Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kadi, Halid Abdulrahim, Terzić, Kasim
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2508.17070
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915460115595264
author Kadi, Halid Abdulrahim
Terzić, Kasim
author_facet Kadi, Halid Abdulrahim
Terzić, Kasim
contents We present a novel goal-conditioned recurrent state space (GC-RSSM) model capable of learning latent dynamics of pick-and-place garment manipulation. Our proposed method LaGarNet matches the state-of-the-art performance of mesh-based methods, marking the first successful application of state-space models on complex garments. LaGarNet trains on a coverage-alignment reward and a dataset collected through a general procedure supported by a random policy and a diffusion policy learned from few human demonstrations; it substantially reduces the inductive biases introduced in the previous similar methods. We demonstrate that a single-policy LaGarNet achieves flattening on four different types of garments in both real-world and simulation settings.
format Preprint
id arxiv_https___arxiv_org_abs_2508_17070
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LaGarNet: Goal-Conditioned Recurrent State-Space Models for Pick-and-Place Garment Flattening
Kadi, Halid Abdulrahim
Terzić, Kasim
Robotics
We present a novel goal-conditioned recurrent state space (GC-RSSM) model capable of learning latent dynamics of pick-and-place garment manipulation. Our proposed method LaGarNet matches the state-of-the-art performance of mesh-based methods, marking the first successful application of state-space models on complex garments. LaGarNet trains on a coverage-alignment reward and a dataset collected through a general procedure supported by a random policy and a diffusion policy learned from few human demonstrations; it substantially reduces the inductive biases introduced in the previous similar methods. We demonstrate that a single-policy LaGarNet achieves flattening on four different types of garments in both real-world and simulation settings.
title LaGarNet: Goal-Conditioned Recurrent State-Space Models for Pick-and-Place Garment Flattening
topic Robotics
url https://arxiv.org/abs/2508.17070