到高校寻找老师带队参加机器人比赛，多数教师回避，李泽湘接了下来。他在港科大开设的机器人比赛与设计课程，要求不同专业、不同年级的学生混编组队，用八个月时间从概念设计做到能上场比赛的机器人原型机，期间必须反复迭代、到深圳采购零部件，在错误中学习。电子系本科生汪滔修了这门课两次，2004至2005年带领港科大战队夺得Robocon香港冠军和亚太区季军。[s4][s8][s6]

汪滔的本科毕业设计选题是遥控直升机的飞行控制系统，但演示没有成功，成绩只拿到C，刚刚及格，因此失去了赴欧美名校深造的机会。李泽湘看中这个年轻人对航模的痴迷和动手能力，破格招收他为硕士研究生，研究方向就是飞控，并提供全额奖学金。在港科大读研期间，汪滔一边读书一边在深圳注册公司，向航模发烧友售卖飞控部件。[s4][s6]

2006年，26岁的汪滔与两位同学在深圳车公庙一间不足20平方米的仓库里创办大疆创新。早期公司困难重重：汪滔招来的工程技术人员陆续离开，有人甚至带走技术另起炉灶。李泽湘的支持是多层的：资金上，他与哈尔滨工业大学的朱晓蕊教授在大疆早期累计提供约100万元；人才上，他从自己创办的哈工大深圳研究生院控制与机电工程学科部抽调学生支援大疆，其中一些人毕业后直接加入；方向上，他曾建议汪滔不要只盯着直升机，也去做"肯定能卖出去"的运动控制卡、驱动器一类产品——汪滔后来在访谈中回忆了这段劝诫，但最终坚持只做飞行器这一件事。[s8][s6]

大疆从飞控部件走向航拍一体机的关键转折，李泽湘也深度参与。公司发现大量客户买了飞控后还要另购GoPro运动相机自行组装航拍，性能不稳定而专业航拍设备学习门槛极高。李泽湘、汪滔等人曾赴硅谷与GoPro洽谈合作，由大疆研发、GoPro销售，对方提出30%利润分成——这在当时对珠三角代工企业已是难得条件，但他们在楼下车库讨论后拒绝了，转而用三个月时间自研出大疆首款航拍一体机精灵（Phantom）。产品问世后国内投资机构反应冷淡，李泽湘拎着两台样机飞往硅谷，敲开红杉资本主席迈克尔·莫里茨（Michael Moritz）的门，为大疆引入顶级资本。莫里茨后来把精灵2代比作"飞行着的Apple II"，精灵系列入选美国《时代》周刊2014年度十大科技产品。大疆管理体系跟不上扩张速度时，李泽湘还带队前往苹果公司取经，帮助完善管理。他曾长期担任大疆董事长，与汪滔一起到野外测试无人机。[s4][s8]

大疆后来成长为全球民用无人机与航空影像技术的最大生产商，占据全球消费级无人机市场超过七成份额，港科大官方简介中的表述是"逾七成市场"，《湾区科创》杂志2026年专访给出的数字是超过75%。2019年，李泽湘与汪滔共同获得IEEE机器人与自动化大奖，成为该奖2002年设立以来首批中国获奖者；IEEE的颁奖词评价他们率先打破了"有用的飞行机器人必然昂贵笨重"的刻板印象，把无人机从奢侈品玩具重新定位为实用工具。[s1][s19][s14]

对李泽湘而言，大疆的意义是模式上的：这是第一个由学生主导、老师在后方辅助的完整案例。他在多个场合说"大疆和汪滔只是一个开始"——真正重要的是把支持汪滔的方法沉淀为可复制的体系。他也坦言自己在大疆只保留少量股份，公司具体运作一直由汪滔团队管理，"有问题找我，我会给建议"。[s6][s8]

**English:** The identity most often attached to Li Zexiang is that of DJI's early guide. The relationship began with a course. Around 2003, a Hong Kong TV station looked for a faculty advisor to lead a robotics competition team; most professors steered clear, but Li accepted. His robotics competition and design course at HKUST mixed students across majors and years into teams that spent eight months moving from concept to a competition-ready prototype, iterating repeatedly, sourcing parts in Shenzhen and learning from mistakes. Electronics undergraduate Frank Wang (Wang Tao) took the course twice and led the HKUST team to the Robocon Hong Kong championship and third place in the Asia-Pacific region in 2004-2005. [s4][s8][s6]

Wang's undergraduate thesis was a flight control system for a remote-controlled helicopter, but the demonstration failed and he received only a C — a passing grade that closed the door to graduate study at top Western universities. Li, struck by the young man's obsession with model aircraft and his hands-on ability, admitted him as a master's student against convention, with flight control as his research direction and a full scholarship. While studying at HKUST, Wang registered a company in Shenzhen, selling flight-control components to hobbyists. [s4][s6]

In 2006, the 26-year-old Wang and two classmates founded DJI in a warehouse of less than 20 square meters in Chegongmiao, Shenzhen. The early years were brutal: engineers Wang hired left one after another, and some took the technology to start rival operations. Li's support ran on multiple levels. Financially, he and Professor Zhu Xiaorui of Harbin Institute of Technology provided about 1 million yuan cumulatively in the early stage. On talent, he dispatched students from the control and mechatronics division he had founded at HIT's Shenzhen graduate school, some of whom joined DJI after graduation. On direction, he once advised Wang not to fixate on helicopters but to also build products "certain to sell," such as motion-control cards and drivers — advice Wang later recalled in interviews, though he ultimately insisted on aircraft alone. [s8][s6]

Li was deeply involved in DJI's pivotal turn from flight-control components to all-in-one aerial cameras. The company noticed that many customers bought flight controllers and then separately purchased GoPro cameras to assemble unstable aerial rigs, while professional aerial gear demanded months of training. Li, Wang and colleagues traveled to Silicon Valley to propose a partnership with GoPro — DJI to develop, GoPro to sell — with GoPro offering a 30 percent profit share, already generous for delta contractors. After debating in a garage downstairs, they declined and spent three months developing DJI's first integrated aerial drone, the Phantom. When domestic investors showed little interest, Li carried two prototypes to Silicon Valley and knocked on the door of Sequoia Capital chairman Michael Moritz, bringing in top-tier capital. Moritz later compared the Phantom 2 to "a flying Apple II," and the Phantom was named among Time magazine's top ten tech products of 2014. When DJI's management could not keep pace with its growth, Li led a team to Apple to study its management practices. He served as DJI's chairman for years and tested drones in the field alongside Wang. [s4][s8]

DJI grew into the world's largest manufacturer of civilian drones and aerial imaging technology, holding more than 70 percent of the global consumer drone market — "over 70 percent" in HKUST's official profile, and "over 75 percent" in a 2026 Bay Area Tech Innovation magazine interview. In 2019, Li and Wang jointly received the IEEE Robotics and Automation Award, the first Chinese recipients since the award's 2002 founding; IEEE's citation credited them with dispelling the stereotype that useful aerial robots must be expensive and unwieldy, repositioning drones from luxury toys to practical tools. [s1][s19][s14]

For Li, DJI's significance was methodological: it was the first complete case in which a student led and the teacher supported from behind. He said repeatedly that "DJI and Wang Tao are only a beginning" — what mattered was distilling the method of supporting Wang into a replicable system. He also noted that he retained only a small stake in DJI and that day-to-day operations always belonged to Wang's team: "When there's a problem they come to me, and I give advice." [s6][s8]

### 5. XbotPark：把孵化做成一门可复制的工程 / XbotPark: Making Incubation a Replicable Engineering Discipline

2014年11月，在东莞松山湖高新区支持下，李泽湘联合香港科技大学工学院原院长高秉强、长江商学院副院长甘洁，共同发起松山湖国际机器人产业基地，即后来的XbotPark机器人基地，由李泽湘团队负责投资、建设和运营。基地的早期资金，部分来自几位发起人出售手中部分大疆股份所得。按照李泽湘的设想，基地要把"基础扎实、受过良好训练、具有国际视野的年轻人+珠三角产业资源+导师的行业经验与国际化资源"组合在一个平台上，帮助更多年轻人完成从技术到公司的跨越。[s2][s12][s6]

XbotPark与传统孵化器的区别，是李泽湘反复强调的。他在2026年《湾区科创》杂志专访中指出，国内绝大多数孵化器的核心模式是"空间租赁加资金扶持"，而XbotPark做的是从教育端开始的全链条赋能：学生先通过科创训练营接触真实的创业过程和前人踩过的坑，再组团队、找方向，在创业导师陪伴和供应链支持下逐步把产品做出来。基地内部，九米多高的车间里排布着设备与3D打印等供应商服务窗口，一个想法从图纸变成实物所需的环节被压缩到步行可达的距离内。生态内部还形成互助循环：固高的控制器可以支持李群自动化，李群的设备又支持大疆的电机生产。[s19][s12][s15]

资金与基金体系同步搭建。2015年，团队成立香港清水湾创业基金，后更名为XBOTPARK基金，为项目提供从探索期、种子期到天使期的资金支持。北京大学工学院的公开资料显示，李泽湘体系自建基金已投资2.8亿元支持60多家在孵初创企业，并撬动外界资本约17亿元跟投。[s8][s3]

