Qwen-Robot Suite: A Foundation Model Suite for Physical-World Intelligence
Original title:Qwen-Robot Suite: A Foundation Model Suite for Physical World Intelligence
AI summary
Qwen released three foundation models: Qwen-RobotNav, Qwen-RobotManip and Qwen-RobotWorld. Nav unifies five task types—instruction following, point/object-goal navigation, target tracking and autonomous driving—through a controllable observation protocol, achieving 76.5% SR on VLN-CE RxR, 75.6% SR on HM3Dv2 object-goal navigation (RGB only), a 90.0% tracking rate on EVT-Bench and 91.4 PDMS on NAVSIM. Manip uses a canonical state-action space to train cross-embodiment on more than 38,100 hours of heterogeneous open-source robot data. World co-trains more than 20 embodiments through a natural-language action interface, predicting the physical future of manipulation, driving and navigation. Together, the three translate general intelligence into physical action.
Key Takeaways
- Qwen released three foundation models: Qwen-RobotNav,Qwen-RobotManip and Qwen-RobotWorld.
- Nav unifies five task types—instruction following,point/object-goal navigation,target tracking and…
- 5% SR on VLN-CE RxR,75.
- 6% SR on HM3Dv2 object-goal navigation (RGB only),a 90.
- 0% tracking rate on EVT-Bench and 91.
Quick Facts
- Published
- Updated
- Source
- Qwen:Blog Retrieval(API)
- Category
- Model Releases
- AI score
- 73
Qwen把导航、操作和世界模型打包成机器人基础套件,用统一语言接口串起来,这是具身智能从单点突破走向系统化的信号,虽然离真实世界还有距离,但方向很清晰。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Qwen-Robot Suite: A Foundation Model Suite for Physical-World Intelligence. June 16, 2026. https://new.nat6.net/en/news/cmqg5kvq2006fslncudi17omu/