LGR2: Language Guided Reward Relabeling for Accelerating Hierarchical Reinforcement Learning

• Published in: arXiv.org
• Main Authors: Singh, Utsav, Bhattacharyya, Pramit, Namboodiri, Vinay P
• Format: Paper
• Language: English
• Published: Ithaca Cornell University Library, arXiv.org 16.06.2024
• Subjects: Cognition & reasoning; Control tasks; Empirical analysis; Interactive control; Interactive systems; Language; Large language models; Robot control; Robot learning; Robotics; Task complexity

Developing interactive systems that leverage natural language instructions to solve complex robotic control tasks has been a long-desired goal in the robotics community. Large Language Models (LLMs) have demonstrated exceptional abilities in handling complex tasks, including logical reasoning, in-context learning, and code generation. However, predicting low-level robotic actions using LLMs poses significant challenges. Additionally, the complexity of such tasks usually demands the acquisition of policies to execute diverse subtasks and combine them to attain the ultimate objective. Hierarchical Reinforcement Learning (HRL) is an elegant approach for solving such tasks, which provides the intuitive benefits of temporal abstraction and improved exploration. However, HRL faces the recurring issue of non-stationarity due to unstable lower primitive behaviour. In this work, we propose LGR2, a novel HRL framework that leverages language instructions to generate a stationary reward function for the higher-level policy. Since the language-guided reward is unaffected by the lower primitive behaviour, LGR2 mitigates non-stationarity and is thus an elegant method for leveraging language instructions to solve robotic control tasks. To analyze the efficacy of our approach, we perform empirical analysis and demonstrate that LGR2 effectively alleviates non-stationarity in HRL. Our approach attains success rates exceeding 70\(\%\) in challenging, sparse-reward robotic navigation and manipulation environments where the baselines fail to achieve any significant progress. Additionally, we conduct real-world robotic manipulation experiments and demonstrate that CRISP shows impressive generalization in real-world scenarios.