Led by Meridian Capital, another Tsinghua-affiliated embodied intelligence company, "Lingyu Intelligence", has completed a seed + round of financing in the tens of millions of yuan.
Recently, Lingyu Intelligence, an embodied intelligence enterprise affiliated with Tsinghua University, announced the completion of a seed + round of financing worth tens of millions of yuan. This round was led by Huaying Capital, with participation from existing shareholders Inno Angel Fund, Shuimu Tsinghua Alumni Seed Fund, and Telescope Venture Capital. The funds from this round of financing will be used for the finalization of the mass - production version of the company's core product, TeleAvatar, and the initial trial production of prototypes.
Lingyu Intelligence was founded in February 2025 by a team with backgrounds from Tsinghua University and Carnegie Mellon University. The company focuses on high - quality teleoperation technology and has launched a teleoperation robot system featuring low cost (in the range of ten thousand yuan), low latency, high precision, and high usability. With the mission of "creating a benchmark for the practical application of embodied intelligence and bringing robots with reliable and flexible operation capabilities into every household", it is committed to promoting the rapid implementation of robots in the service and industrial and commercial sectors. Through a three - step strategy of "selling the robot body - selling data - selling capabilities", the company aims to ultimately drive the development of fully autonomous robots.
Team background of "top - notch scientific research + industrial implementation"
Entering the embodied intelligence industry from the teleoperation direction
Lingyu Intelligence is another embodied intelligence enterprise affiliated with Tsinghua University. Its founding team is characterized by "top - notch scientific research + industrial implementation".
Mo Yilin, the founder & chief scientist, is currently a tenured associate professor in the Department of Automation at Tsinghua University. He was previously an assistant professor in the EEE Department at Nanyang Technological University, Singapore. He holds a bachelor's degree from the Department of Automation at Tsinghua University, a doctorate from the Department of Computer Engineering at Carnegie Mellon University, and completed his post - doctoral research at the California Institute of Technology. He studied under Professor R. M. Murray, a member of the US National Academy of Engineering and a pioneer in the field of robot manipulation. Jin Ge, the CEO, also graduated from the Department of Automation at Tsinghua University and has dual experience in high - tech entrepreneurship investment and corporate management. He has served as a managing partner at Telescope Venture Capital and the vice - president of Aoliang Photonics. He is skilled at promoting the commercialization of technologies from the laboratory.
For nearly two decades, Mo Yilin has been dedicated to research in the field of control. Despite the increasing complexity of modern control theory, the PID control method, which was refined last century, is still widely used in the industrial sector. In response to the challenges of implementing control theory in practical applications, Professor He Yuqi from Harvard University and a member of the US National Academy of Engineering even once asked the profound question, "Control is dead?" To break through this predicament, Mo Yilin began to focus on fields closely related to control theory with broader applications, and robot technology came into his view. However, the specific direction to enter was still unclear at that time.
It wasn't until March 2024 that research institutions such as Stanford University officially released ALOHA 2, which caught Mo Yilin's attention and focused it on teleoperation. ALOHA 2 is an upgraded version of ALOHA, a low - cost open - source hardware system. Its emergence marked a new era in remote operation technology. Compared with ALOHA, ALOHA 2 has significantly improved performance, ergonomic design, and robustness, while its cost is only about 190,000 yuan.
In Mo Yilin's view, the development path of humanoid robots can be compared to that of autonomous driving, which needs to go through the stages from L0 to L2 and then to L4. Teleoperation is like the assisted - driving stage and is a necessary step for fully autonomous robots. "I think embodied intelligence will stay in the L2 stage (comparable to autonomous driving) for a long time, which is a stage of human - machine co - driving."
So, at the beginning of founding Lingyu Intelligence, the founding team decided to focus on the teleoperation direction.
Develop a
"Easy - to - operate and user - friendly" robot within half a year
Set the lowest selling price in the industry
Currently, the implementation of teleoperation faces two key constraints - "latency" and "cost". The Lingyu Intelligence team has achieved significant breakthroughs in both aspects through years of technological accumulation.
Regarding low latency, the company can now provide an almost "imperceptible" operating experience. According to actual tests, the image latency of its teleoperation system is as low as 100 milliseconds, and the operation latency is only 50 milliseconds. This indicator has reached the level of "no obvious perception of latency in human operation" and can meet the requirements of scenarios with high real - time requirements, such as precision assembly and remote control.
Regarding low cost, Lingyu Intelligence has reduced costs across the entire chain from hardware to data collection.
On the hardware side, the company's TA lite (TeleAvatar Lite) model has its price controlled within 100,000 yuan, which is much lower than similar teleoperation devices in the industry. Just three months after its establishment, Lingyu Intelligence launched a robot prototype with "a dexterous operation end + dual seven - axis robotic arms + a wheeled chassis". In July, the company launched the TA lite with a starting price of 79,900 yuan, setting the lowest price for currently mass - produced and retailed wheeled humanoid robots. Lingyu TA lite integrates the latest scientific research results of Lingyu Intelligence in core fields such as robot motion control and artificial intelligence algorithms. In addition to its low price, it also has features such as high reliability, high versatility, high practicality, high - value data, and ease of use.
