Meta-Reinforcement Learning via Language Instructions
Authors:
- 29 May 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5985-5991
Abstract
Although deep reinforcement learning has recently been very successful at learning complex behaviors, it requires a tremendous amount of data to learn a task. One of the fundamental reasons causing this limitation lies in the nature of the trial-and-error learning paradigm of reinforcement learning, where the agent communicates with the environment and pro-gresses in the learning only relying on the reward signal. This is implicit and rather insufficient to learn a task well. On the con-trary, humans are usually taught new skills via natural language instructions. Utilizing language instructions for robotic motion control to improve the adaptability is a recently emerged topic and challenging. In this paper, we present a meta-RL algorithm that addresses the challenge of learning skills with language instructions in multiple manipulation tasks. On the one hand, our algorithm utilizes the language instructions to shape its in-terpretation of the task, on the other hand, it still learns to solve task in a trial-and-error process. We evaluate our algorithm on the robotic manipulation benchmark (Meta-World) and it significantly outperforms state-of-the-art methods in terms of training and testing task success rates. Codes are available at https://tumi6robot.wixsite.com/million.Keywords
All Related Versions
Funding Information
- Regional Development and Energy (StMWi) (DIK0249)
This publication has 16 references indexed in Scilit:
- An Accurate GRU-Based Power Time-Series Prediction Approach With Selective State Updating and Stochastic OptimizationIEEE Transactions on Cybernetics, 2021
- Concept2Robot: Learning manipulation concepts from instructions and human demonstrationsThe International Journal of Robotics Research, 2021
- Complex Robotic Manipulation via Graph-Based Hindsight Goal GenerationIEEE Transactions on Neural Networks and Learning Systems, 2021
- Energy-efficient and damage-recovery slithering gait design for a snake-like robot based on reinforcement learning and inverse reinforcement learningNeural Networks, 2020
- Grandmaster level in StarCraft II using multi-agent reinforcement learningNature, 2019
- Reinforcement Learning, Fast and SlowTrends in Cognitive Sciences, 2019
- End to End Learning of Spiking Neural Network Based on R-STDP for a Lane Keeping Vehicle2018
- Human-level control through deep reinforcement learningNature, 2015
- Glove: Global Vectors for Word Representation2014
- MuJoCo: A physics engine for model-based control2012