LegoGPT, developed by Carnegie Mellon University researchers, enables users to create LEGO designs by inputting text prompts. The AI generates physically stable models that can be built with existing bricks or custom ones. It utilizes a dataset called StableText2Lego, comprising over 47,000 LEGO structures, associating each with descriptive captions. The model predicts brick placement and checks for validity, ensuring each brick fits correctly. Additional features include assigning colors and surface textures, making it versatile for both human assembly and robotic construction.
The AI generates stable LEGO designs based on text inputs, allowing users to construct custom models with existing or new bricks through an intuitive design plan.
Training on over 47,000 LEGO structures, LegoGPT can predict brick placement effectively, ensuring physical stability and offering features like color and texture assignment.
Collection
[
|
...
]