Publication
Beyond static perception: Integrating temporal context into VLMs for cloth folding
Conference Article
Conference
ICRA Workshop on Representing and Manipulating Deformable Objects (RMDO)
Edition
2025
Pages
1-4
Doc link
https://deformable-workshop.github.io/icra2025/spotlight/01_04_13_Barbany_beyond.pdf
File
Abstract
Manipulating clothes is challenging due to their complex dynamics, high deformability, and frequent selfocclusions. Garments exhibit a nearly infinite number of configurations, making explicit state representations difficult to define. In this paper, we analyze BiFold, a model that predicts language-conditioned pick-and-place actions from visual observations, while implicitly encoding garment state through end-to-end learning. To address scenarios such as crumpled garments or recovery from failed manipulations, BiFold leverages temporal context to improve state estimation. We examine the internal representations of the model and present evidence that its fine-tuning and temporal context enable effective alignment between text and image regions, as well as temporal consistency
Categories
artificial intelligence, computer vision.
Author keywords
VLM, Robotic Cloth Folding, Deformable Object Manipulation, LoRA Fine-Tuning, Temporal Consistency
Scientific reference
O. Barbany, A. Colomé and C. Torras. Beyond static perception: Integrating temporal context into VLMs for cloth folding, 2025 ICRA Workshop on Representing and Manipulating Deformable Objects, 2025, , pp. 1-4.
Follow us!