By treating natural language as a universal action interface，Qwen-RobotWorld bridges the gap betwee...
TL;DR · AI 摘要
通义千问推出Qwen-RobotWorld，通过自然语言作为统一动作接口，实现多种机器人动作的联合训练与高效控制。
核心要点
- Qwen-RobotWorld通过自然语言统一控制20+种机器人动作类型和500+动作类别。
- 模型基于8.6M视频-文本对和200M+帧的Embodied World Knowledge语料库进行训练。
- Qwen-RobotWorld在多个基准测试中表现优异，包括EWMBench、DreamGen等。
结构提纲
按章节快速跳转。
- §引言
介绍Qwen-RobotWorld的核心目标和应用场景。
- ·核心技术
通过自然语言作为统一动作接口，实现多种机器人动作的联合训练。
- ›训练数据
基于8.6M视频-文本对和200M+帧的Embodied World Knowledge语料库进行训练。
- ·性能表现
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- Qwen-RobotWorld
- 核心技术
- 自然语言作为统一动作接口
- 支持20+种机器人动作类型和500+动作类别
- 训练数据
- 8.6M视频-文本对
- 200M+帧的Embodied World Knowledge语料库
- 性能表现
- 在EWMBench、DreamGen等基准测试中表现优异
金句 / Highlights
值得收藏与分享的关键句。
通过自然语言作为统一动作接口，Qwen-RobotWorld实现了多种机器人动作的联合训练。
模型基于8.6M视频-文本对和200M+帧的Embodied World Knowledge语料库进行训练。
Qwen-RobotWorld在多个基准测试中表现优异，包括EWMBench、DreamGen等。
Qwen on X: "By treating natural language as a universal action interface，Qwen-RobotWorld bridges the gap between general video generation models and domain-specific embodied models — this converts end-effector poses, steering commands, and navigation waypoints into a single interface, https://t.co/bb54GFYvi3" / X
Qwen
@Alibaba_Qwen
Replying to
By treating natural language as a universal action interface，Qwen-RobotWorld bridges the gap between general video generation models and domain-specific embodied models — this converts end-effector poses, steering commands, and navigation waypoints into a single interface, enabling 20+ embodiment types and 500+ action categories to be co-trained under the Embodied World Knowledge corpus (8.6M video-text pairs, 200M+ frames), with each domain's physical knowledge reinforcing the others. Qwen-RobotWorld performs strongly across EWMBench/DreamGen/WorldModelBench/PBench benchmarks. Blog Link：
qwen.ai/blog?id=qwen-r…
1:07 PM · Jun 16, 2026
9.5K
Views
1
5
6
61
9
Read 1 reply