LARG, Language-based Automatic Reward and Goal Generation

TL;DR

LARG leverages large language models to automatically generate reward and goal functions from textual task descriptions, enabling scalable training of robotic manipulation policies without handcrafted rewards.

Contribution

The paper introduces LARG, a novel method that converts text descriptions into reward and goal functions using LLMs, reducing manual effort in reward design for robot learning.

Findings

Successfully trained robotic manipulation policies using LARG-generated rewards.

Demonstrated scalability and effectiveness in diverse manipulation tasks.

Eliminated need for handcrafted reward functions in experiments.

Abstract

Goal-conditioned and Multi-Task Reinforcement Learning (GCRL and MTRL) address numerous problems related to robot learning, including locomotion, navigation, and manipulation scenarios. Recent works focusing on language-defined robotic manipulation tasks have led to the tedious production of massive human annotations to create dataset of textual descriptions associated with trajectories. To leverage reinforcement learning with text-based task descriptions, we need to produce reward functions associated with individual tasks in a scalable manner. In this paper, we leverage recent capabilities of Large Language Models (LLMs) and introduce \larg, Language-based Automatic Reward and Goal Generation, an approach that converts a text-based task description into its corresponding reward and goal-generation functions We evaluate our approach for robotic manipulation and demonstrate its ability to train and execute policies in a scalable manner, without the need for handcrafted reward functions.