Reproduce “MAESTRO: Orchestrating Robotics Modules with Coding Agents for Zero-Shot Generalist Robots” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Junyao Shi, Rujia Yang, Kaitian Chao, Selina Bingqing Wan, Yifei Shao, Jiahui Lei, Jianing Qian, Long Le and 4 more
The paper introduces MAESTRO, which uses a vision-language model to write code combining perception, geometry, grasping, and control tools. An orchestration harness helps the robot check feedback and continue or revise its code. On “Cut banana with knife,” MAESTRO reaches mean task progress of 92.0/100 versus 14.0/100 for π₀.₅, a human-scored partial-credit measure on this task.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.