ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals

Imitation Learning (IL) is a promising approach for teaching tasks to robots by human demonstrations, although it faces challenges from long-horizon tasks and diverse goals in real-world settings. These issues stem from (i) a distribution mismatch between demonstrations and real-world execution and...

Full description

Saved in:
Bibliographic Details
Main Authors: Cynthia Ochoa, Hanbit Oh, Yuhwan Kwon, Yukiyasu Domae, Takamitsu Matsubara
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10811934/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846099116714622976
author Cynthia Ochoa
Hanbit Oh
Yuhwan Kwon
Yukiyasu Domae
Takamitsu Matsubara
author_facet Cynthia Ochoa
Hanbit Oh
Yuhwan Kwon
Yukiyasu Domae
Takamitsu Matsubara
author_sort Cynthia Ochoa
collection DOAJ
description Imitation Learning (IL) is a promising approach for teaching tasks to robots by human demonstrations, although it faces challenges from long-horizon tasks and diverse goals in real-world settings. These issues stem from (i) a distribution mismatch between demonstrations and real-world execution and (ii) existing policy models that typically focus on prelearned final goals, limiting efficiency with diverse goals. To address this situation, we propose Interactive Sub-Goal-Planning Imitation Learning (ISPIL), an IL framework that learns hierarchical, goal-conditioned policies. Specifically, a high-level policy sets reachable sub-goals for the final goals, and a low-level policy executes the required actions. ISPIL interactively collects two types of demonstration data based on the novelty criteria: meta-sub-goal data, which represent with symbols the causal relationships between sub-goals, and action data, which consist of the physical robotic actions required to achieve these sub-goals. Meta-sub-goal data enable effective planning using a Regression Planning Network (RPN), and a sub-goal switching function helps reduce unnecessary data queries at the high level. We validate ISPIL through simulations and real-robot experiments in a kitchen-like environment and demonstrate improved task execution and generalizability across diverse goals.
format Article
id doaj-art-c4e2f9fb0f224dc2955fde8cd4eb1816
institution Kabale University
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c4e2f9fb0f224dc2955fde8cd4eb18162025-01-01T00:01:52ZengIEEEIEEE Access2169-35362024-01-011219761619763110.1109/ACCESS.2024.352130210811934ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse GoalsCynthia Ochoa0https://orcid.org/0009-0007-9220-2345Hanbit Oh1https://orcid.org/0000-0002-2368-3321Yuhwan Kwon2https://orcid.org/0000-0001-9058-1379Yukiyasu Domae3https://orcid.org/0000-0002-1366-9657Takamitsu Matsubara4https://orcid.org/0000-0003-3545-4814Division of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanIndustrial Cyber-Physical Systems Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Fukushima, JapanDivision of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanDivision of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanDivision of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanImitation Learning (IL) is a promising approach for teaching tasks to robots by human demonstrations, although it faces challenges from long-horizon tasks and diverse goals in real-world settings. These issues stem from (i) a distribution mismatch between demonstrations and real-world execution and (ii) existing policy models that typically focus on prelearned final goals, limiting efficiency with diverse goals. To address this situation, we propose Interactive Sub-Goal-Planning Imitation Learning (ISPIL), an IL framework that learns hierarchical, goal-conditioned policies. Specifically, a high-level policy sets reachable sub-goals for the final goals, and a low-level policy executes the required actions. ISPIL interactively collects two types of demonstration data based on the novelty criteria: meta-sub-goal data, which represent with symbols the causal relationships between sub-goals, and action data, which consist of the physical robotic actions required to achieve these sub-goals. Meta-sub-goal data enable effective planning using a Regression Planning Network (RPN), and a sub-goal switching function helps reduce unnecessary data queries at the high level. We validate ISPIL through simulations and real-robot experiments in a kitchen-like environment and demonstrate improved task execution and generalizability across diverse goals.https://ieeexplore.ieee.org/document/10811934/Interactive imitation learninglearning-to-planhierarchical policy
spellingShingle Cynthia Ochoa
Hanbit Oh
Yuhwan Kwon
Yukiyasu Domae
Takamitsu Matsubara
ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
IEEE Access
Interactive imitation learning
learning-to-plan
hierarchical policy
title ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
title_full ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
title_fullStr ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
title_full_unstemmed ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
title_short ISPIL: Interactive Sub-Goal-Planning Imitation Learning for Long-Horizon Tasks With Diverse Goals
title_sort ispil interactive sub goal planning imitation learning for long horizon tasks with diverse goals
topic Interactive imitation learning
learning-to-plan
hierarchical policy
url https://ieeexplore.ieee.org/document/10811934/
work_keys_str_mv AT cynthiaochoa ispilinteractivesubgoalplanningimitationlearningforlonghorizontaskswithdiversegoals
AT hanbitoh ispilinteractivesubgoalplanningimitationlearningforlonghorizontaskswithdiversegoals
AT yuhwankwon ispilinteractivesubgoalplanningimitationlearningforlonghorizontaskswithdiversegoals
AT yukiyasudomae ispilinteractivesubgoalplanningimitationlearningforlonghorizontaskswithdiversegoals
AT takamitsumatsubara ispilinteractivesubgoalplanningimitationlearningforlonghorizontaskswithdiversegoals