Daxiao Robotics Unveils Mundo, a New World Model for Robot Training
几张照片就能搭建虚拟训练场，全新机器人世界模型Mundo来了！
Unlike previous reconstruction methods, Mundo embeds physical parameters such as mass and friction to simulate realistic collisions and interactions like pushing chairs or opening drawers. This capability allows robots to practice navigation and manipulation tasks in simulated settings, reducing real-world trial-and-error costs while supporting long-duration exploration and multi-angle review of interaction details.
Source: Chinese robotics — Company news discovery · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S3
Reported numbers
photo captures
1–12
View original evidence
仅用几张照片，Mundo就能重建办公室三维空间，自由生成新视角画面，支撑机器人仿真训练 实际使用中，只需要1～12张实拍图片，Mundo就能够还原办公室、居家房间等室内空间。
Open source S7
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
简单来讲，这项新技术只需要少量照片或者短视频，就可以搭建出一套既能自由漫游又支持物理碰撞交互的三维虚拟环境，给机器人提供仿真训练场地，帮助机器人更好地适应现实生活中的各类场景。
Open source S3
仅用几张照片，Mundo就能重建办公室三维空间，自由生成新视角画面，支撑机器人仿真训练 实际使用中，只需要1～12张实拍图片，Mundo就能够还原办公室、居家房间等室内空间。
Open source S7
机器人能够在这个复刻出来的环境中练习导航移动、抓取摆放物品，系统还可以同步输出机器人第一视角、第三视角画面，完整留存运动轨迹与交互数据。
Open source S12
研发人员可以反复调试机器人策略，把大量训练放在虚拟环境完成，降低现实世界的试错成本。
Open source S13
模型既可以顺着设定的相机路线自由漫游，补全照片没有拍到的角落；面对更大空间，还可以生成数十秒甚至分钟级的长镜头探索画面。
Open source S8
以往不少三维重建技术只能做到“看得过去”，画面可以切换视角，但物体不会真实发生碰撞、推拉、掉落等物理反应。
Open source S4
Source：Chinese robotics — Company news discovery · finance.sina.com.cn