Xiaomi reveals human-like bionic hand for ‘CyberOne’ robot: Here’s how it works
The Xiaomi team unveiled a redesigned bionic hand for its CyberOne robot, shrinking it by 60% to human-size while adding full-palm touch sensors and ‘sweat glands’
The transformation of Xiaomi's CyberOne from a shaky prototype into a high-performance industrial humanoid is centered on a unique biological inspiration. To prevent high-performance 100W motors from overheating in a compact space, Xiaomi developed a liquid cooling system.
Utilizing 3D-printed metal channels, the hand evaporates 0.5 ml of water per minute to provide 100W of active cooling. This “sweat” allows the robot to perform high-load, gruelling tasks for hours without thermal throttling or performance loss.
The hand assembly is 60% smaller, achieving a 1:1 scale with a human hand, which is crucial for training the AI. It now features an 83% increase in active Degrees of Freedom (DOF), closely mimicking the 22-27 DOFs of a biological human hand. While many prototypes fail at 10,000 cycles, CyberOne’s hardware has endured over 150,000 grasping cycles.
The robot is covered in 8,200 $mm 2$ of tactile sensors, allowing it to feel its environment and work effectively even when its line of sight is blocked. In automotive trials, it maintained a 90.2% success rate in fastening nuts over a three-hour period, proving it is ready for the factory floor. Xiaomi has open-sourced its TacRefineNet framework and released 61 hours of tactile data to help the global robotics community improve sim-to-real transitions.
-
Anthropic, Gates Foundation collaborates to expand AI partnership in health education sector
-
OpenAI reviews antitrust action against Apple; Claims report
-
Anthropic overtakes OpenAI in business AI adoption
-
Tencent, Alibaba turn to local AI chips as Nvidia uncertainty grows
-
Microsoft faces UK antitrust probe over business software practices
-
Google unveils Googlebook: Here’s everything you need to know
-
Halupedia explained: Why AI Wikipedia clone is raising red flags
-
Who shapes AI’s answers? Ex-Meta news chief raises concerns
