The Art of Imitation: Learning Long-Horizon Manipulation Tasks from Few Demonstrations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: von Hartz, Jan Ole, Welschehold, Tim, Valada, Abhinav, Boedecker, Joschka
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909359483650048
author von Hartz, Jan Ole
Welschehold, Tim
Valada, Abhinav
Boedecker, Joschka
author_facet von Hartz, Jan Ole
Welschehold, Tim
Valada, Abhinav
Boedecker, Joschka
contents Task Parametrized Gaussian Mixture Models (TP-GMM) are a sample-efficient method for learning object-centric robot manipulation tasks. However, there are several open challenges to applying TP-GMMs in the wild. In this work, we tackle three crucial challenges synergistically. First, end-effector velocities are non-Euclidean and thus hard to model using standard GMMs. We thus propose to factorize the robot's end-effector velocity into its direction and magnitude, and model them using Riemannian GMMs. Second, we leverage the factorized velocities to segment and sequence skills from complex demonstration trajectories. Through the segmentation, we further align skill trajectories and hence leverage time as a powerful inductive bias. Third, we present a method to automatically detect relevant task parameters per skill from visual observations. Our approach enables learning complex manipulation tasks from just five demonstrations while using only RGB-D observations. Extensive experimental evaluations on RLBench demonstrate that our approach achieves state-of-the-art performance with 20-fold improved sample efficiency. Our policies generalize across different environments, object instances, and object positions, while the learned skills are reusable.
format Preprint
id arxiv_https___arxiv_org_abs_2407_13432
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle The Art of Imitation: Learning Long-Horizon Manipulation Tasks from Few Demonstrations
von Hartz, Jan Ole
Welschehold, Tim
Valada, Abhinav
Boedecker, Joschka
Robotics
Machine Learning
Task Parametrized Gaussian Mixture Models (TP-GMM) are a sample-efficient method for learning object-centric robot manipulation tasks. However, there are several open challenges to applying TP-GMMs in the wild. In this work, we tackle three crucial challenges synergistically. First, end-effector velocities are non-Euclidean and thus hard to model using standard GMMs. We thus propose to factorize the robot's end-effector velocity into its direction and magnitude, and model them using Riemannian GMMs. Second, we leverage the factorized velocities to segment and sequence skills from complex demonstration trajectories. Through the segmentation, we further align skill trajectories and hence leverage time as a powerful inductive bias. Third, we present a method to automatically detect relevant task parameters per skill from visual observations. Our approach enables learning complex manipulation tasks from just five demonstrations while using only RGB-D observations. Extensive experimental evaluations on RLBench demonstrate that our approach achieves state-of-the-art performance with 20-fold improved sample efficiency. Our policies generalize across different environments, object instances, and object positions, while the learned skills are reusable.
title The Art of Imitation: Learning Long-Horizon Manipulation Tasks from Few Demonstrations
topic Robotics
Machine Learning
url https://arxiv.org/abs/2407.13432