Understand this story
On October 27, 2026, Xingyuan Intelligence released "Chorus Zhongxing," a cross-embodiment data engine for robot pre-training. According to a Leikeji report, the engine can convert human manipulation videos captured by smart glasses, chest-mounted cameras, wrist cameras, UMI devices, smart headbands, and other devices into pre-training data adapted to different robot embodiments. Tests show that Chorus, through adaptation to 16 robot embodiments, reduces the processing cost of effective robot trajectories to 15 yuan per hour, raises the average usable data conversion rate to 85%, and, combined with 3x instruction augmentation, can achieve approximately 40x expansion of effective training data, with an equivalent cost of 5 yuan/hour per unit of training sample. A TMTPost·AGI report on the same day mentioned that Yang Shiqi of Xingyuan Intelligence said the team insists on not building hardware embodiments, and instead uses the System Scaling path to connect data, models, and real-world deployment; under the current test configuration, 1 hour of human video can generate approximately 13.6 hours of cross-embodiment motion trajectories. No further progress has been disclosed at present.
AI-generated from public reports
Developments
1 confirmed developments
Xingyuan Intelligence releases cross-embodiment data engine Chorus Zhongxing
Source:English text is being prepared.
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 27
TMTPost · AGI
Xingyuan Intelligence's Yang Shiqi: We Don't Build Robot Bodies, We Bet on Robot "Tokens" and the Cross-Embodiment Data Engine Chorus
Xingyuan Intelligence will launch the cross-embodiment data engine "Chorus Zhongxing" on October 27, which cleans and 3D-reconstructs human operation videos and then adapts them to multiple robot embodiments. Under the current test configuration, Chorus achieves an effective data conversion rate of about 85%; 1 hour of human video can generate about 13.6 hours of cross-embodiment action trajectories, and combined with 3x instruction augmentation, it achieves about 40x training data expansion. The team insists on not building hardware bodies, and uses the System Scaling path to connect data, models, and real-world deployment.
Leitech
Xingyuan Intelligence Releases Cross-Embodiment Data Engine Chorus, Claiming Approximately 40x Training Data Expansion
Xingyuan Intelligence has released Chorus Zhongxing, a cross-embodiment data engine for robot pretraining, which can convert human manipulation videos captured by smart glasses, chest-worn cameras, wrist cameras, UMI devices, smart headbands, and other devices into pretraining data adapted to different robot embodiments. Tests show that Chorus, through adaptation to 16 robot embodiments, reduces the processing cost per hour of effective robot trajectories to 15 yuan, raises the average usable data conversion rate to 85%, and, combined with 3x instruction augmentation, can achieve approximately 40x effective training data expansion, with an equivalent cost per unit of training samples of 5 yuan/hour.