这套体系产出的企业群像，覆盖了硬科技的多个方向。工业端有2011年成立的李群自动化（轻量型工业机器人，李泽湘早期以个人名义提供启动资金）、2012年成立的逸动科技（船艇电动化）、2016年成立的海柔创新（"货箱到人"仓储机器人，李泽湘任首席顾问，估值超20亿美元）、本末科技（机器人动力模组）、松灵机器人（移动机器人底盘）、奇诺动力（外骨骼机器人）；消费端有2016年前后进入体系的云鲸智能（张峻彬创办，扫拖机器人，据报道2019年发布产品后次年销售额达6亿元）、2017年成立的正浩创新EcoFlow（王雷创办，储能电源）；还有睿魔智能、未知星球（据报道2023年推出liberlive无弦吉他后销售增长迅速）等。据公开报道，企业成活率长期保持在约80%，且几乎全部为硬科技公司；早期资料称约15%成长为独角兽或准独角兽（不同时间点统计口径不一）。[s8][s13][s11][s3]

数字的口径随时间更新。2021年前后公开报道多使用"孵化60多家科技企业"；2024年李泽湘回湖南创办的希迪智驾冲刺IPO时，报道口径为体系参与培育250家以上企业；2026年2月广东省高质量发展大会上，李泽湘给出的数字是累计孵化超270家硬科技企业、总估值超3500亿元，其中4家上市、12家为独角兽或准独角兽；到2026年8月中国经济网的长篇报道中，数字更新为累计孵化超280家、总估值超5000亿元。口径差异主要来自统计范围（仅松山湖基地，还是包含宁波、重庆、长沙、常州、广州、北京等全部落地平台）和时间点的不同。[s3][s8][s11][s12]

资本市场的兑现集中发生在2023至2025年。2023年8月固高科技在深交所创业板上市；2025年12月19日，希迪智驾（3881.HK）登陆港交所主板，成为港股首家商用车智能驾驶上市公司，发行价263港元，募资净额约13.09亿港元，李泽湘为控股股东（IPO前持股43.64%）；11天后的12月30日，2015年成立、2017年进入XbotPark体系的卧安机器人（06600.HK）在港交所挂牌，李泽湘以非执行董事身份持股超10%，出席敲钟仪式。三家公司分属工业控制、商用车自动驾驶和家庭具身机器人赛道，被媒体视为同一套孵化路径在不同方向上的重复验证。[s2][s16][s13]

**English:** In November 2014, with support from the Songshan Lake High-Tech Zone in Dongguan, Li co-founded the Songshan Lake International Robotics Industry Base — later XbotPark — with Gao Bingqiang, former dean of HKUST's School of Engineering, and Gan Jie, deputy dean of the Cheung Kong Graduate School of Business. Li's team handled investment, construction and operations. Early funding came partly from the founders selling part of their DJI shares. Li's design combined well-trained young people with international vision, the Pearl River Delta's industrial resources, and mentors' industry experience and global networks on one platform, helping more young people cross the gap from technology to company. [s2][s12][s6]

Li repeatedly stressed how XbotPark differed from conventional incubators. In a 2026 Bay Area Tech Innovation interview, he noted that most domestic incubators ran on "space rental plus funding," whereas XbotPark delivered end-to-end empowerment starting from education: students first encounter real entrepreneurship and the pitfalls of predecessors through bootcamps, then form teams, find directions, and build products with mentors and supply-chain support. Inside the base, nine-meter-high workshops house equipment and supplier service windows such as 3D printing, compressing everything needed to turn a drawing into a physical object within walking distance. An internal mutual-aid loop took shape: Googol's controllers support Liqun Automation, whose equipment in turn supports DJI's motor production. [s19][s12][s15]

The funding system was built in parallel. In 2015, the team set up the Hong Kong Clearwater Bay Fund, later renamed the XBOTPARK Fund, providing capital from exploration and seed stages through angel rounds. Peking University's public profile records that Li's self-managed funds invested 280 million yuan in more than 60 incubated startups, leveraging roughly 1.7 billion yuan in follow-on outside capital. [s8][s3]

The resulting portfolio spans hard-tech directions. On the industrial side: Liqun Automation (2011, lightweight industrial robots, with Li personally providing early seed money), ePropulsion (2012, electric boat propulsion), Hai Robotics (2016, box-to-person warehouse robots, with Li as chief advisor and a valuation above US$2 billion), AgileX Robotics, LimX Dynamics (robotics power modules),松灵机器人 (mobile robot chassis) and Kenurity (exoskeletons). On the consumer side: Narwal (founded by Zhang Junbin, entering the ecosystem around 2016; its robot mop-vacuum was reported to reach 600 million yuan in sales the year after its 2019 product launch), EcoFlow (founded by Wang Lei in 2017, portable power stations), plus Remo-robot and Unknown Planet (whose liberlive stringless guitar, launched in 2023, was reported to see rapid sales growth). According to public reports, survival rates have held around 80 percent, almost entirely in hard tech; earlier profiles put about 15 percent becoming unicorns or quasi-unicorns (figures vary by measurement date). [s8][s13][s11][s3]

The headline figures have been updated over time. Reports around 2021 cited "more than 60 incubated companies"; when CiDi headed for IPO in 2024, coverage put the ecosystem's cumulative portfolio at over 250 companies; at the February 2026 Guangdong High-quality Development Conference, Li reported more than 270 hard-tech companies incubated with a combined valuation above 350 billion yuan, including 4 listed firms and 12 unicorns or quasi-unicorns; by an August 2026 China Economic Net feature, the figures stood at over 280 companies and a combined valuation above 500 billion yuan. The differences mainly reflect scope (Songshan Lake alone versus all platforms in Ningbo, Chongqing, Changsha, Changzhou, Guangzhou, Beijing and elsewhere) and the date of measurement. [s3][s8][s11][s12]

Capital-market realization clustered in 2023-2025. Googol listed on Shenzhen's ChiNext in August 2023. On December 19, 2025, CiDi (3881.HK) listed on the HKEX main board as Hong Kong's first commercial-vehicle intelligent driving company, priced at HK$263 with net proceeds of about HK$1.309 billion; Li was the controlling shareholder with 43.64 percent pre-IPO. Eleven days later, on December 30, Woan Technology (06600.HK) — founded in 2015 and in the XbotPark system from 2017 — listed on HKEX, with Li holding more than 10 percent as a non-executive director and attending the bell-ringing. The three companies sit in industrial control, commercial-vehicle autonomous driving and home embodied robotics respectively, and the media read them as repeated validation of one incubation path across different directions. [s2][s16][s13]

### 6. 新工科教育：从一门机器人课到深圳科创学院 / New-Engineering Education: From a Robotics Course to Shenzhen InnoX Academy

李泽湘的孵化实践始终建立在一个教育判断之上：传统工科教育培养不出能够驾驭从0到1创新的创业者。他在混沌学园2017年的课程《科技创业的孵化密码》中直言："以前我招学生，都是找最好的学校、成绩最好的，最后发现错了，工程意识比成绩更重要。"他所说的工程意识有三个支柱：设计思维，即通过观察发现真问题；动手能力，即在实践中快速迭代；数学工具，即用理论简化复杂问题。[s18]

教学改革的起点是机器人比赛课程。2004年，李泽湘把ROBOCON机器人竞赛引入香港科技大学课堂，课程要求学生在八个月里完成从概念到原型机的全过程，不同专业混编组队，为了成绩必须不断迭代设计，并到深圳采购零部件，从而熟悉珠三角产业链。汪滔是这门课最著名的产物，但不是唯一一个。同年，李泽湘在哈尔滨工业大学深圳研究生院创办控制与机电工程学科部，从2004年开始每年招生约50人，七年培养350多名学生，鼓励他们创业、动手、到现场发现问题。此后他又先后与广东工业大学、湖南大学、常州大学、重庆大学等高校合作创办机器人学院或教改班。[s8][s2][s6]

3126实验室成为这套方法的早期样本。实验室的名字在后来的报道中被解释为"3个老师带12个学生，6天做出原型机"。三十多年间，从这里毕业的100多位学生中超过三分之一走上创业道路，创办科技公司50多家，包括大疆创新的汪滔、李群自动化的石金博、逸动科技的陶师正。李泽湘在实验室里很少指定学生"你做什么"，而是问"你想做什么"。[s12][s5][s10]

2021年，教育实验升级为独立机构。在深圳市政府支持下，深圳科创学院正式成立，李泽湘担任创始院长，学院被列入深圳市"十四五"规划重点支持项目，与郑泉水院士领衔的深圳零一学院同属深圳全新机制创新创业学院体系。2021年9月15日，首批50名新生在坪山区入学。学院设置智能驾驶、智能建造、智慧生活、智慧健康和柔性制造五大研究中心，每年举办四期科创训练营，每期3至6周，从中选拔学生并提供一年全额奖学金；学生入学即围绕真实市场需求做项目，在"做中学"。李泽湘在开学第一课中说："最后能够成就一番事业的，不是成绩最好的，也不是一毕业就考了研的，而是喜欢折腾、喜欢鼓捣、不安于现状，在传统教育模式下不是特别受待见的那一拨人。"[s10][s19]

到2026年，学院已举办数十期科创训练营与创业营，吸引全球300余所高校数万名学生参与，与香港科技大学、香港大学、斯坦福大学、沙特国王科技大学等20余所高校建立深度合作，形成覆盖本硕博阶段的培养网络。学院在孵近千名创业者与约百个创业项目，人才覆盖剑桥、帝国理工、清华、北大等学府及大疆、华为等企业，项目80%布局AI硬件方向、80%主攻海外市场。2026年，学院与哈尔滨工业大学共建具身智能（大湾区班），面向本科招生，中国工程院外籍院士黄捷任班主任，李泽湘任首席顾问。[s19][s20]

