Reproduce “FreeSpeed: Training-Free Speed Control for Generative Robot Policies” 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
Yuxuan Hu, Shilin Shan, Qiheng Wang, Jinghan Yang, Junqiao Fan, Hao Wan, Jianfei Yang
FreeSpeed adjusts action chunks from a frozen robot policy at test time, using changes in consecutive motion directions to adapt step sizes. In real-robot tests, success averaged 94.0% across six non-default commands, versus 93.8% for the unchanged policy. Realized rates ranged from 0.38× to 1.97×; requested rates were not always met, particularly during contact-rich phases.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.