Reproduce “TEXEDO : Test Time Scaling for Controller-aware Language-conditioned Humanoid Motion Generation” 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

Jianuo Cao, Yuxin Chen, Yuzhen Song, Masayoshi Tomizuka, Chenran Li, Thomas Tian

TEXEDO selects among motions sampled by a frozen language-conditioned generator, using a learned verifier trained on rollouts of a humanoid controller to filter predicted infeasible candidates before ranking survivors by language alignment. In simulation, with 32 candidates, it achieves 0.984 tracking success and a 6.054 alignment score, versus 0.873 and 5.722 for one unselected sample; feasibility is predicted, not certified.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.