李泽湘还用"创业浓度"来检验教改效果：他在2026年广东省高质量发展大会上给出对比数据，麻省理工、斯坦福学生的创业浓度不足1%，而常州大学教改班超过10%，重庆大学明月班达到20%，广州美术学院教改班达到26%。他将这套跨城市落地的框架总结为"1地+1（N）校+1平台+1园区"：一个孵化基地、一所或多所合作高校、一个基金平台、一个产业园，已在全国8个城市落地。2026年4月，他受聘为东莞国家卓越工程师创新研究院名誉院长，把新工科方法接入国家卓越工程师培养体系；该研究院已与中国科学技术大学、华南理工大学等十余所985高校及科研院所签署联合培养协议，累计招收培养工程硕博士1900余名。[s11][s19]

**English:** Li Zexiang's incubation practice rests on an educational judgment: traditional engineering education cannot produce entrepreneurs capable of driving zero-to-one innovation. In his 2017 Hundun course "The Incubation Code of Tech Entrepreneurship," he put it bluntly: "I used to recruit students from the best schools with the best grades, and eventually found I was wrong — engineering awareness matters more than grades." The engineering awareness he described has three pillars: design thinking, discovering real problems through observation; hands-on capability, iterating quickly in practice; and mathematical tools, using theory to simplify complexity. [s18]

The teaching reform began with the robotics competition course. In 2004, Li brought ROBOCON into HKUST's curriculum: mixed-major teams spent eight months moving from concept to prototype, iterating designs for results and sourcing parts in Shenzhen, learning the delta supply chain firsthand. Frank Wang was the course's most famous product, but not the only one. The same year, Li founded the Control and Mechatronic Engineering division at HIT's Shenzhen Graduate School, enrolling about 50 students a year from 2004 and training more than 350 over seven years, encouraging them to start ventures, build things and find problems in the field. He later co-founded robotics schools or reform classes with Guangdong University of Technology, Hunan University, Changzhou University, Chongqing University and others. [s8][s2][s6]

Lab 3126 became the early sample of this method. Its name was later glossed in reports as "three teachers, twelve students, six days to a prototype." Over three decades, more than a third of the 100-plus students who graduated from the lab became entrepreneurs, founding more than 50 technology companies — including Frank Wang of DJI, Shi Jinbo of Liqun Automation and Tao Shizheng of ePropulsion. In the lab, Li rarely told students what to do; he asked what they wanted to do. [s12][s5][s10]

In 2021, the educational experiment became an independent institution. With Shenzhen government support, the Shenzhen InnoX Academy was formally established with Li as founding dean. Listed as a key supported project in Shenzhen's 14th Five-Year Plan, it sits alongside the Shenzhen X-Institute led by academician Zheng Quanshui in the city's new-mechanism innovation academy system. On September 15, 2021, the first 50 students enrolled in Pingshan. The academy set up five research centers — intelligent driving, intelligent construction, smart living, smart health and flexible manufacturing — running four bootcamps a year of three to six weeks each, selecting students for a one-year full-scholarship program in which they work on real market needs from day one, learning by doing. In his opening lecture, Li said: "Those who eventually accomplish something are not the top graders or those who go straight to graduate school, but the ones who love tinkering, who are restless with the status quo — the group not especially favored under the traditional education model." [s10][s19]

By 2026, the academy had run dozens of bootcamps and entrepreneurship programs, drawing tens of thousands of students from more than 300 universities worldwide, and built deep partnerships with over 20 universities including HKUST, HKU, Stanford and KAUST, forming a training network spanning bachelor's, master's and doctoral levels. It was incubating nearly 1,000 entrepreneurs and about 100 projects, with talent from Cambridge, Imperial College, Tsinghua and Peking universities and from companies such as DJI and Huawei; 80 percent of projects were in AI hardware and 80 percent targeted overseas markets. In 2026, the academy co-founded the Embodied Intelligence (Greater Bay Area) Class with Harbin Institute of Technology for undergraduates, with CAE foreign academician Huang Jie as class head teacher and Li as chief advisor. [s19][s20]

Li also uses "entrepreneurship concentration" to measure reform outcomes. At the 2026 Guangdong conference, he reported that the entrepreneurship concentration at MIT and Stanford was under 1 percent, versus over 10 percent in the Changzhou University reform class, 20 percent in Chongqing University's Mingyue class and 26 percent in the Guangzhou Academy of Fine Arts reform class. He summarized the cross-city framework as "1 base + 1 (N) schools + 1 platform + 1 park": an incubation base, one or more partner universities, a fund platform and an industrial park, now operating in eight cities. In April 2026, he was named honorary dean of the Dongguan National Academy of Outstanding Engineers, plugging the new-engineering method into the national outstanding-engineer training system; the academy had signed joint-training agreements with more than ten 985 universities and research institutes including USTC and South China University of Technology, enrolling over 1,900 engineering master's and doctoral students. [s11][s19]

### 7. 方法论与判断：大湾区供应链、工程师红利与商业闭环 / Method and Judgments: Bay Area Supply Chains, the Engineer Dividend, and the Commercial Loop

在三十多年的产学研实践中，李泽湘形成了一套反复陈述的判断。

关于区域优势，他最常引用的是粤港澳大湾区的供应链效率。他在2025粤港澳大湾区人工智能与机器人产业大会和2026年广东省高质量发展大会上都给出同一组对比：大湾区是世界上唯一同时汇聚机电一体化技术和人工智能技术的地区，即便在硅谷能找齐供应链，大湾区的产品迭代速度也是硅谷的10至30倍，而成本仅为十分之一。他形容大湾区的科研与供应链融合优势"非常了不起，怎么说都不为过"，并称这里是"the Hollywood of makers"（创客的好莱坞）。他同时指出硅谷、波士顿、以色列工程师的短板：迭代过程太慢，没有像中国科技公司那样在市场中快速学习。[s13][s11][s4]

关于创业方法，他总结过成功科技创业的若干共性：领头人要有混合式交叉背景，既懂机械、电子，也懂软件和设计，同时是产品经理、市场经理、人事经理和运营经理；好的创意来自设计思维、批判思维、市场调研和头脑风暴；团队要随产品试错持续迭代；要做与众不同的产品而非山寨；必须具备不易被取代的核心竞争力。他教创业者去见代工厂老板时"别说技术多先进，就说能帮工厂省多少成本、赚多少钱"。[s18][s8]

关于商业与资本的关系，他的表述在硬科技投资语境中相当克制："创业不是讲故事拿融资的游戏。只有在商业闭环的过程中，你的技术、产品才能够持续迭代。光是一个秀，这是给资本市场看的，但资本市场最终还是要回报，哪怕你上市了，你得给股民创造价值，更重要的是给用户创造价值。"他的孵化体系因此愿意给慢项目时间：厨鲸团队想尽快推产品上市时，他建议"思考全面一点，把后面的路想清楚"，团队最终花三四年打磨产品；他也极少给学生标准答案，而是反问问题，让团队自己把没想清楚的环节补齐。[s12]

关于人才，他长期持一个反成绩至上的立场："创业能力和考分高低没关系，关键是要有好奇心和执行力。"他招人的偏好是"喜欢玩"、喜欢折腾的学生，认为判断创新人才最好的标准是看他能否完成从0到1、再从1到100的创业过程。他本人热爱骑行、徒步等户外运动，孵化基地里流传着"李教授的硕士生必须能一天徒步五十公里，博士生要一百公里"的说法。[s8][s7]

关于宏观趋势，他在2017年的课程中就提出中国正经历从"人口红利"向"工程师红利"的转变，并在后来多次重申"过去40年我们靠人口红利，未来40年靠工程师红利"。2026年初，他进一步提出"新制造"概念：中国消费至少还有4.6万亿美元的增量空间，应鼓励年轻人在C端创业——C端创业者普遍在30岁以下、最懂新一代消费者——以国际知名消费品牌反向拉动B端产业链升级。他把年轻人C端创业总结为"一举六得"：创业者更具优势、反向推动新工科教育、引领新制造体系、强化以内需为主导的经济结构、打造中国全球化创新品牌、投资回报更好。[s18][s5][s11]

关于身份，他在2026年《中国企业家》访谈中被问及如果只能保留一个身份会选什么，回答是："我们就做人才的培养、孵化这一件事，帮助企业一步一步发展起来。具体是什么身份，都不重要。"[s5]

**English:** Across more than three decades of industry-academia practice, Li Zexiang has developed a set of repeatedly stated judgments.

