Excited to share our new work! #CoRL2024 RoVi-Aug: Robot and Viewpoint Augmentation for Cross-Embodiment Robot Learning RoVi-Aug can zero-shot deploy on a different robot w/ significantly different camera angles and achieve robot-skill cross-product✨ 🔗rovi-aug.github.io

Please also check out the amazing concurrent work VISTA led by Stephen Tian also leveraging zero-shot novel view synthesis techniques for robot learning 🥳 Glad to see corroborative findings!

Thanks TechXplore for writing a really nice article about our RoVi-Aug work (rovi-aug.github.io)! Check it out! 👀

Excited to kick off the X-Embodiment workshop Conference on Robot Learning in the morning (9am @ room Terra)! We have an exciting lineup of speakers, including Shuran Song, Kevin Black, Ryan Julian, Kiana Ehsani, Yuke Zhu, Edward Johns, Sergey Levine

Everything you love about generative models — now powered by real physics! Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics

Introducing Phantom 👻: a method to train robot policies without collecting any robot data — using only human video demonstrations. Phantom turns human videos into "robot" demonstrations, making it significantly easier to scale up and diversify robotics data. 🧵1/9

How to use simulation data for real-world robot manipulation? We present sim-and-real co-training, a simple recipe for manipulation. We demonstrate that sim data can significantly enhance real-world performance, even with notable differences between the sim and the real. (1/n)

Co-training with both real and sim data; great effort to squeeze as much juice out of different sources of data so the whole is larger than the sum of parts! Great having Soroush Nasiriany Zhenyu Jiang Abhi Maddukuri Lawrence Yunliang Chen on the pod with me & Chris Paxton !

Super fun chatting with Chris Paxton and Michael Cho - Rbt/Acc about our recent study on Sim-and-Real Co-Training: A Simple Recipe for Vision-Based Robotic Manipulation co-training.github.io, also to be presented at RSS next month Robotics: Science and Systems Check it out!