Archon Robotics Releases Anchor 1 for Precise Humanoid Control
刚刚，源策未来发布 Anchor 1：让人形机器人真正做到精准控制
Archon Robotics released Anchor 1, a whole-body control model designed to enable precise and stable manipulation for humanoid robots. The system utilizes five anchor points (pelvis, wrists, ankles) to accept sparse instructions without online redirection and decouples error feedback from motion tokens to correct deviations in real-time. In real-world tests on Unitree G1 robots, Anchor 1 demonstrated lower wrist and ankle position errors compared to prior methods, improving teleoperation success rates for tasks such as kicking boxes and picking up slippers. The model serves as a foundational 'cerebellum' layer within Archon's embodied AI architecture.
Source: Chinese robotics — Hardware and sensing · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported success rate
37.5%
Reported trials
Reported success rate
98%
Reported trials
successful demonstrations per hour
9
successful demonstrations per hour
43
operator heights
165
View original evidence
四名身高从 165 厘米到 188 厘米、体重从 50 公斤到 80 公斤不等的操作员，先后用动作捕捉设备驱动同一台机器人：弯腰触地、单膝跪地、侧身去摸脚踝、原地小步横移。
Open source S8operator weights
50
View original evidence
四名身高从 165 厘米到 188 厘米、体重从 50 公斤到 80 公斤不等的操作员，先后用动作捕捉设备驱动同一台机器人：弯腰触地、单膝跪地、侧身去摸脚踝、原地小步横移。
Open source S8
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
用它做遥操作，踢纸箱任务的示范采集成功率从 37.5% 提升到 98.0%，每小时采到的成功示范从 9 条增加到 43 条。
Open source S17
几组实验数据展示了 Anchor 1 在精度与稳定性上的提升：在 72 段全身操作动作的真机测试中，它的手腕和脚踝全局位置误差，比此前表现最好的对比方法分别低约 28% 和 39%。
Open source S16
四名身高从 165 厘米到 188 厘米、体重从 50 公斤到 80 公斤不等的操作员，先后用动作捕捉设备驱动同一台机器人：弯腰触地、单膝跪地、侧身去摸脚踝、原地小步横移。
Open source S8
Source：Chinese robotics — Hardware and sensing · finance.sina.com.cn