Orienting Novel 3D Objects Using Self-Supervised Learning of Rotation Transforms

Orienting objects is a critical component in the automation of many packing and assembly tasks. We present an algorithm to orient novel objects given a depth image of the object in its current and desired orientation. We formulate a self-supervised objective for this problem and train a deep neural...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2021-05
Hauptverfasser: Devgon, Shivin, Ichnowski, Jeffrey, Balakrishna, Ashwin, Zhang, Harry, Goldberg, Ken
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Orienting objects is a critical component in the automation of many packing and assembly tasks. We present an algorithm to orient novel objects given a depth image of the object in its current and desired orientation. We formulate a self-supervised objective for this problem and train a deep neural network to estimate the 3D rotation as parameterized by a quaternion, between these current and desired depth images. We then use the trained network in a proportional controller to re-orient objects based on the estimated rotation between the two depth images. Results suggest that in simulation we can rotate unseen objects with unknown geometries by up to 30{\deg} with a median angle error of 1.47{\deg} over 100 random initial/desired orientations each for 22 novel objects. Experiments on physical objects suggest that the controller can achieve a median angle error of 4.2{\deg} over 10 random initial/desired orientations each for 5 objects.
ISSN:2331-8422