Reproduce “MAGAR: Multimodal Agents for GUI-Based 3D Mechanical Assembly and Minimal Rework” 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
Jingxuan Wei, Jiahao Fu, Fuqiang Wang, Xiaoming Wu, Yuwen Liu, Kaiwen Wei, Ming Yang, Junnan Zhu
MAGAR improves predictions for three tasks in Unity-based 3D assembly by proposing task-specific corrections to a supervised-fine-tuned policy and retaining its original prediction when a selector’s thresholds are not met. On recorded states from six held-out assembly families, its task-balanced Overall score is 41.39, versus 37.50 for Vanilla-SFT. Evaluation does not include long-horizon online assembly.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.