← Back to Blog

Alibaba Launches Qwen-Robot Series: Deep Dive into the Manip + Nav + World Three-Model Matrix

2026-06-16 By Superdata RobotAI
Qwen-RobotVLAVLNWorld ModelEmbodied AIAlibabaQwenModel Release

On June 16, 2026, Alibaba officially launched the Qwen-Robot series — the first complete embodied AI model matrix from the Qwen family, covering VLA manipulation (Qwen-RobotManip), VLN navigation (Qwen-RobotNav), and world modeling (Qwen-RobotWorld).

Model Matrix

ModelTypeCore Capability
Qwen-RobotManipVLAManipulation & Grasping
Qwen-RobotNavVLNMobility & Pathfinding
Qwen-RobotWorldWorld ModelPhysics Prediction

Key Highlights

Qwen-RobotManip

  • Trained on 38,000+ hours of purely open-source data — no proprietary robot data used
  • Ranked #1 and #2 on RoboChallenge Table30 v1 real-robot benchmark (45% success rate)
  • Diffusion Action Head for smooth trajectory generation

Qwen-RobotNav

  • Unified framework for 5 navigation tasks: instruction following, object navigation, tracking, autonomous driving, multi-modal wayfinding
  • Native integration with LangChain, AutoGPT, MetaGPT

Qwen-RobotWorld

  • Predicts future visual/physical states for training data augmentation, action pre-simulation, and anomaly prediction
  • Works in tandem with Manip in a generate-evaluate-select loop

Industry Implications

  • Victory for open data: achieving SOTA with 100% open-source training data
  • Hardware-agnostic: no GPU or simulator lock-in unlike NVIDIA GR00T
  • First complete model matrix: Manip + Nav + World in a single family

References

AI 助手
输入需求,AI 帮你在 58 个数据集、19 个标准、18 个工具中智能匹配