X Square Robot公布了具身智能的进展，展示了其WALL-B系统，该系统在包裹分拣方面比Figure 03的性能提高了45%，同时成本降低了70%。此外，该公司开源了HOST，这是一个使人形机器人能够从简短的视频演示中学习新技能的框架，学习时间最短可达29秒，成功率达到62%。这种方法将具身智能训练从漫长的离线过程转变为快速、即时的模仿。 AI
影响 通过实现快速技能获取和降低硬件成本，加速了具身智能的发展。
排序理由 一家AI机器人公司发布了重要的产品和性能公告。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
In a one-hour livestream, X Square Robot's dual-arm system with standard grippers sorted 1,816 parcels, a short-burst throughput 45% above Figure 03's 200-hour benchmark of 1,248 parcels per hour, at roughly 70% lower hardware cost. The WALL-B world unified model carries the comp…
X Square Robot has open-sourced HOST, an inference-time learning framework that lets a humanoid robot watch a 29-second human demonstration and reproduce the skill at 62 percent success. The approach flips the embodied-AI recipe from offline fine-tuning to on-the-fly imitation.
X Square Robot has open-sourced HOST, a framework allowing humanoid robots to learn skills from a 29-second video and reproduce them on the first attempt with 62 percent success. The approach replaces hours of offline fine-tuning with inference-time learning, potentially transfor…
<table> <tr><td> <a href="https://www.reddit.com/r/singularity/comments/1vp7wcm/x_square_robot_demonstrates_embodied_ai_in/"> <img alt="X Square Robot Demonstrates Embodied AI in Real-World Logistics Operations" src="https://external-preview.redd.it/ogY9gpWlt87KHzfmG-Wi8SvgVRxiHI…