- TeleAvatar Lite -
The reason for such a low selling price is mainly that the team started from the perspective of practical task goals and combined its technical advantages in the field of motion control. While ensuring the working ability, it reduced some "human - like" and "aesthetic" requirements and redesigned and redefined the robot body with practicality as the core goal.
On the data collection side, Lingyu Intelligence uses a VR device solution with a cost as low as a few thousand yuan. This solution avoids the pain points of "million - level cost + space limitations" of motion - capture devices and solves the problems of "many occlusions and small range" of pure vision solutions, achieving a balance between low cost and high - precision data collection.
Mo Yilin believes that VR devices will become the "mainstream direction" of data collection. "The industry is converging towards VR - based data collection. Leading companies such as Tesla and NVIDIA are all deploying this route. The VR solution can balance cost and precision and conforms to human operation habits, and will become the core method of future robot data collection."
Whether it is reducing latency or cost, Lingyu Intelligence's core goal is to create a "easy - to - operate and user - friendly" robot. Because only by lowering the operation threshold of teleoperation can large - scale and consistent real - machine data be quickly accumulated, which is the core barrier for the iterative improvement of robot autonomy.
- TeleAvatar Lite Practical operation demonstration -
Currently, data remains a bottleneck in the embodied intelligence industry. The amount of data in the global robotics field is 3 - 5 orders of magnitude lower than that in autonomous driving and large - language models, and it faces three problems: poor quality, low efficiency, and high cost. The breakthrough of embodied intelligence depends on data, but data collection needs to meet two core conditions : first, the scenarios should be close to reality, and second, the teleoperation actions should be close to human habits.
Regarding the debate between "real - machine data and simulation data", Mo Yilin said: Simulation data is an amplifier of real - machine data. Without the "1" of real - machine data, no matter how many "0"s of simulation data there are, they are meaningless. He believes that the core advantage of real - machine data lies in "accuracy", while simulation data can help real - machine data achieve low - cost generalization (such as generating multi - scenario data through optical and physical simulations).
After clarifying the progressive relationship among hardware, data, and autonomous capabilities, Lingyu Intelligence has designed a clear commercialization path. Through a three - step strategy of "increasing hardware quantity - accumulating data - outputting capabilities", it gradually builds long - term competitive barriers.
Specifically, in the first step, the company will quickly capture the market by "selling the robot body". Starting with robot hardware featuring "high usability and high cost - effectiveness", it will quickly cover service industries (such as remote inspection and unmanned retail) and industrial and commercial sectors (such as workshop assembly and logistics sorting) to accumulate initial users and scenario bases through large - scale implementation.
When the market share of the robots reaches a certain scale, the "differentiated real - machine data" generated in various scenarios will become the core asset, and the company will enter the "selling data" stage. This data can not only feed back into the iterative improvement of intelligent models but also form a scarce data barrier in the industry, avoiding homogeneous competition.
With the accumulation of data and the optimization of models, Lingyu Intelligence will start to output solutions and enter the "selling capabilities" stage. It will upgrade from "a hardware manufacturer" to "an intelligent - capability provider", outputting standardized teleoperation control capabilities and scenario - based autonomous operation algorithms to the industry, and ultimately achieving the flywheel effect of "data - capabilities - business".
Currently, Lingyu Intelligence is actively expanding in the "selling the robot body" stage. The first batch of practical test scenarios include special industries, the commercial sector, and the industrial field including logistics sorting. Relevant customers have entered the stage of waiting for prototype testing.
Regarding this round of financing, Jin Ge, the CEO of Lingyu Intelligence, said: "This round of financing has added an important fuel for the cold - start of Lingyu Intelligence, which has just entered the embodied intelligence track, and helps us further consolidate our technological advantages. Within the next year, we will complete the mass production of our high - quality robots. Combined with our self - developed autonomous operation model, we will put TeleAvatar into specific production and living scenarios to contribute to the next milestone of robot data and intelligence."
Ji Wei, the founding managing partner of Huaying Capital, said: Huaying Capital has always firmly believed in the long - term strategic value of the embodied intelligence track and has been focusing on finding high - quality entrepreneurial teams that can systematically improve the overall technical level of embodied devices and solve the difficulties in operation and collaboration and the pain points in robot data collection. The Lingyu Intelligence team is an excellent example that meets the above two points: Professor Mo, the founder, has the world's top - notch scientific research capabilities in the field of robot motion control, and CEO Jin Ge has high - efficiency execution ability on the business side. As one of the earliest investors to contact the team, we witnessed the company's development from 0 to 1 in researching and producing a prototype of a low - latency, high - cost - effectiveness teleoperation embodied intelligence system within just a few months. We believe that with the team's advantage of "top - notch scientific research + industrial implementation", the company will become the driving force for the upgrade of basic technologies in the field of embodied intelligence and has the potential to grow into an indispensable core technology supporter and leader in the industry.