On regional advantage, he most often cites the supply-chain efficiency of the Guangdong-Hong Kong-Macao Greater Bay Area. At both the 2025 Greater Bay Area AI and Robotics Industry Conference and the 2026 Guangdong High-quality Development Conference, he offered the same comparison: the Bay Area is the only place in the world where mechatronics and artificial intelligence converge; even if one could assemble a full supply chain in Silicon Valley, the Bay Area iterates 10 to 30 times faster at one-tenth the cost. He called the region's fusion of research and supply chain "remarkable, impossible to overstate" and described it as "the Hollywood of makers." At the same time, he noted the weakness of engineers in Silicon Valley, Boston and Israel: their iteration was too slow, lacking the rapid in-market learning of Chinese technology companies. [s13][s11][s4]

On entrepreneurial method, he has summarized common traits of successful tech startups: leaders need hybrid cross-disciplinary backgrounds — understanding mechanics, electronics, software and design, while serving as product, marketing, HR and operations managers; good ideas come from design thinking, critical thinking, market research and brainstorming; teams must iterate through product trial and error; build distinctive products rather than copies; and hold core competitiveness that is hard to replace. He teaches founders meeting contract manufacturers: "Don't talk about how advanced your technology is — talk about how much cost you save the factory and how much money it can make." [s18][s8]

On the relationship between business and capital, his language is notably restrained in the hard-tech investment context: "Entrepreneurship is not a game of telling stories to raise funding. Only within a commercial loop can your technology and products keep iterating. A mere show is for the capital markets, but capital markets ultimately want returns; even if you list, you must create value for shareholders — and more importantly, for users." His incubation system accordingly gives slow projects time: when the Chuqing team wanted to rush a product to market, he advised them to "think comprehensively and figure out the road ahead," and the team spent three to four years polishing the product. He rarely gives students standard answers; he asks questions in return, letting teams fill in the gaps themselves. [s12]

On talent, he has long held an anti-grade-centric position: "Entrepreneurial ability has nothing to do with exam scores; what matters is curiosity and execution." He prefers students who love to play and tinker, holding that the best test of innovative talent is whether someone can complete the zero-to-one and then one-to-hundred process of entrepreneurship. He himself loves cycling and hiking, and a saying circulates in the incubation base: "Professor Li's master's students must be able to hike fifty kilometers a day; doctoral students, one hundred." [s8][s7]

On macro trends, he argued as early as his 2017 course that China was moving from a "population dividend" to an "engineer dividend," later restating that "the past 40 years we relied on the population dividend; the next 40 years rely on the engineer dividend." In early 2026, he went further with a "new manufacturing" concept: China's consumption still had at least US$4.6 trillion of incremental headroom, and young people should be encouraged to start consumer-facing companies — C-end founders are generally under 30 and understand the new generation of consumers best — with global consumer brands pulling B-end supply-chain upgrades in reverse. He summed up youth C-end entrepreneurship as "one move, six gains": stronger founders, a reverse push for new-engineering education, leadership of a new manufacturing system, an economy led by domestic demand, Chinese global innovation brands, and better investment returns. [s18][s5][s11]

On identity, when asked in a 2026 China Entrepreneur interview which single title he would keep, he answered: "We just do this one thing — cultivating and incubating talent, helping companies develop step by step. What the exact title is doesn't matter." [s5]

## 卷尾 Editorial Conclusion

他常说，大学里最重要的成果应该是学生，而不是论文。他没有创办出最大的公司，但他创办了一种能批量"长出"公司的人。

学术上，他是IEEE会士、国家自然科学奖获得者，北大工学院称他创立了机器人非完整约束运动规划这一重要学术领域，被国际同行普遍采用或引用。产业上，固高2023年登陆深交所创业板，被媒体称为他"人生第一个IPO"；2019年他与汪滔共同获得IEEE机器人与自动化大奖，成为该奖2002年设立以来首批中国获奖者；松山湖XbotPark走出了大疆之外的李群自动化、逸动科技、海柔创新等一批硬科技企业，模式复制到宁波、重庆、长沙等八个城市。

他的命题仍在接受检验：创业者能不能被系统化培养？大湾区的供应链效率（他称迭代速度是硅谷的10到30倍、成本仅十分之一）能不能持续孵化出下一个大疆？这些问题尚无定论，但他的方法论已经留下了一层清晰的地基。

他的精神遗产或许可以归纳为三件事：第一，他证明了硬科技创业可以从课程开始，像一门学科一样被拆解、传授、复制；第二，他用固高二十四年、大疆十余年的案例链条，确立了"教授可以走下讲台去做供应链"的范式；第三，他把松山湖从一个东莞小镇变成了中国硬科技创业版图上绕不过去的地理坐标。这种遗产不以公司市值衡量，而以他培养了多少代人的思维方式为刻度。独立编委会评审认为，李泽湘的独特价值在于他既是学者又是系统架构师，用四十余年在学术和产业之间搭起了一座可以反复通行的桥梁。

"过去40年我们靠人口红利，未来40年靠工程师红利。"一个从湖南蓝山教师家庭走出来的少年，用四十多年把自己活成了工程师红利的铺路人。终身档案编码已为这份记录建立永久存证，档案记录他的方法、成果与待验证的部分，不作结论。

He often says the most important output of a university should be students, not papers. He never founded the biggest company himself—instead, he created a way of producing people who can found companies, batch after batch.

In academia he is an IEEE Fellow and a winner of China's National Natural Science Award; Peking University's College of Engineering notes that he founded the important field of nonholonomic motion planning in robotics, work widely adopted and cited internationally. In industry, Googol listed on the Shenzhen ChiNext board in 2023, called by the media his "first IPO in life"; in 2019 he and Frank Wang jointly received the IEEE Robotics and Automation Award, becoming its first Chinese laureates since the prize was established in 2002. Beyond DJI, XbotPark at Songshan Lake has produced hard-tech firms including Agile Robots, ePropulsion and Hai Robotics, and the model has been replicated across eight cities including Ningbo, Chongqing and Changsha.

His central proposition is still being tested: can entrepreneurs be systematically cultivated? Can the supply-chain efficiency of the Greater Bay Area—where, he says, iteration runs 10 to 30 times faster than in Silicon Valley at a tenth of the cost—keep producing the next DJI?

"For the past 40 years we relied on the demographic dividend; for the next 40, we rely on the engineer dividend." A boy from a teacher's family in Lanshan, Hunan, spent more than four decades becoming the one who paves the road for that dividend. This archive records his methods, results and the parts still awaiting verification—no verdict attached.

## 金句 Pull Quote

> 大学里最重要的成果应该是学生，而不是论文。

> The most important output of a university should be students, not papers.

