Skip to content
View original post on X: 👩‍💻 Paige Bailey
Published
35/100

Research team uses NanoBanana image generation to plan rope manipulation

AIA research team uses NanoBanana, an image generation model, to synthesize goal images for deformable object manipulation from a scene and a text prompt, such as passing a rope through a red slot. The method is presented as a visual imagination step for planning, and the paper is listed at Spotlight S5 and Poster S2 #45.
👩‍💻 Paige BaileyVerified on X
@DynamicWebPaige
View on X

Part of a thread · earlier post

Using image generation as sort of a visual imagination for planning! 🍌

Given a scene + prompt (ex: "pass rope through red slot"), this team uses @NanoBanana to synthesize the goal image for deformable object manipulation.

📍 Spotlight S5 | Poster S2 #45
🔗 https://openreview.net/forum?id=3hsq75Gmr9

Source details

👩‍💻 Paige Bailey · x.comPublished

Original title: Using image generation as sort of a visual imagination for planning! 🍌

Topics

Image generationEmbodied AIPapers & research