① Against the backdrop of a rising tide in interdisciplinary research, popular fields such as AI and embodied intelligence are gaining significant attention from academia, with major companies like Tencent and Xiaomi actively supporting fundamental scientific research; ② Companies such as Unitree Robotics and DeepRobotics have expressed strong willingness to collaborate with academia, aiming for differentiated breakthroughs; ③ Several interviewed scientists stated they are using or learning AI tools, with some researchers having adapted to domestic computing power and collaborating with Huawei.
Cailian Press, September 25 (Reporter Jing Fu) “Many of today’s questions are about embodied intelligence. People increasingly see me as an industry expert. But essentially, I am still a scientist and engineer, so I can only share my own perspectives.” Professor Fan Zhiyong from the Department of Electronic and Computer Engineering and the Department of Chemical and Biological Engineering at the Hong Kong University of Science and Technology, along with his team, is exploring potential applications of cutting-edge biomimetic optoelectronic and sensor devices in the field of embodied intelligence. During a group interview at the 2025 New Cornerstone 50² Forum, he responded to frequent media inquiries on embodied intelligence with these remarks.
The New Cornerstone 50² Forum, co-hosted by Tencent’s Sustainable Social Value Division (SSV), the New Cornerstone Science Foundation, and Southern University of Science and Technology, has continuously provided a cross-academic exchange platform for scientists in various fields of fundamental research and frontier technologies over the past five years, with a commitment to advancing breakthroughs in basic scientific research and fostering deep interdisciplinary collaboration.
Cailian Press reporters learned during the forum that, against the background of a surge in interdisciplinary research, popular domains such as AI and embodied intelligence have also attracted significant attention from frontline researchers, forming new perspectives distinct from those within the industrial sector.
It is understood that although embodied intelligence is currently at the forefront, its industrial implementation progress has not met expectations due to constraints such as the absence of large-scale application scenarios and the need for improved intelligence levels. Many leading robotics companies are now engaging closely with academia to seek new breakthroughs.
Fan Zhiyong revealed during the interview, “I know there are perhaps three to four hundred companies in China working on embodied intelligence. This year, I have visited around ten such companies, including Unitree Robotics, DeepRobotics, Dobot, Standard Robots, and Juniper Automation (688306.SH). I have reviewed and personally evaluated many of these enterprises, all of which have expressed a strong desire for collaboration.”
“Nowadays, everyone might be equipped with cameras, dexterous hands, or other inertial sensors, all seeking differentiated advantages. Novel sensors represent one aspect of this.” He further elaborated.
Does the spotlight on the embodied intelligence industry also bring pressure? Fan Zhiyong told Cailian Press, “Companies engaged in embodied intelligence certainly face some pressure. The current phase is still about ‘attracting attention.’ Who will buy so many robots, placing orders for tens of thousands or even hundreds of thousands of units at once? Truly large-scale application scenarios for embodied intelligence have yet to emerge.”
From the perspective of bionics, he analyzed the current state of sensor devices in the field of embodied intelligence: “Embodied intelligence itself serves as an excellent platform, integrating a wide variety of sensors, including vision, hearing, and touch. However, it lacks certain sensors, such as olfactory sensors that convert chemical and biological signals into electrical ones. Humans possess olfactory and gustatory sensors, but embodied intelligence does not, pointing to future research directions. Even the sensors already in use in embodied intelligence, such as visual sensors, are still based on planar CCD or CMOS technology, which differs significantly from human visual perception.”
He illustrated by stating that the human skin contains a vast number of sensors, whereas embodied intelligence tactile sensors are predominantly used in dexterous hands at this stage, without full coverage of the robot’s skin. This involves numerous challenges such as ‘not being able to produce them cheaply, covering large areas, achieving sufficient flexibility, and processing such a multitude of signals.’
In addition, reporters from Cailian Press learned that the olfactory chip and electronic nose chip solutions proposed by Fan Zhiyong’s team have been adopted by Junpu Intelligence and have caused a sensation overseas.
Not only has bionics provided a direction for embodied intelligence, but reporters from Cailian Press also learned at the 2025 New Foundation 50² Forum that the currently popular AI has garnered significant attention in the scientific community, with multiple scientists indicating they are conducting interdisciplinary research in related fields.
Gao Yiqin, a professor at Peking University specializing in molecular science computing, introduced, “The amount of data in the biomedicine field approximately doubles every six months. A large amount of data has not been adequately analyzed, particularly regarding the underlying molecular mechanisms. There is substantial work to be done. The advancement of AI technology may bring new opportunities for organizing and understanding this data. An important mission of our research group is to build a scientific computing platform within the AI framework, which will enable highly precise quantum chemistry calculations to obtain microscopic information and perform dynamic calculations, among others.”
At the computational power level, Gao Yiqin also mentioned that his research group has been working with several domestic hardware manufacturers such as Huawei Kunpeng, Ascend, Hygon, Moore Threads, and Muxi for compatibility. When discussing the results, he stated, “With current software and hardware co-optimization, we can now achieve computational capabilities comparable to NVIDIA A100. Our collaboration with the Huawei team, fully utilizing domestically produced software and hardware for training from scratch, has achieved results on par with or even surpassing those of Google’s team, both in single structures and complex predictions.”
Xie Xiaoliang, an academician of the Chinese Academy of Sciences and director of Changping Laboratory, also stated that big data and artificial intelligence are profoundly transforming the research paradigm of biomedicine, providing sustainable solutions to long-standing health risks faced by humanity, such as emerging epidemics, cancer, and Alzheimer’s disease.
Zheng Nanfeng, an academician of the Chinese Academy of Sciences and professor at Xiamen University, who has long been engaged in research on surface and interface chemistry of functional materials, told reporters from Cailian Press and other media outlets that his academic research “encounters a hurdle roughly every three years because what I’ve learned is no longer sufficient.” According to him, his current research also employs AI methods, “including material development and new energy fluctuation prediction, all using AI.”
Xi Dan, Senior Vice President and Chief Talent Officer of Tencent, stated during an interview with Cailian Press and other media that interdisciplinary integration and mutual permeation between fields are becoming the core engine of technological innovation and may even define new research paradigms of the future.
It is worth noting that many leading technology companies are placing emphasis on funding fundamental scientific research. For example: one of the highest-funded youth science talent sponsorship programs in China, the ‘Science Exploration Award,’ is funded by Tencent and, by the end of 2025, will have supported 347 outstanding young scientists; the Xiaomi Foundation for Public Welfare, affiliated with Beijing Xiaomi, jointly established the public welfare talent sponsorship program ‘Xiaomi Young Scholars’ with universities, planning to donate a total of RMB 500 million to support full-time young teachers and researchers at universities, with a focus on computer science, communications, electronics, and other fields.
Xi Dan stated that Tencent hopes to leverage the tolerance and flexibility of private funding while continuing to foster its innovative spirit as a tech company, serving as a beneficial supplement to national scientific investment. “When Tencent decided to establish SSV, it prioritized funding fundamental scientific research as one of its core directions in creating societal value, without any commercial requirements or profit objectives. We hope to explore a new path for private funds and enterprises to sponsor basic research and aim to set a benchmark.”