Journal article Open Access

Semi-Automatic Generation of Training Data for Neural Networks for 6D Pose Estimation and Robotic Grasping

Johannes Rauer; Mohamed Aburaia; Wilfried Wöber

Machine-learning-based approaches for
pose estimation are trained using annotated groundtruth
data – images showing the object and information
of its pose. In this work an approach to semiautomatically
generate 6D pose-annotated data, using
a movable marker and an articulated robot, is
presented. A neural network for pose estimation is
trained using datasets varying in size and type. The
evaluation shows that small datasets recorded in the
target domain and supplemented with augmented images
lead to more robust results than larger synthetic
datasets. The results demonstrate that a mobile manipulator
using the proposed pose-estimation system
could be deployed in real-life logistics applications
to increase the level of automation.

Files (374.6 kB)
Name Size
Semi-Automatic Generation of Training Data for Neural Networks.pdf
374.6 kB Download
All versions This version
Views 112112
Downloads 2828
Data volume 10.5 MB10.5 MB
Unique views 101101
Unique downloads 2525


Cite as