Neural Implicit Representation for Building Digital Twins of Unknown Articulated Objects

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Weng, Yijia, Wen, Bowen, Tremblay, Jonathan, Blukis, Valts, Fox, Dieter, Guibas, Leonidas, Birchfield, Stan
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916279145725952
author Weng, Yijia
Wen, Bowen
Tremblay, Jonathan
Blukis, Valts
Fox, Dieter
Guibas, Leonidas
Birchfield, Stan
author_facet Weng, Yijia
Wen, Bowen
Tremblay, Jonathan
Blukis, Valts
Fox, Dieter
Guibas, Leonidas
Birchfield, Stan
contents We address the problem of building digital twins of unknown articulated objects from two RGBD scans of the object at different articulation states. We decompose the problem into two stages, each addressing distinct aspects. Our method first reconstructs object-level shape at each state, then recovers the underlying articulation model including part segmentation and joint articulations that associate the two states. By explicitly modeling point-level correspondences and exploiting cues from images, 3D reconstructions, and kinematics, our method yields more accurate and stable results compared to prior work. It also handles more than one movable part and does not rely on any object shape or structure priors. Project page: https://github.com/NVlabs/DigitalTwinArt
format Preprint
id arxiv_https___arxiv_org_abs_2404_01440
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Neural Implicit Representation for Building Digital Twins of Unknown Articulated Objects
Weng, Yijia
Wen, Bowen
Tremblay, Jonathan
Blukis, Valts
Fox, Dieter
Guibas, Leonidas
Birchfield, Stan
Computer Vision and Pattern Recognition
Artificial Intelligence
Graphics
Robotics
We address the problem of building digital twins of unknown articulated objects from two RGBD scans of the object at different articulation states. We decompose the problem into two stages, each addressing distinct aspects. Our method first reconstructs object-level shape at each state, then recovers the underlying articulation model including part segmentation and joint articulations that associate the two states. By explicitly modeling point-level correspondences and exploiting cues from images, 3D reconstructions, and kinematics, our method yields more accurate and stable results compared to prior work. It also handles more than one movable part and does not rely on any object shape or structure priors. Project page: https://github.com/NVlabs/DigitalTwinArt
title Neural Implicit Representation for Building Digital Twins of Unknown Articulated Objects
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Graphics
Robotics
url https://arxiv.org/abs/2404.01440