Qwen把自然语言当遥控器，一个模型搞定机器臂、自动驾驶、无人机等20多种动作，还赢了多个基准，挺有意思。
Qwen-RobotWorld由阿里巴巴Qwen团队发布，将自然语言视为通用动作接口，统一末端执行器姿态、转向命令和导航点。该模型在具身世界知识语料库（860万视频-文本对、超2亿帧）上联合训练20+种具身类型和500+个动作类别。在EWMBench、DreamGen、WorldModelBench、PBench等基准测试中表现强劲。这一方法桥接了通用视频生成模型与领域专用具身模型之间的鸿沟。
By treating natural language as a universal action interface，Qwen-RobotWorld bridges the gap betwee...
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 🔄 1 ❤️ 37 👀 6007 📊 5 ⚡