## 履历时间线 Career Timeline

- **1961** 出生于湖南蓝山教师家庭 / Born into a teaching family in Lanshan, Hunan
  - 李泽湘出生于湖南省南部边界的蓝山县，母亲在小学任教，父亲是中学物理老师。[s2][s7]
  - EN: Li Zexiang was born in Lanshan County, on the southern border of Hunan; his mother taught at a primary school and his father was a middle-school physics teacher. [s2][s7]
- **1978** 考入中南矿冶学院 / Admitted to Central South Institute of Mining and Metallurgy
  - 恢复高考后的第一年，李泽湘考入中南矿冶学院（后发展为中南工业大学、中南大学）。[s2][s7]
  - EN: In the first year of the restored national college entrance examination, Li was admitted to the Central South Institute of Mining and Metallurgy, the predecessor of Central South University. [s2][s7]
- **1979** 作为中国首批公派本科生赴美 / Among China's first government-sponsored undergraduates sent to the U.S.
  - 美国铝业公司访华时提出为中方培养两名本科生，18岁的大一新生李泽湘被选中，赴卡内基梅隆大学就读，成为中国首批公派赴美本科生之一。他几乎没有英语基础，约六个月后即可跟班上课。[s1][s7]
  - EN: When ALCOA visited China and offered to sponsor two Chinese undergraduates, 18-year-old freshman Li was selected to study at Carnegie Mellon University, becoming one of China's first government-sponsored undergraduates in the U.S. With almost no English, he was able to follow classes after about six months. [s1][s7]
- **1983** 获卡内基梅隆大学双学士学位 / Earned dual B.S. degrees at Carnegie Mellon
  - 李泽湘作为优秀毕业生获卡内基梅隆大学电机工程与经济学双学士学位，并获大学学者奖（University Scholar Award）。[s2][s3]
  - EN: Li graduated with distinction from Carnegie Mellon University with dual bachelor's degrees in electrical engineering and economics, and received the University Scholar Award. [s2][s3]
- **1986** 获加州大学伯克利分校数学硕士学位 / Earned M.A. in mathematics at UC Berkeley
  - 李泽湘在加州大学伯克利分校继续深造，1986年获得数学硕士学位，其后转向电机工程与计算机方向。[s1][s2]
  - EN: Li pursued graduate studies at UC Berkeley, earning an M.A. in mathematics in 1986 before moving into electrical engineering and computer sciences. [s1][s2]
- **1989** 获伯克利电机工程与计算机博士学位 / Earned Ph.D. in EECS at UC Berkeley
  - 1989年李泽湘获加州大学伯克利分校电机工程与计算机博士学位，同年获得伯克利E.I. Jury奖。[s2][s3]
  - EN: In 1989, Li received his Ph.D. in electrical engineering and computer sciences from UC Berkeley, and was awarded the E.I. Jury Award the same year. [s2][s3]
- **1989** 在MIT人工智能实验室与纽约大学从事研究教学 / Research and teaching at MIT AI Lab and NYU
  - 博士毕业后，李泽湘先后在麻省理工学院人工智能实验室任研究科学家、在纽约大学机器人与制造实验室任助理教授；1990年获美国国家科学基金会研究创新奖（Research Initiation Award）。[s3][s2]
  - EN: After his doctorate, Li worked as a research scientist at the MIT Artificial Intelligence Laboratory and as an assistant professor in the Robotics and Manufacturing Laboratory at New York University; he received the NSF Research Initiation Award in 1990. [s3][s2]
- **1992** 加入香港科技大学，创办数控研究实验室与自动化技术中心 / Joined HKUST and founded the CNC Lab and Automation Technology Center
  - 1992年，李泽湘放弃美国教职，加入创校翌年的香港科技大学电子与计算机工程学系，创办数控研究实验室及自动化技术中心（3126实验室）并任主任。他选择香港的原因是这里毗邻内地，可以做产学研结合的事。[s1][s6][s5]
  - EN: In 1992, Li gave up a U.S. faculty position to join the Department of Electronic and Computer Engineering at HKUST, founded the year after the university's establishment. He set up and directed the CNC Research Laboratory and the Automation Technology Center (Lab 3126), choosing Hong Kong for its proximity to mainland China and its potential for industry-academia collaboration. [s1][s6][s5]
- **1993** 合著《机器人操作的数学导论》出版 / Co-authored A Mathematical Introduction to Robotic Manipulation
  - 李泽湘与Richard M. Murray、S. Shankar Sastry合著的《A Mathematical Introduction to Robotic Manipulation》（中文译名《机器人操作的数学导论》）出版，以微分几何、李群与旋量理论构建机器人操作的统一数学框架，成为机器人研究生课程的经典教材；其中译本1998年由机械工业出版社出版。[s3][s15]
  - EN: Li co-authored A Mathematical Introduction to Robotic Manipulation with Richard M. Murray and S. Shankar Sastry, building a unified mathematical framework for robotic manipulation on differential geometry, Lie groups and screw theory. It became a classic graduate robotics textbook; its Chinese translation was published by China Machine Press in 1998. [s3][s15]
- **1997** 获中国国家自然科学三等奖 / Received China's National Natural Science Award (Third Class)
  - 李泽湘因在机器人学领域的研究获得中国国家自然科学三等奖。他在机器人非完整约束环境下的运动规划方向发表了该领域第一篇论文、组织了第一次国际学术研讨会。[s3][s2]
  - EN: Li received China's National Natural Science Award (Third Class) for his robotics research. He published the first paper and organized the first international workshop on motion planning for robots under nonholonomic constraints. [s3][s2]
- **1999** 与高秉强、吴宏创办固高科技 / Co-founded Googol Technology with Gao Bingqiang and Wu Hong
  - 借深圳市政府、北京大学、香港科技大学共建深港产学研基地的契机，李泽湘与港科大同事高秉强、吴宏创办固高科技（Googol Technology），成为亚太区首家专注运动控制及控制系统的高科技公司。公司英文名GOOGOL意为10的100次方，创意来自李泽湘的数学家妻子。[s2][s6][s1]
  - EN: Seizing the opportunity of the Shenzhen-Hong Kong Industry-Education-Research Base co-founded by the Shenzhen government, Peking University and HKUST, Li co-founded Googol Technology with HKUST colleagues Gao Bingqiang and Wu Hong — Asia-Pacific's first high-tech company focused on motion control systems. The name GOOGOL, meaning 10 to the power of 100, was suggested by Li's wife, a mathematician. [s2][s6][s1]
- **2003** 机器人比赛课程中结识汪滔 / Met Wang Tao through the robotics competition course
  - 香港电视台到学校寻找老师带队参加机器人比赛，李泽湘接受了挑战；成绩一般但动手能力强的电子系本科生汪滔报名参加。汪滔2004至2005年带领港科大战队夺得Robocon香港冠军和亚太区季军，引起李泽湘注意。[s4][s8]
  - EN: When a Hong Kong TV station sought a faculty advisor for a robotics competition, Li took the challenge; Wang Tao, an average-grading but hands-on electronics undergraduate, signed up. Wang led the HKUST team to the Robocon Hong Kong championship and third place in the Asia-Pacific region in 2004-2005, catching Li's attention. [s4][s8]
- **2004** 创办哈工大深圳研究生院控制与机电工程学科部，破格招收汪滔 / Founded HIT Shenzhen control discipline division; admitted Wang Tao against the rules
  - 李泽湘与哈尔滨工业大学深圳研究生院合作创办控制与机电工程学科部并开始招生，每年约50人，七年培养350多名学生。同年他将ROBOCON机器人竞赛引入港科大课堂。汪滔本科毕业设计做直升机飞控系统仅得C，失去赴欧美名校深造机会，李泽湘破格招收其为硕士研究生，并提供全额奖学金。[s6][s8][s4]
  - EN: Li co-founded the Control and Mechatronic Engineering division at the Harbin Institute of Technology Shenzhen Graduate School, enrolling about 50 students a year and training more than 350 over seven years. The same year he brought the ROBOCON competition into HKUST's curriculum. When Wang Tao's undergraduate thesis on helicopter flight control received only a C, closing the door to top Western universities, Li admitted him as a master's student against convention with a full scholarship. [s6][s8][s4]
- **2005** 受聘长江学者讲座教授 / Appointed Changjiang (Cheung Kong) Scholar Chair Professor
  - 李泽湘获聘为教育部长江学者讲座教授。[s2][s3]
  - EN: Li was appointed a Changjiang (Cheung Kong) Scholar Chair Professor by China's Ministry of Education. [s2][s3]
- **2006** 支持汪滔创办大疆创新 / Backed Wang Tao in founding DJI
  - 26岁的汪滔与两位同学在深圳车公庙一间不足20平方米的仓库里创办大疆创新。李泽湘提供资金支持（早期与哈工大朱晓蕊教授累计提供约100万元）、人才支援（从哈工大深研院调派学生）和方向指导，并在其后担任大疆董事长。[s4][s8][s6]
  - EN: At 26, Wang Tao and two classmates founded DJI in a warehouse of less than 20 square meters in Chegongmiao, Shenzhen. Li provided funding (about 1 million yuan cumulatively in the early stage with Professor Zhu Xiaorui of HIT), personnel (students from HIT Shenzhen) and strategic guidance, and later served as DJI's chairman. [s4][s8][s6]
- **2007** 当选IEEE Fellow，任ICRA 2011总主席 / Elevated to IEEE Fellow; served as General Chair of ICRA 2011
  - 李泽湘当选国际电气与电子工程师学会会士（IEEE Fellow），为该学会最高等级会员；据松山湖管委会2018年发布的资料，当时其学术著作被引用次数已超过1.1万次，参与建立的工业机械臂数学建模理论为工业机器人界常用标准模型之一。他还长期服务学术共同体，2001至2004年任IEEE Transactions on Robotics and Automation副编辑，并担任2011年IEEE国际机器人与自动化大会（ICRA）总主席。[s3][s15][s2]
  - EN: Li was elevated to Fellow of the IEEE, the society's highest membership grade. According to 2018 data released by the Songshan Lake administration, his works had by then been cited more than 11,000 times, and the robotic-arm mathematical modeling theory he helped build is a standard model in industrial robotics. He served the community as an associate editor of IEEE Transactions on Robotics and Automation (2001-2004) and as General Chair of IEEE ICRA 2011. [s3][s15][s2]
- **2013** 携大疆精灵样机赴硅谷引入红杉资本 / Took Phantom prototypes to Silicon Valley and brought in Sequoia Capital
  - 大疆推出航拍一体机精灵（Phantom）系列后，国内投资机构兴趣有限，李泽湘拎着两台样机飞往硅谷，找到红杉资本主席迈克尔·莫里茨（Michael Moritz），为大疆引入顶级资本。莫里茨后将精灵2代比作飞行着的Apple II；精灵系列入选美国《时代》周刊2014年度十大科技产品。[s4][s8]
  - EN: After DJI launched the all-in-one Phantom aerial drone, domestic investors showed limited interest. Li carried two prototypes to Silicon Valley and met Sequoia Capital chairman Michael Moritz, bringing top-tier capital into DJI. Moritz later compared the Phantom 2 to a flying Apple II; the Phantom was named among Time magazine's top 10 tech products of 2014. [s4][s8]
- **2014** 联合创办松山湖国际机器人产业基地（XbotPark） / Co-founded the Songshan Lake XbotPark robotics base
