MISCGrasp: Leveraging Multiple Integrated Scales and Contrastive Learning for Enhanced Volumetric Grasping

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Fan, Qingyu, Cai, Yinghao, Li, Chao, Jiao, Chunting, Zheng, Xudong, Lu, Tao, Liang, Bin, Wang, Shuo
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911036882288640
author Fan, Qingyu
Cai, Yinghao
Li, Chao
Jiao, Chunting
Zheng, Xudong
Lu, Tao
Liang, Bin
Wang, Shuo
author_facet Fan, Qingyu
Cai, Yinghao
Li, Chao
Jiao, Chunting
Zheng, Xudong
Lu, Tao
Liang, Bin
Wang, Shuo
contents Robotic grasping faces challenges in adapting to objects with varying shapes and sizes. In this paper, we introduce MISCGrasp, a volumetric grasping method that integrates multi-scale feature extraction with contrastive feature enhancement for self-adaptive grasping. We propose a query-based interaction between high-level and low-level features through the Insight Transformer, while the Empower Transformer selectively attends to the highest-level features, which synergistically strikes a balance between focusing on fine geometric details and overall geometric structures. Furthermore, MISCGrasp utilizes multi-scale contrastive learning to exploit similarities among positive grasp samples, ensuring consistency across multi-scale features. Extensive experiments in both simulated and real-world environments demonstrate that MISCGrasp outperforms baseline and variant methods in tabletop decluttering tasks. More details are available at https://miscgrasp.github.io/.
format Preprint
id arxiv_https___arxiv_org_abs_2507_02672
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MISCGrasp: Leveraging Multiple Integrated Scales and Contrastive Learning for Enhanced Volumetric Grasping
Fan, Qingyu
Cai, Yinghao
Li, Chao
Jiao, Chunting
Zheng, Xudong
Lu, Tao
Liang, Bin
Wang, Shuo
Robotics
Computer Vision and Pattern Recognition
Robotic grasping faces challenges in adapting to objects with varying shapes and sizes. In this paper, we introduce MISCGrasp, a volumetric grasping method that integrates multi-scale feature extraction with contrastive feature enhancement for self-adaptive grasping. We propose a query-based interaction between high-level and low-level features through the Insight Transformer, while the Empower Transformer selectively attends to the highest-level features, which synergistically strikes a balance between focusing on fine geometric details and overall geometric structures. Furthermore, MISCGrasp utilizes multi-scale contrastive learning to exploit similarities among positive grasp samples, ensuring consistency across multi-scale features. Extensive experiments in both simulated and real-world environments demonstrate that MISCGrasp outperforms baseline and variant methods in tabletop decluttering tasks. More details are available at https://miscgrasp.github.io/.
title MISCGrasp: Leveraging Multiple Integrated Scales and Contrastive Learning for Enhanced Volumetric Grasping
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2507.02672