In-Context Learning Enables Robot Action Prediction in LLMs

TL;DR

This paper introduces RoboPrompt, a framework that leverages in-context learning in large language models to predict robot actions directly from textual descriptions without additional training, demonstrating improved performance in both simulated and real-world tasks.

Contribution

The paper presents RoboPrompt, a novel method that enables off-the-shelf LLMs to predict robot actions through ICL, bypassing the need for training on specific robot data.

Findings

RoboPrompt outperforms zero-shot baselines in robot action prediction.

The approach is effective in both simulated and real-world environments.

Structured ICL demonstrations improve prediction accuracy.

Abstract

Recently, Large Language Models (LLMs) have achieved remarkable success using in-context learning (ICL) in the language domain. However, leveraging the ICL capabilities within LLMs to directly predict robot actions remains largely unexplored. In this paper, we introduce RoboPrompt, a framework that enables off-the-shelf text-only LLMs to directly predict robot actions through ICL without training. Our approach first heuristically identifies keyframes that capture important moments from an episode. Next, we extract end-effector actions from these keyframes as well as the estimated initial object poses, and both are converted into textual descriptions. Finally, we construct a structured template to form ICL demonstrations from these textual descriptions and a task instruction. This enables an LLM to directly predict robot actions at test time. Through extensive experiments and analysis, RoboPrompt shows stronger performance over zero-shot and ICL baselines in simulated and real-world settings. Our project page is available at https://davidyyd.github.io/roboprompt.