According to TechFlowPost, Alibaba (09988.HK) announced the Qwen-Robot embodied intelligence model series on June 16. The release includes three models: the VLA manipulation model Qwen-RobotManip, the VLN navigation model Qwen-RobotNav, and the world model Qwen-RobotWorld.
Three Models for Robotic Hands, Feet and Brain
The announcement describes the three models as corresponding to different core capabilities for robots. Qwen-RobotManip is positioned around manipulation, giving robots more capable “hands”; Qwen-RobotNav is focused on navigation and movement, described as helping robots find their way; Qwen-RobotWorld is presented as the world model component, compared with a brain that can think.
This is the first complete embodied intelligence model series within the Qwen large model family. The three models can be deployed separately or operate together, providing what the announcement calls a reliable “general foundation” for different forms of robots to move toward real-world deployment. The item was credited to Jinshi and carried by TechFlowPost under its technology updates category.