Hangzhou, China – November 01, 2026 – World Physical AI 2026 & Unitree Partner Ecosystem Conference took place on October 22 at Phase II of the Hangzhou International Expo Center. Under the theme “Decade Forward, Grow Together,” the event brought together more than 3,000 embodied AI experts, representatives of leading companies, and ecosystem partners from around the world to present advances in physical AI and industry collaboration.
https://www.youtube.com/watch?v=2aMcKpHu63E
CHINGMU took part in several robot sports demonstrations at the conference. Working with Unitree Robotics and EndlessAI, the company supported table tennis, tennis, and basketball demonstrations with high-precision optical motion capture for tracking fast-moving objects and measuring their positions in 3D space. CHINGMU’s motion capture systems also recorded human movements used to support robot learning and training for the humanoid-versus-human go-kart race and the robot combat demonstration.
From tracking fast-moving balls to capturing a wide range of human movements, CHINGMU provides motion data and technical support for robot learning, motion analysis, and related research and development.
01 High Precision Motion Tracking for Robot Table Tennis and Tennis
On the morning of the opening day, Chinese basketball legend Yao Ming and Unitree founder Wang Xingxing were among the visitors to the joint CHINGMU and EndlessAI booth. They watched live demonstrations of robots playing tennis, table tennis, and basketball, making the exhibit one of the most popular at the conference.
Fast table tennis rallies and repeated tennis returns may look straightforward, but they pose interconnected challenges for robot perception, motion prediction, real-time response, and motion control.
In the joint demonstrations, CHINGMU’s motion capture system measured the 3D positions of the balls and robots in real time, continuously tracking the trajectories of fast-moving balls. These external measurements supplied information about the physical environment to the models and algorithms responsible for predicting motion, deciding when and how to strike the ball, and generating movements. The robot control systems then executed those movements. CHINGMU’s spatial measurements and motion tracking data were a key input to this process.
For EndlessAI, competitive sports provide a way to continually refine humanoid robots’ whole-body coordination across perception, decision-making, and execution. In an interview, Chen Guang, founder of EndlessAI, described CHINGMU’s two key contributions: “First, establishing a world coordinate system for the court; second, tracking the ball’s trajectory in real time.”
By establishing a shared spatial reference frame and continuously capturing target motion, high-precision motion capture provides a quantitative foundation for developing, debugging, and validating models and control algorithms.
CHINGMU also supported motion data collection for the go-kart racing and robot combat demonstrations. Capturing these complex movements provides data for robots to learn new motions and develop their physical capabilities.
The humanoid-versus-human go-kart race was presented by CHINGMU partner Symbiosis Robotics. Driving a go-kart requires a robot to coordinate its body movements with vehicle control, supported by a workflow that connects data collection, model training, and deployment. CHINGMU contributed motion data collection and performance evaluation, providing evidence for analyzing interactive movements, training models, and refining the approach.
Across high-speed ball tracking, racing, and combat, CHINGMU applies precise spatial measurement and motion data services to specific tasks, supporting robot motion learning, performance evaluation, and system improvement.
02 CHINGMU Joins the First Group of Companies in the Unitree Startup and Innovation Program
The Unitree Startup & Innovation Program was launched globally at the conference’s main forum. The program is aimed at entrepreneurs working to bring embodied AI into commercial use, with a focus on industry solutions, developer tools and ecosystem software, and data and simulation platforms. It is expected to generate several billion yuan in collaborative business activity over the next five years and aims to incubate more than 200 promising companies within the ecosystem.
CHINGMU founder and CEO Zhang Haiwei attended the launch ceremony as a representative of one of the program’s first participating companies. CHINGMU’s inclusion in Unitree’s new initiative for ecosystem development and startup investment reflects partners’ recognition of its technical innovation and ability to bring embodied AI technologies into practical use.
03 Motion Data as Infrastructure for Physical AI
At the embodied AI developers’ forum, Zhang Haiwei delivered a keynote titled “Opening the Physical World to AI: Infrastructure for the Physical AI Era.” He explained that, as AI moves from digital environments into the physical world, high-quality motion data is becoming a key factor in advancing robot intelligence. Just as data underpins the internet, precise, multimodal, reusable motion data is essential infrastructure for physical AI.
Robots operating in real environments must perceive spatial relationships, learn complex movements, and test models and control strategies through physical execution. Real-world motion data provides an essential basis for each of these tasks.
Building on high-precision optical motion capture, CHINGMU is expanding its capabilities in multimodal data acquisition, motion retargeting, teleoperation, and motion performance evaluation. Together, these capabilities turn real-world movement into data resources for robotics research and development.
The aim is to capture real movement accurately, make motion data useful in practice, and provide objective measurements of robot motion performance.
By turning measured motion into data that supports robot intelligence, CHINGMU is exploring further applications of high-precision motion data technology in physical AI.
04 Expanding Collaboration with Unitree from Data Collection to Real World Applications
The joint demonstrations at the conference were the latest result of CHINGMU’s ongoing collaboration with Unitree Robotics.
The two companies have previously worked together on capturing professional athletes’ movements, interactive robot motion, teleoperation, and 3D motion measurement, applying high-precision motion capture across research and demonstration settings.
Capturing Professional Athletes’ Movements
In projects associated with the CMG World Robot Competition – Mecha Fighting Series and the UFC Performance Institute, CHINGMU used optical motion capture to record professional combat athletes performing boxing, Muay Thai, and jiu-jitsu techniques. The resulting motion data supports robot learning and research into robot movement.
Interactive Robot Sports
For robot table tennis demonstrations at the 2026 World Robot Conference (WRC) and AI & Digital Bridge 2026 in Kazakhstan, CHINGMU combined high-speed ball tracking with 3D robot position tracking to support Unitree robots’ dynamic interactions.
Teleoperation and Motion Measurement
In robot demonstrations and research projects, CHINGMU’s motion capture technology has also been used to reconstruct human body poses, map movements to robots, and measure robot motion in 3D space, providing data for technical exploration and validation.
These collaborations, spanning human motion data collection, dynamic robot interaction, and motion measurement, demonstrate the range of applications for high-precision motion capture in embodied AI research and development.
05 Working with Partners to Expand Physical AI Applications
Progress in physical AI depends on continued collaboration across robot hardware, algorithms, data, and real-world applications. CHINGMU will continue to improve the precision, real-time performance, and usability of motion data. Working with Unitree Robotics, EndlessAI, Symbiosis Robotics, and other industry partners, the company aims to expand the use of high-precision motion capture and motion data in robotics research, interactive motion, and application validation.
CHINGMU looks forward to working with its partners to capture real-world movements accurately and put the resulting data to effective use. By connecting perception, data, and intelligence, the company aims to make motion data a reliable foundation for robots to understand the physical world, learn complex movements, and validate control strategies.
The real world is an essential training ground for physical AI, and high-quality motion data connects that world with machine intelligence.
About CHINGMU
CHINGMU is a high-tech enterprise specializing in optical motion capture, with full-stack in-house development across algorithms, hardware, and software. Centered on computer vision and AI, CHINGMU has built an integrated technology system for 3D intelligent perception and human-computer interaction, with its core technologies developed independently. Its business covers robotics, engineering, life sciences, virtual reality, entertainment, and education.
Media ContactCompany Name: SHANGHAI CHINGMU VISION TECHNOLOGY CO., LTDContact Person: CHINGMUEmail: Send EmailPhone: 13310099680City: ShanghaiCountry: ChinaWebsite: https://en.chingmu.com/
