8 Hits in 2.4 sec

Self-supervised 6D Object Pose Estimation for Robot Manipulation [article]

Xinke Deng, Yu Xiang, Arsalan Mousavian, Clemens Eppner, Timothy Bretl, Dieter Fox
2020 arXiv   pre-print
In this work, we introduce a robot system for self-supervised 6D object pose estimation.  ...  Starting from modules trained in simulation, our system is able to label real world images with accurate 6D object poses for self-supervised learning.  ...  PoseRBPF is a Rao-Blackwellized particle filter combined with a learned auto-encoder network [8] for 6D object pose estimation.  ... 
arXiv:1909.10159v2 fatcat:2pjdmi5ai5czlhwti2dkcc2fny

VIPose: Real-time Visual-Inertial 6D Object Pose Tracking [article]

Rundong Ge, Giuseppe Loianno
2021 arXiv   pre-print
In this work, we introduce a novel Deep Neural Network (DNN) called VIPose, that combines inertial and camera data to address the object pose tracking problem in real-time.  ...  Estimating the 6D pose of objects is beneficial for robotics tasks such as transportation, autonomous navigation, manipulation as well as in scenarios beyond robotics like virtual and augmented reality  ...  PoseRBPF [12] proposes a Rao-Blackwellized particle filter approach for 6D pose tracking, which decouples 3D rotation and 3D translation and takes the uncertainty and object symmetry into account, achieving  ... 
arXiv:2107.12617v2 fatcat:rukwliflyvbrti7pzkw6m5loym

MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking

Nicola A Piga, Fabrizio Bottarel, Claudio Fantacci, Giulia Vezzani, Ugo Pattacini, Lorenzo Natale
2021 Frontiers in Robotics and AI  
This paper proposes a 6D object pose tracking algorithm, called MaskUKF, that combines deep object segmentation networks and depth information with a serial Unscented Kalman Filter to track the pose and  ...  Tracking the 6D pose and velocity of objects represents a fundamental requirement for modern robotics manipulation tasks.  ...  PoseRBPF (Deng et al., 2019 ) is a 6D object pose tracking algorithm adopting a deep autoencoder for implicit orientation encoding (Martin et al., 2018) and a particle filter to track the position and  ... 
doi:10.3389/frobt.2021.594583 pmid:33996920 pmcid:PMC8114180 fatcat:ibzdmo54c5gkzpgxp6x234l2by

Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains [article]

Bowen Wen, Chaitanya Mitash, Kostas Bekris
2022 arXiv   pre-print
Tracking the 6D pose of objects in video sequences is important for robot manipulation. This work presents se(3)-TrackNet, a data-driven optimization approach for long term, 6D pose tracking.  ...  Code, data and supplementary video for this project are available at  ...  Recent work [4] proposed a data-driven Rao-Blackwellized particle filter, achieving promising results on the YCB Video benchmark.  ... 
arXiv:2105.14391v2 fatcat:6pxpnun7ffe77iyn47zzhchbhy

se(3)-TrackNet: Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains [article]

Bowen Wen, Chaitanya Mitash, Baozhang Ren, Kostas E. Bekris
2020 arXiv   pre-print
Tracking the 6D pose of objects in video sequences is important for robot manipulation.  ...  This work proposes a data-driven optimization approach for long-term, 6D pose tracking.  ...  Recent work [2] proposed a Rao-Blackwellized particle filter, which decouples the translational and rotational uncertainty, achieving state-of-art 6D pose tracking performance on the YCB Video benchmark  ... 
arXiv:2007.13866v1 fatcat:pskz3r4wereujkpggpqs2dv23i

A Multi-Hypothesis Approach to Pose Ambiguity in Object-Based SLAM [article]

Jiahui Fu, Qiangqiang Huang, Kevin Doherty, Yue Wang, John J. Leonard
2021 arXiv   pre-print
In object-based Simultaneous Localization and Mapping (SLAM), 6D object poses offer a compact representation of landmark geometry useful for downstream planning and manipulation tasks.  ...  In particular, we (1) present a learned pose estimation network that provides multiple hypotheses about the 6D pose of an object; (2) by treating the output of our network as components of a mixture model  ...  PoseRBPF [24] employs a Rao-Blackwellized particle filter to track object poses across consecutive frames in the frontend, demonstrating robustness to object shape symmetry when a sufficient number of  ... 
arXiv:2108.01225v1 fatcat:qtsqbkfa3vhr3mluq436nt7yka

Deep Learning on Monocular Object Pose Detection and Tracking: A Comprehensive Overview [article]

Zhaoxin Fan, Yazhi Zhu, Yulin He, Qi Sun, Hongyan Liu, Jun He
2022 arXiv   pre-print
Among methods for object pose detection and tracking, deep learning is the most promising one that has shown better performance than others.  ...  detection, category-level monocular object pose detection, and monocular object pose tracking.  ...  Specifically, it combines Rao-Blackwellized particle filtering with a learned auto encoder network to update object pose.  ... 
arXiv:2105.14291v2 fatcat:2kxd4owthvf7tbcbnlqlqu4r3m

A Survey of 6D Object Detection Based on 3D Models for Industrial Applications

Felix Gorschlüter, Pavel Rojtberg, Thomas Pöllabauer
The second contribution is a collection of quantitative evaluation results for several different 6D object detection methods trained with synthetic data and the comparison and analysis thereof.  ...  Six-dimensional object detection of rigid objects is a problem especially relevant for quality control and robotic manipulation in industrial contexts.  ...  [33] considers rotation and translation separately, using a Rao-Blackwellized particle filtering framework. He et al.  ... 
doi:10.26083/tuprints-00021027 fatcat:rebesb3ozjbg7e5v2s24mti7g4