  - 2014年11月，李泽湘联合港科大工学院原院长高秉强、长江商学院副院长甘洁等在东莞松山湖发起成立松山湖国际机器人产业基地（XbotPark机器人基地），由其团队负责投资、建设和运营管理，专注机器人及相关行业的创业孵化。基地早期资金部分来自几位发起人出售部分大疆股份所得。[s2][s15][s12]
  - EN: In November 2014, Li co-founded the Songshan Lake International Robotics Industry Base (XbotPark) in Dongguan with former HKUST engineering dean Gao Bingqiang and Cheung Kong Graduate School of Business deputy dean Gan Jie. His team ran its investment, construction and operations, focusing on robotics startup incubation. Early funding came partly from the founders selling part of their DJI shares. [s2][s15][s12]
- **2015** 设立香港清水湾创业基金（后更名XBOTPARK基金） / Established Clearwater Bay venture fund (later XBOTPARK Fund)
  - 为配套松山湖基地的孵化，李泽湘团队成立香港清水湾创业基金，后更名为XBOTPARK基金，为创业者提供从探索期、种子期到天使期的资金支持。同年香港科技大学机器人研究所成立，李泽湘为重要成员。[s8][s1]
  - EN: To support the Songshan Lake base, Li's team set up the Hong Kong Clearwater Bay startup fund, later renamed the XBOTPARK Fund, providing funding from exploration and seed stages through angel rounds. HKUST's Robotics Institute was also established that year, with Li as a key member. [s8][s1]
- **2016** 发起HONG KONG X科技创业平台 / Launched the Hong Kong X-Tech Startup Platform
  - 2016年7月，李泽湘与红杉资本全球执行合伙人沈南鹏、香港大学陈冠华教授等联合香港高校科技界十余人共同发起HONG KONG X科技创业平台暨青年创业服务系统；马化腾等参与支持。云鲸智能创始人张峻彬同年进入松山湖体系创业。[s2][s8][s11]
  - EN: In July 2016, Li co-launched the Hong Kong X-Tech Startup Platform and youth entrepreneurship service system with Sequoia Capital's Neil Shen (Shen Nanpeng), HKU professor Chen Guanhua and more than a dozen Hong Kong academics, with support including Pony Ma. Narwal (Yunjing) founder Zhang Junbin entered the Songshan Lake ecosystem to start his business the same year. [s2][s8][s11]
- **2017** 任香港政府创新、科技与再工业化委员会成员；创立希迪智驾 / Joined Hong Kong's I&T Committee; co-founded CiDi
  - 李泽湘被任命为香港政府创新、科技与再工业化委员会成员及港深（河套）科技园董事。同年10月，他与德州仪器原高管马潍博士在湖南长沙共同创立长沙智能驾驶研究院（希迪智驾，CiDi），聚焦封闭场景商用车自动驾驶。[s2][s17]
  - EN: Li was appointed a member of Hong Kong's Committee on Innovation, Technology and Re-industrialization and a director of the Hong Kong-Shenzhen Innovation and Technology Park in the Lok Ma Chau Loop. In October the same year, he and former Texas Instruments executive Dr. Ma Wei co-founded Changsha Intelligent Driving Institute (CiDi) in Hunan, focusing on autonomous driving for commercial vehicles in closed scenarios. [s2][s17]
- **2018** 获香港创新领军人物大奖；宣布获2019 IEEE机器人与自动化大奖 / Won Hong Kong Innovation Leadership Award; named for 2019 IEEE R&A Award
  - 2018年3月李泽湘获团结香港基金颁发的香港创新领军人物大奖。同年7月，IEEE宣布李泽湘与汪滔共同获得2019年IEEE机器人与自动化大奖，表彰二人对民用无人机发展与商业化、航空影像技术及机器人工程进步的贡献，这是该奖2002年设立以来首次授予中国学者与企业家。[s2][s14][s15]
  - EN: In March 2018, Li received the Hong Kong Innovation Leadership Award from the Our Hong Kong Foundation. In July, IEEE announced that Li and Frank Wang would jointly receive the 2019 IEEE Robotics and Automation Award for contributions to the development and commercialization of civilian drones, aerial imaging technology and robotics engineering — the first time the award, established in 2002, went to Chinese recipients. [s2][s14][s15]
- **2019** 在蒙特利尔ICRA领取IEEE大奖；宁波两院一园开园 / Received IEEE Award at ICRA Montreal; Ningbo campus opened
  - 2019年5月，李泽湘与汪滔在加拿大蒙特利尔ICRA 2019大会上正式领取IEEE机器人与自动化大奖。同年7月20日，宁波智能技术学院、宁波清水湾智能技术研究院和清水湾（宁波）智能产业园开园，XbotPark模式落地宁波。[s14][s7]
  - EN: In May 2019, Li and Wang formally received the IEEE Robotics and Automation Award at ICRA 2019 in Montreal. On July 20 the same year, the Ningbo Institute of Intelligent Technology, Ningbo Clearwater Bay Research Institute and Clearwater Bay (Ningbo) Intelligent Industrial Park opened, extending the XbotPark model to Ningbo. [s14][s7]
- **2020** 发起重庆明月湖国际智能产业科创基地；获深圳40年40人表彰 / Launched Chongqing Mingyue Lake base; named among Shenzhen's 40 in 40 years
  - 2020年6月，李泽湘与重庆两江新区共同发起明月湖国际智能产业科创基地，探索硬科技创新创业与新工科教育融合。同年他被评为深圳经济特区建立40周年创新创业人物和先进模范人物（深圳40年40人）。[s2][s3]
  - EN: In June 2020, Li co-launched the Mingyue Lake International Intelligent Industry Innovation and Entrepreneurship Base with Chongqing's Liangjiang New Area, integrating hard-tech incubation with new-engineering education. The same year he was named among the 40 innovation and entrepreneurship figures for the 40th anniversary of the Shenzhen Special Economic Zone. [s2][s3]
- **2021** 创办深圳科创学院并开学 / Founded and opened Shenzhen InnoX Academy
  - 2021年9月15日，李泽湘发起创办的深圳科创学院在深圳坪山开学，首批50名新生入学，探索以创业人才培养为核心导向的新工科教育；学院被列入深圳十四五规划重点支持项目，设置智能驾驶、智能建造、智慧生活、智慧健康和柔性制造五大研究中心。[s10][s19]
  - EN: On September 15, 2021, the Shenzhen InnoX Academy, founded by Li, opened in Pingshan, Shenzhen with its first cohort of 50 students, exploring a new-engineering education centered on cultivating entrepreneurs. Listed as a key supported project in Shenzhen's 14th Five-Year Plan, the academy set up five research centers covering intelligent driving, intelligent construction, smart living, smart health and flexible manufacturing. [s10][s19]
- **2023-08** 固高科技在深交所创业板上市 / Googol Technology listed on ChiNext, Shenzhen Stock Exchange
  - 2023年8月，由李泽湘、高秉强、吴宏共同创办的固高科技股份有限公司在深交所创业板上市（证券代码301510），成为李泽湘体系内首家A股上市公司，也被视为其人生第一个IPO。[s2][s13]
  - EN: In August 2023, Googol Technology, co-founded by Li, Gao Bingqiang and Wu Hong, listed on the ChiNext board of the Shenzhen Stock Exchange (stock code 301510), becoming the first A-share listed company in Li's ecosystem — described as his first IPO. [s2][s13]
- **2023-12** XbotPark机器人基地新总部在松山湖开园 / New XbotPark headquarters opened at Songshan Lake
  - 2023年12月21日，由李泽湘、高秉强、甘洁联合创办的XbotPark机器人基地新总部基地在东莞松山湖落成开园，形成集车间、供应商服务窗口、零部件供应链于一体的孵化空间。[s2][s12]
  - EN: On December 21, 2023, the new XbotPark headquarters, co-founded by Li, Gao Bingqiang and Gan Jie, opened at Songshan Lake in Dongguan — an incubation space integrating workshops, supplier service windows and components supply chains within walking distance. [s2][s12]
- **2025-12** 希迪智驾与卧安机器人先后登陆港交所 / CiDi and Woan Technology listed on HKEX in succession
  - 2025年12月19日，希迪智驾（3881.HK）在港交所主板挂牌，成为港股首家商用车智能驾驶上市公司，发行价263港元，募资净额约13.09亿港元；公司2018年起历经8轮融资、累计获投近15亿元，2024年初最后一轮融资估值约90亿元[s9]，李泽湘为控股股东（IPO前持股43.64%）。12月30日，2015年成立、2017年进入XbotPark体系孵化的卧安机器人（06600.HK）在港交所上市，李泽湘以非执行董事身份持股超10%并出席敲钟。[s16][s17][s13][s9]
  - EN: On December 19, 2025, CiDi (3881.HK) listed on the HKEX main board as Hong Kong's first listed commercial-vehicle intelligent driving company, priced at HK$263 per share with net proceeds of about HK$1.309 billion; after eight funding rounds since 2018 raising nearly 1.5 billion yuan, it was valued at about 9 billion yuan in its early-2024 round [s9], with Li as controlling shareholder (43.64% pre-IPO). On December 30, Woan Technology (06600.HK) — founded in 2015 and incubated in XbotPark from 2017 — listed on HKEX, with Li holding more than 10% as a non-executive director and attending the bell-ringing. [s16][s17][s13][s9]
- **2026-02** 在广东省高质量发展大会阐述新制造与C端创业 / Addressed Guangdong High-quality Development Conference on new manufacturing
  - 2026年2月24日，李泽湘在广东省高质量发展大会上表示中国消费至少还有4.6万亿美元增量空间，提出以国际知名消费品牌反向拉动B端产业链升级的新制造概念，并给出大湾区产品迭代速度为硅谷10至30倍、成本仅为十分之一的判断。其团队当时累计孵化超270家硬科技企业，总估值超3500亿元。[s11]
  - EN: On February 24, 2026, Li told the Guangdong High-quality Development Conference that China's consumption had at least US$4.6 trillion of incremental headroom, proposing a new-manufacturing concept in which global consumer brands pull B-end supply-chain upgrades. He noted the Greater Bay Area's product iteration speed was 10 to 30 times that of Silicon Valley at one-tenth the cost. His team had by then incubated more than 270 hard-tech companies with a combined valuation exceeding 350 billion yuan. [s11]
- **2026-07** 深圳科创学院与哈工大共建具身智能大湾区班 / InnoX Academy and HIT launched Embodied Intelligence GBA Class
  - 2026年，深圳科创学院联合哈尔滨工业大学设立具身智能（大湾区班），面向本科招生，中国工程院外籍院士黄捷任班主任，李泽湘任首席顾问。学院当时在孵近千名创业者与约百个创业项目，项目80%布局AI硬件、80%主攻海外市场，已与全球20余所高校建立深度合作。[s20]
  - EN: In 2026, InnoX Academy and the Harbin Institute of Technology launched the Embodied Intelligence (Greater Bay Area) Class for undergraduates, with CAE foreign academician Huang Jie as class head teacher and Li as chief advisor. The academy was then incubating nearly 1,000 entrepreneurs and about 100 startup projects — 80% in AI hardware and 80% targeting overseas markets — with deep partnerships with more than 20 universities worldwide. [s20]
- **2026-09-02** 入选《亚洲封面人物·格局篇·总第018期》，入档封面传记，封卷存档 / Selected for Asia Cover Figure · Vision · Vol.018; accessioned into the Cover Archive and sealed
  - 入选《亚洲封面人物·格局篇·总第018期》，分配封面传记唯一编码 ACF-00-00119，入档封面传记，封卷存档。
  - EN: Selected for Asia Cover Figure (Vision issue, Vol.018); assigned Cover Archive Unique Code ACF-00-00119, accessioned into the Cover Archive and sealed.

