Reproduce “Model Predictive Control of Tensegrity Robots via Contact-Aware Graph Neural Dynamics Model” 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
Nelson Chen, Patrick Meng, Charles Tang, Angelina Degay, Zachary Brei, Rebecca Kramer-Bottiglio, Kostas E. Bekris, Mridul Aanjaneya
The authors develop a contact-aware graph neural network for a tensegrity robot and use it in model predictive path integral control. It encodes planar contacts and self-collisions, while a hybrid controller adds turns when sampling struggles. In MuJoCo, it reached 84% success on a held-out composite course versus 35% for MPPI alone, within the paper’s planar setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.