Gizmo: Turns Images into Sim Scenes in Minutes to Train Robots

eigenron · x · 2026-07-16

A developer demonstrated using the Gizmo tool to convert text into a robotic arm grasping scene within about 10 minutes and export it to a simulator. Subsequently, 500 sets of demonstration data were generated in a few hours, and an ACT policy was successfully trained. The tool aims to enable one-click conversion from real images/videos to simulation scenes.

Related event: Gizmo: Generate Robot Simulation Scenes from Text in Minutes(4 posts)→

Original post →

More from Embodied

Embodied channel →