Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Qianxu, Deng, Congyue, Lum, Tyler Ga Wei, Chen, Yuanpei, Yang, Yaodong, Bohg, Jeannette, Zhu, Yixin, Guibas, Leonidas
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910678434971648
author Wang, Qianxu
Deng, Congyue
Lum, Tyler Ga Wei
Chen, Yuanpei
Yang, Yaodong
Bohg, Jeannette
Zhu, Yixin
Guibas, Leonidas
author_facet Wang, Qianxu
Deng, Congyue
Lum, Tyler Ga Wei
Chen, Yuanpei
Yang, Yaodong
Bohg, Jeannette
Zhu, Yixin
Guibas, Leonidas
contents One-shot transfer of dexterous grasps to novel scenes with object and context variations has been a challenging problem. While distilled feature fields from large vision models have enabled semantic correspondences across 3D scenes, their features are point-based and restricted to object surfaces, limiting their capability of modeling complex semantic feature distributions for hand-object interactions. In this work, we propose the \textit{neural attention field} for representing semantic-aware dense feature fields in the 3D space by modeling inter-point relevance instead of individual point features. Core to it is a transformer decoder that computes the cross-attention between any 3D query point with all the scene points, and provides the query point feature with an attention-based aggregation. We further propose a self-supervised framework for training the transformer decoder from only a few 3D pointclouds without hand demonstrations. Post-training, the attention field can be applied to novel scenes for semantics-aware dexterous grasping from one-shot demonstration. Experiments show that our method provides better optimization landscapes by encouraging the end-effector to focus on task-relevant scene regions, resulting in significant improvements in success rates on real robots compared with the feature-field-based methods.
format Preprint
id arxiv_https___arxiv_org_abs_2410_23039
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping
Wang, Qianxu
Deng, Congyue
Lum, Tyler Ga Wei
Chen, Yuanpei
Yang, Yaodong
Bohg, Jeannette
Zhu, Yixin
Guibas, Leonidas
Robotics
Computer Vision and Pattern Recognition
One-shot transfer of dexterous grasps to novel scenes with object and context variations has been a challenging problem. While distilled feature fields from large vision models have enabled semantic correspondences across 3D scenes, their features are point-based and restricted to object surfaces, limiting their capability of modeling complex semantic feature distributions for hand-object interactions. In this work, we propose the \textit{neural attention field} for representing semantic-aware dense feature fields in the 3D space by modeling inter-point relevance instead of individual point features. Core to it is a transformer decoder that computes the cross-attention between any 3D query point with all the scene points, and provides the query point feature with an attention-based aggregation. We further propose a self-supervised framework for training the transformer decoder from only a few 3D pointclouds without hand demonstrations. Post-training, the attention field can be applied to novel scenes for semantics-aware dexterous grasping from one-shot demonstration. Experiments show that our method provides better optimization landscapes by encouraging the end-effector to focus on task-relevant scene regions, resulting in significant improvements in success rates on real robots compared with the feature-field-based methods.
title Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.23039