Saved in:
Bibliographic Details
Main Authors: Baert, Mattijs, Leroux, Sam, Simoens, Pieter
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.10096
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909426935398400
author Baert, Mattijs
Leroux, Sam
Simoens, Pieter
author_facet Baert, Mattijs
Leroux, Sam
Simoens, Pieter
contents Learning from Demonstrations (LfD) and Reinforcement Learning (RL) have enabled robot agents to accomplish complex tasks. Reward Machines (RMs) enhance RL's capability to train policies over extended time horizons by structuring high-level task information. In this work, we introduce a novel LfD approach for learning RMs directly from visual demonstrations of robotic manipulation tasks. Unlike previous methods, our approach requires no predefined propositions or prior knowledge of the underlying sparse reward signals. Instead, it jointly learns the RM structure and identifies key high-level events that drive transitions between RM states. We validate our method on vision-based manipulation tasks, showing that the inferred RM accurately captures task structure and enables an RL agent to effectively learn an optimal policy.
format Preprint
id arxiv_https___arxiv_org_abs_2412_10096
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Reward Machine Inference for Robotic Manipulation
Baert, Mattijs
Leroux, Sam
Simoens, Pieter
Robotics
Machine Learning
Learning from Demonstrations (LfD) and Reinforcement Learning (RL) have enabled robot agents to accomplish complex tasks. Reward Machines (RMs) enhance RL's capability to train policies over extended time horizons by structuring high-level task information. In this work, we introduce a novel LfD approach for learning RMs directly from visual demonstrations of robotic manipulation tasks. Unlike previous methods, our approach requires no predefined propositions or prior knowledge of the underlying sparse reward signals. Instead, it jointly learns the RM structure and identifies key high-level events that drive transitions between RM states. We validate our method on vision-based manipulation tasks, showing that the inferred RM accurately captures task structure and enables an RL agent to effectively learn an optimal policy.
title Reward Machine Inference for Robotic Manipulation
topic Robotics
Machine Learning
url https://arxiv.org/abs/2412.10096