Roboalign Language Action Reasoning

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.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit 998b5718d2

Frequently asked questions

npx get-skill add adu2021/roboalign-language-action-reasoning