The same tech long used to animate human movement is now scaling to power armies of humanoid robots.
The popular Chinese motion capture actress, Xixiyu Sakana (曦曦鱼SAKANA), recently controlled six Unitree G1 humanoid robots with her body. The footage caught the attention of MC Hammer because robots are training to move so well, they apparently CAN touch this.

Xixiyu Sakana controls six Unitree humanoid robots with her body (Source: CHINGMU/AMD)
Her high-tech suit is covered in markers for cameras to track her movements in real time. As she moves, the system turns her body motion into digital data, which AI converts into movement commands for the Unitree G1 humanoids to follow.
The performer demonstrated the capabilities during a live event in Shanghai, held by CHINGMU and the US chipmaker AMD. Shanghai-based CHINGMU developed the motion-capture gear and core algorithms while AMD served as the lead computing partner.

It was part of a larger showcase that captured motion data from 100 performers at once. Their movements were converted into a large motion-data library to teach robots to move in more natural, humanlike ways. According to a press release, the system performed real-time 3D reconstruction, market identification, and skeleton solving with an end-to-end latency under 12 milliseconds. The setup included 76 CHINGMU optical cameras that tracked about 5,300 markers across the 100 bodies.
The companies said it was the first publicly verified 100-person real-time motion-capture system. Representatives from the Shanghai New Hongqiao Notary Public Office officially witnessed and documented the achievement, confirming the authenticity of the technical data it produced.
The movements usually cannot be copied directly, since a robot’s joints, weight, balance, and limb proportions are different from a human’s. The MoCap data is commonly used as a reference for imitation learning, teleoperation, and training in simulation.

Sakana’s robot dancing routine was almost certainly built around what the humanoids could safely handle. The performance still showed impressive real-time control, even if it was executed in a carefully choreographed movement range.