## 常问问答 FAQ

**Q1: 李泽湘是谁？他的主要身份是什么？**

A: 李泽湘1961年生于湖南蓝山，是机器人与自动化领域学者，现任香港科技大学电子与计算机工程学系教授、林高演教授席教授，同时担任XbotPark机器人基地发起人、深圳科创学院创始院长、固高科技董事长等职。他1992年加入创校翌年的香港科技大学，1999年创办运动控制企业固高科技，2006年起作为导师支持学生汪滔创办大疆创新并曾任大疆董事长，2014年创办松山湖国际机器人产业基地（XbotPark），2021年创办深圳科创学院。媒体常以"大疆教父"或"硬科技创业教父"称呼他，但他本人在2026年访谈中表示，具体身份不重要，"我们就做人才的培养、孵化这一件事"。[s1][s2][s5]

**Q1 (EN): Who is Li Zexiang, and what are his main roles?**

A: Born in Lanshan, Hunan in 1961, Li Zexiang is a robotics and automation scholar who serves as professor in the Department of Electronic and Computer Engineering at HKUST and holder of the Lam Ko Yin Professorship. He is also founder of the XbotPark robotics base, founding dean of the Shenzhen InnoX Academy and chairman of Googol Technology. He joined HKUST in 1992, co-founded the motion-control company Googol in 1999, mentored his student Frank Wang in founding DJI from 2006 and served as DJI's chairman, founded the Songshan Lake XbotPark base in 2014 and the InnoX Academy in 2021. Media often call him the "godfather of DJI" or "godfather of hard-tech entrepreneurship," but in a 2026 interview he said the title does not matter: "We just do this one thing — cultivating and incubating talent." [s1][s2][s5]

**Q2: 李泽湘的教育背景是怎样的？**

A: 李泽湘1978年考入中南矿冶学院（现中南大学），1979年作为中国首批公派赴美本科生之一进入卡内基梅隆大学，1983年获电机工程与经济学双学士学位；随后在加州大学伯克利分校深造，1986年获数学硕士学位，1989年获电机工程与计算机博士学位。博士毕业后，他在麻省理工学院人工智能实验室任研究科学家，并在纽约大学任助理教授，1992年回到亚洲加入香港科技大学。[s1][s2][s3]

**Q2 (EN): What is Li Zexiang's educational background?**

A: Li entered the Central South Institute of Mining and Metallurgy (now Central South University) in 1978. In 1979, as one of China's first government-sponsored undergraduates in the U.S., he enrolled at Carnegie Mellon University, earning dual B.S. degrees in electrical engineering and economics in 1983. He then pursued graduate study at UC Berkeley, receiving an M.A. in mathematics in 1986 and a Ph.D. in electrical engineering and computer sciences in 1989. After his doctorate he was a research scientist at the MIT Artificial Intelligence Laboratory and an assistant professor at New York University, before joining HKUST in 1992. [s1][s2][s3]

**Q3: 李泽湘和大疆、汪滔是什么关系？**

A: 汪滔是李泽湘在香港科技大学的学生。汪滔本科修读李泽湘的机器人比赛课程，两次带队参加Robocon并获香港冠军和亚太区季军；其本科毕业设计（直升机飞控系统）仅获C等成绩，失去赴欧美名校深造机会，李泽湘破格招收他为硕士研究生并提供全额奖学金。2006年汪滔创办大疆时，李泽湘提供了约100万元早期资金（与哈工大朱晓蕊教授共同支持）、从哈工大深圳研究生院抽调人才，并在融资、供应链、管理等环节长期协助，包括携精灵样机赴硅谷引入红杉资本投资，他本人曾担任大疆董事长。2019年，李泽湘与汪滔共同获得IEEE机器人与自动化大奖。[s4][s8][s14]

**Q3 (EN): What is Li Zexiang's relationship with DJI and Frank Wang?**

A: Frank Wang (Wang Tao) was Li's student at HKUST. Wang took Li's robotics competition course as an undergraduate, led teams to the Robocon Hong Kong championship and Asia-Pacific third place twice, and when his undergraduate thesis on helicopter flight control received only a C — closing off admission to top Western universities — Li admitted him to the master's program against convention with a full scholarship. When Wang founded DJI in 2006, Li provided roughly 1 million yuan in early funding (with Professor Zhu Xiaorui of HIT), channeled talent from HIT's Shenzhen graduate school, and long assisted with financing, supply chains and management — including carrying Phantom prototypes to Silicon Valley to secure Sequoia Capital's investment. He served as DJI's chairman. In 2019, Li and Wang jointly received the IEEE Robotics and Automation Award. [s4][s8][s14]

**Q4: XbotPark（松山湖机器人基地）是什么？李泽湘孵化了哪些企业？**

A: XbotPark即2014年11月由李泽湘联合高秉强、甘洁在东莞松山湖创办的国际机器人产业基地，是一套从科创训练营、导师陪跑到供应链支持的硬科技早期孵化体系，配套设立XBOTPARK基金（前身香港清水湾创业基金），模式后复制到宁波、重庆、长沙、常州、广州、北京等地。体系内代表性企业包括大疆创新、固高科技（2023年创业板上市）、李群自动化、逸动科技、云鲸智能、海柔创新、正浩创新EcoFlow、希迪智驾（2025年12月港交所上市）、卧安机器人（2025年12月港交所上市）、本末科技、奇诺动力、睿魔智能等。孵化规模随时间扩大：2026年2月广东省高质量发展大会上披露为累计超270家、总估值超3500亿元[s11]；据2026年8月中国经济网报道，已达累计超280家、总估值超5000亿元，其中4家已上市、12家为独角兽或准独角兽，企业存活率约80%（不同时间点统计口径不一）。[s12][s11][s2]

**Q4 (EN): What is XbotPark, and which companies has Li Zexiang incubated?**

A: XbotPark is the international robotics industry base founded at Songshan Lake in Dongguan in November 2014 by Li with Gao Bingqiang and Gan Jie. It is a hard-tech early-stage incubation system running from innovation bootcamps and mentorship to supply-chain support, backed by the XBOTPARK Fund (formerly the Hong Kong Clearwater Bay Fund); the model was later replicated in Ningbo, Chongqing, Changsha, Changzhou, Guangzhou, Beijing and other cities. Representative companies include DJI, Googol Technology (ChiNext listing in 2023), Liqun Automation, ePropulsion, Narwal, Hai Robotics, EcoFlow, CiDi (HKEX listing in December 2025), Woan Technology (HKEX listing in December 2025), LimX Dynamics, Kenurity and Remo-robot. The portfolio grew over time: at the February 2026 Guangdong High-quality Development Conference, Li reported over 270 incubated companies with a combined valuation above 350 billion yuan [s11]; according to an August 2026 China Economic Net report, the figure exceeded 280 companies with a combined valuation above 500 billion yuan, including 4 listed firms and 12 unicorns or quasi-unicorns, with a survival rate around 80 percent (figures vary by measurement date). [s12][s11][s2]

**Q5: 李泽湘的新工科教育是什么？深圳科创学院是什么机构？**

A: 新工科教育是李泽湘从2004年前后在港科大机器人比赛课程和哈工大深圳研究生院教改班中发展出的培养方法：以真实产业问题为起点，通过跨学科混编、项目制学习和快速迭代，培养能完成从0到1创新的工程创业者，核心能力是他所说的"工程意识"（设计思维、动手能力、数学工具）。深圳科创学院是2021年在深圳市政府支持下成立、由李泽湘任创始院长的新型创新创业学院，列入深圳十四五规划，2021年9月首批50名学生入学，设智能驾驶、智能建造、智慧生活、智慧健康、柔性制造五大研究中心，通过科创训练营选拔学生并提供全额奖学金。到2026年，学院已与全球20余所高校（含港科大、港大、斯坦福、沙特国王科技大学等）深度合作，在孵近千名创业者和约百个项目，项目80%布局AI硬件、80%主攻海外市场。[s18][s10][s19][s20]

