28 June 2024

Text2Robot:

Evolutionary Robot Design from Text Descriptions

Preprint

Ryan Ringel
Ryan Ringel Duke University
Zachary Charlick
Zachary Charlick Duke University zacharycharlick.com
Jiaxun Liu
Jiaxun Liu Duke University www.jiaxunliu.com
Boxi Xia
Boxi Xia Duke University
Boyuan Chen
Boyuan Chen Duke University boyuanchen.com

Overview

Robot design has traditionally been costly and labor-intensive. Despite advancements in automated processes, it remains challenging to navigate a vast design space while producing physically manufacturable robots. We introduce Text2Robot, a framework that converts user text specifications and performance preferences into physical quadrupedal robots. Within minutes, Text2Robot can use text-to-3D models to provide strong initializations of diverse morphologies. Within a day, our geometric processing algorithms and body-control co-optimization produce a walking robot by explicitly considering real-world electronics and manufacturability. Text2Robot enables rapid prototyping and opens new opportunities for robot design with generative models.

Video (Click to YouTube)

Video Figure

Paper

Check out our paper linked here.

Codebase

Check out our codebase at https://github.com/generalroboticslab/Text2Robot

Citation

@article{ringel2024text2robot,
      title={Text2Robot: Evolutionary Robot Design from Text Descriptions}, 
      author={Ryan P. Ringel and Zachary S. Charlick and Jiaxun Liu and Boxi Xia and Boyuan Chen},
      year={2024},
      eprint={2406.19963},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2406.19963}, 
}         

Acknowledgment

This work is supported by DARPA FoundSci program under award HR00112490372, by ARL STRONG program under awards W911NF2320182 and W911NF2220113.

Contact

If you have any questions, please feel free to contact Ryan Ringel, Zachary Charlick, or Jiaxun Liu. These three authors contributed equally to this work.

Categories

Computational Robot Design Robot Learning