“Code as Policy” (CaP) is a research framework where language model-generated programs (LMPs) represent robot policies, allowing language models to directly write executable code for robot control. It translates natural language instructions into code that can handle spatial-geometric reasoning, generalize to new tasks, and define robot behaviors like picking, placing, and trajectory control.
Code as Policies: Language Model Programs for Embodied Control
This paper presents code as policies: a robot-centric formalization of language model generated programs (LMPs) that can represent reactive policies… read more
Code as Policies: Language Model Programs for Embodied Control
by J Liang · 2022 · Cited by 1707 — This paper presents code as policies: a robot-centric formulation of language model generated programs (LMPs) that can represent… read more
Code as Policies is a framework that leverages large language models (LLMs) to generate programs acting as policies for robotic control.
