Gasgoo Munich- At 2026 World Robot Conference kicking off on August 19, CasiaSilicon Robotics, an incubatee of the Institute of Automation, Chinese Academy of Sciences, is marking its fourth consecutive appearance. The company brought its CasiaHand dextrous hand series, embodied "large and small brain" models, the "Wuji" embodied intelligent industrial robot, and data acquisition products to the show floor.
Image Source: CasiaHand Robotics
In the dextrous hand exhibition area, CasiaHand Robotics made a collective debut of six products spanning its M series, X series, and the industrial-grade, three-finger G series. The M6 features an anthropomorphic, lightweight design with 6 degrees of freedom, multimodal perception, and a 10-kilogram payload. The M11 combines worm gears with cable-driven transmission, enabling hybrid rigid-soft control. Meanwhile, the X12—built for industrial scenarios—uses a full direct-drive architecture to deliver 12 degrees of freedom and a 30-kilogram payload.
A standout feature was the world's first 24-hour livestream of an industry-grade dextrous hand in continuous operation. On the M6 cyclic grasping platform, the unit autonomously picked up objects of varying textures—milk cartons, medicine boxes, and cosmetics—while maintaining a stable operating temperature around 30°C. Powered by proprietary perception algorithms and motion control, the system adapts its grip to target characteristics, executing over 1,000 grasps per hour.
Image Source: CasiaHand Robotics
On the algorithmic front, the CasiaHand Brain-Si 0.5 model employs a layered architecture: a "large brain" for environmental understanding and task planning, and a "small brain" for motion control and joint coordination. The universal grasping platform requires no pre-set object positioning, instead autonomously closing the loop on recognition, positioning, decision-making, and execution.
The "Wuji" embodied intelligent industrial robot demonstrated loading and unloading tasks on a production line. It autonomously handled operations such as opening machine doors and swapping trays—all without requiring extensive modifications to existing infrastructure. Meanwhile, a data glove interaction zone allowed visitors to experience "human-machine mapping" in real time.
Incubated by the National Key Laboratory of Multimodal Artificial Intelligence Systems at the CAS Institute of Automation, CasiaSilicon has achieved 100% domestic sourcing for core components. For this conference, the company joined forces with supply chain partners—including Panxi, Rui'erman, Luoshi, and GigaDevice—to showcase their wares and build an open ecosystem for embodied intelligence. CEO Zhang Tianyi noted that industry competition is shifting from pure hardware performance to a combined "hardware plus intelligence" capability.