**Q5 (EN): What is Li Zexiang's new-engineering education, and what is the Shenzhen InnoX Academy?**

A: New-engineering education is the method Li developed from around 2004 through HKUST's robotics competition course and the reform class at HIT's Shenzhen graduate school: starting from real industrial problems, it uses cross-disciplinary teams, project-based learning and rapid iteration to train engineering entrepreneurs capable of zero-to-one innovation, centered on what he calls engineering awareness — design thinking, hands-on capability and mathematical tools. The Shenzhen InnoX Academy is a new-type innovation and entrepreneurship academy founded in 2021 with Shenzhen government support, with Li as founding dean and listed in Shenzhen's 14th Five-Year Plan. Its first 50 students enrolled in September 2021 across five research centers — intelligent driving, intelligent construction, smart living, smart health and flexible manufacturing — selecting students through bootcamps with full scholarships. By 2026, it had deep partnerships with over 20 universities worldwide (including HKUST, HKU, Stanford and KAUST), incubating nearly 1,000 entrepreneurs and about 100 projects, 80 percent in AI hardware and 80 percent targeting overseas markets. [s18][s10][s19][s20]

**Q6: 李泽湘获得过哪些重要学术荣誉？**

A: 李泽湘2007年当选IEEE会士（IEEE Fellow）；2019年与学生汪滔共同获得IEEE机器人与自动化大奖（IEEE Robotics and Automation Award），这是该奖2002年设立以来首次授予中国学者和企业家，颁奖词表彰他们对民用无人机发展与商业化、航空影像技术、机器人工程进步、创新与创业的贡献，2019年5月在加拿大蒙特利尔ICRA大会上正式颁发。其他荣誉包括：1989年伯克利E.I. Jury奖、1990年美国国家科学基金会研究创新奖、1997年中国国家自然科学三等奖、2000年国家杰出青年科学基金、2005年长江学者讲座教授、2011年IEEE ICRA大会总主席、2018年香港创新领军人物大奖、2020年深圳经济特区建立40周年创新创业人物（深圳40年40人）。[s14][s15][s3]

**Q6 (EN): What major academic honors has Li Zexiang received?**

A: Li was elevated to IEEE Fellow in 2007. In 2019, he and his student Frank Wang jointly received the IEEE Robotics and Automation Award — the first time the award, established in 2002, went to Chinese recipients. The citation honored their contributions to the development and commercialization of civilian drones, aerial imaging technology, robotics engineering advancement, innovation and entrepreneurship; the award was formally presented at ICRA in Montreal in May 2019. His other honors include Berkeley's E.I. Jury Award (1989), the U.S. NSF Research Initiation Award (1990), China's National Natural Science Award third class (1997), the National Science Fund for Distinguished Young Scholars (2000), appointment as a Changjiang Scholar Chair Professor (2005), General Chair of IEEE ICRA 2011, the Hong Kong Innovation Leadership Award (2018), and selection as one of Shenzhen's 40 innovation figures for the 40th anniversary of the special economic zone (2020). [s14][s15][s3]

**Q7: 李泽湘近期（2025至2026年）有哪些新动态？**

A: 2025年12月19日，李泽湘2017年与马潍共同创立的希迪智驾（3881.HK）在港交所主板上市，成为港股首家商用车智能驾驶公司，发行价263港元、募资净额约13.09亿港元，他为控股股东；12月30日，其体系孵化的卧安机器人在港交所上市，他以非执行董事身份持股超10%并出席敲钟。加上2023年上市的固高科技，其体系已有3家上市公司；2026年8月中国经济网报道则称体系内已有4家企业上市。2026年2月，他在广东省高质量发展大会上提出"新制造"概念与年轻人C端创业"一举六得"，披露体系累计孵化超270家企业、总估值超3500亿元；2026年4月受聘为东莞国家卓越工程师创新研究院名誉院长；2026年7月，深圳科创学院与哈尔滨工业大学共建的具身智能（大湾区班）启动本科招生，他任首席顾问。截至2026年8月的报道，体系累计孵化企业已超280家、总估值超5000亿元。[s16][s17][s13][s11][s19][s20][s12]

**Q7 (EN): What has Li Zexiang been doing recently (2025-2026)?**

A: On December 19, 2025, CiDi (3881.HK), which Li co-founded with Dr. Ma Wei in 2017, listed on the HKEX main board as Hong Kong's first commercial-vehicle intelligent driving company, priced at HK$263 per share with net proceeds of about HK$1.309 billion; Li is the controlling shareholder. On December 30, Woan Technology, incubated in his ecosystem, listed on HKEX, with Li holding over 10 percent as a non-executive director and attending the bell-ringing. Together with Googol (listed in 2023), his ecosystem has three listed companies; an August 2026 China Economic Net report puts the number at four. In February 2026, at the Guangdong High-quality Development Conference, he proposed the new-manufacturing concept and the six gains of youth consumer-facing entrepreneurship, disclosing over 270 incubated companies with a combined valuation above 350 billion yuan. In April 2026 he was named honorary dean of the Dongguan National Academy of Outstanding Engineers; in July 2026 the Embodied Intelligence (Greater Bay Area) Class co-built by InnoX Academy and HIT began undergraduate enrollment with Li as chief advisor. By August 2026 reports, the ecosystem had incubated over 280 companies with a combined valuation above 500 billion yuan. [s16][s17][s13][s11][s19][s20][s12]

## 来源清单 Sources

1. [s1] 香港科技大学官网 HKUST · Named Professorship at HKUST：林高演教授席 李泽湘教授 — https://np.hkust.edu.hk/sc/top_minds_details.php?id=59
2. [s2] 头条百科 · 李泽湘（固高科技有限公司董事长）百科词条 — https://m.baike.com/wiki/%E6%9D%8E%E6%B3%BD%E6%B9%98/1480166
3. [s3] 北京大学工学院官网 · 李泽湘——北京大学工学院先进制造与机器人系客座教授简介 — https://www.coe.pku.edu.cn/teaching/all_time/10772.html
4. [s4] 南方日报 南方+ · 学院派创业公司是如何崛起的？大疆孵化者李泽湘告诉你 — http://static.nfapp.southcn.com/content/201802/06/c954391.html
5. [s5] 虎嗅网（转自《中国企业家》） · 对话李泽湘：如何批量制造硬科技创业者 — https://m.huxiu.com/article/4881926.html
6. [s6] 人民网 · 大疆创新董事长李泽湘：大疆和汪滔只是一个开始 — http://hm.people.com.cn/n1/2017/0619/c42272-29348000.html
7. [s7] 宁波日报 海曙新闻报 · 在宁波，孵化更多大疆——记专攻机器与自动化领域的学者李泽湘 — http://hsxwb.cnnb.com.cn/resfile/2022-11-14/04/04.pdf
8. [s8] 澎湃新闻 · 大疆教父，正在成为IPO发动机 — https://m.thepaper.cn/newsDetail_forward_32250756
9. [s9] 创业邦 · 大疆教父回湖南老家，打造90亿估值独角兽，即将IPO — https://m.cyzone.cn/article/781758.html
10. [s10] 深圳新闻网 坪山发布 · 50名新生首批登陆科创星球：深圳科创学院坪山开学，率先探索新工科教育 — https://www.sznews.com/content/mb/2021-09/15/content_24573453.htm
11. [s11] 南方+ · 深圳科创学院院长李泽湘：进一步发挥年轻人的作用，打造全球科创的广东样板 — https://www.nfnews.com/content/LozB1BZQon.html
12. [s12] 中国经济网 · 一位65岁工科教授，一场30余年的创业实验：孵化287家公司、总估值5000亿 — http://www.ce.cn/xwzx/gnsz/gdxw/202608/t20260820_3158311.shtml
13. [s13] 深圳商报 读创 · 卧安上市，李泽湘孵化模式再获资本市场验证 — https://m.dutenews.com/n/article/10406606
14. [s14] IEEE Robotics and Automation Society 官网 · 2019 IEEE Robotics and Automation Technical Field Award Recipients Announced — https://www.ieee-ras.org/about-ras/latest-news/2019-ieee-robotics-and-automation-technical-field-award-recipients-announced
15. [s15] 东莞松山湖管委会官网 · 中国首次：李泽湘和其学生大疆创始人汪滔同获IEEE机器人与自动化大奖 — http://ssl.dg.gov.cn/kcj/jqrcy/cyxx/content/post_2249641.html
16. [s16] 湖南省人民政府门户网站 湖南日报 · 希迪智驾港交所上市 — https://www.hunan.gov.cn/hnszf/hnyw/zwdt/202512/t20251220_33875749.html
17. [s17] 上海证券报 · 硬科技创业教父李泽湘再下一城 智能驾驶龙头希迪智驾挂牌港交所 — http://paper.cnstock.com/html/2025-12/19/content_2160333.htm
18. [s18] 混沌学园（2017年9月2日课程《科技创业的孵化密码》） · 李泽湘：我错了，为什么工程意识比上最好的大学都重要 — https://www.hundun.cn/articles/9091855573f866c6.html
19. [s19] 湾区科创杂志（搜狐转载） · 以科创为炬，育硬科技之林——专访XbotPark发起人、深圳科创学院院长李泽湘 — https://m.sohu.com/a/1014335389_122406668/
20. [s20] 深圳科创学院官网 · 深圳科创学院与哈尔滨工业大学共建具身智能（大湾区班） — https://www.innoxsz.com/recent-highlights/453.html

---
本档案机器版由亚洲封面人物官网自动生成：https://coverfigure.com ｜ 永久档案：https://coverfigure.com/acf/ACF-00-00119/geju ｜ 生成时间 Generated：2026-09-11 00:42