Guangdong Zhishen Data Information Technology has raised 30 million yuan, or roughly $4.2 million, in a first-round strategic investment from Universal Intelligent Control (Hong Kong) International R&D Co., Ltd. The announcement came on November 14, 2026, and the money is earmarked for one narrow but increasingly valuable problem: there isn't enough real-world data of humans walking, climbing and balancing for robot makers to train on.
The company builds wearable capture devices that record human leg and lower-body movement outside a lab. That distinction matters more than it sounds. Most motion-capture systems used to train humanoid robots today rely on fixed camera rigs in controlled studios, which means the data they produce looks nothing like a sidewalk, a staircase or a cluttered warehouse floor.
Why does leg data matter so much right now?
Humanoid robot companies have spent the past two years mostly solving for hands and arms — grasping, assembly, manipulation. Walking reliably across uneven, unpredictable terrain has lagged behind. Training a robot to balance and recover from a stumble requires thousands of hours of footage showing how real human bodies shift weight, adjust stride length and respond to obstacles, and that footage is in short supply.
"This partnership will address the critical shortage of high-quality human motion datasets for humanoid robot model training while accelerating algorithm iteration and product commercialization," the companies said in their joint announcement, according to Sina Finance.
Who is the investor, and why does it matter?
Universal Intelligent Control is not a typical venture fund. According to the same report, its client roster already includes Tesla, Unitree, ABB and UBTech — four of the most prominent names building humanoid and industrial robots globally. That gives Guangdong Zhishen Data a built-in distribution channel: rather than selling datasets cold, it can plug directly into an investor that already has relationships with robot makers hungry for training data.
Guangdong Zhishen Data says it will use the capital to expand hardware R&D, iterate multi-part wearable capture products that go beyond legs to cover more of the body, scale up real-world data production capacity, and drive down the cost per hour of usable footage.
How crowded is the embodied-data market already?
Very. China alone has seen roughly 15 independent embodied-data companies raise a combined 4.47 billion yuan (about $625 million) over the past year, according to industry tracking cited by Chinese tech outlets. Rivals include Zhiyuan Robotics subsidiary Mifeng Technology, which claims 1-millimeter trajectory restoration accuracy with its MEgo wearable system, and Jianzhi Robotics, which has raised hundreds of millions of yuan from Ant Group and Didi for its Gen DAS capture hardware.
That crowding is itself evidence of demand. Humanoid robot manufacturers in China and elsewhere are shipping hardware faster than they can generate the training data needed to make that hardware useful, and wearable capture — strap sensors on a person and send them outside — is emerging as the cheapest way to close that gap compared to building elaborate motion-capture stages.
What comes next?
Guangdong Zhishen Data's near-term task is proving its leg-focused approach can scale beyond a narrow use case into the multi-part, full-body capture that competitors like Mifeng already offer. With four of its investor's robot-maker clients watching, the company has a fast feedback loop to find out whether real-world leg data actually improves walking performance — and whether that's enough to stand out in an increasingly packed field chasing the same underlying shortage.
