roboalign-language-action-reasoning

【效率与协作】Bridge the modality gap between language reasoning and low-level robot actions through two-stage alignment training. Use RL with action-accuracy rewards to ground linguistic reasoning in precise motor control, achieving 17.5%-106.6% improvements over SFT baselines.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准