MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking
Tracking the 6D pose and velocity of objects represents a fundamental requirement for modern robotics manipulation tasks. 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 Filt...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2021-03-01
|
| Series: | Frontiers in Robotics and AI |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/frobt.2021.594583/full |
| _version_ | 1818596840628551680 |
|---|---|
| author | Nicola A. Piga Nicola A. Piga Fabrizio Bottarel Fabrizio Bottarel Claudio Fantacci Giulia Vezzani Ugo Pattacini Lorenzo Natale |
| author_facet | Nicola A. Piga Nicola A. Piga Fabrizio Bottarel Fabrizio Bottarel Claudio Fantacci Giulia Vezzani Ugo Pattacini Lorenzo Natale |
| author_sort | Nicola A. Piga |
| collection | DOAJ |
| description | Tracking the 6D pose and velocity of objects represents a fundamental requirement for modern robotics manipulation tasks. 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 the velocity of an object in real-time. MaskUKF achieves and in most cases surpasses state-of-the-art performance on the YCB-Video pose estimation benchmark without the need for expensive ground truth pose annotations at training time. Closed loop control experiments on the iCub humanoid platform in simulation show that joint pose and velocity tracking helps achieving higher precision and reliability than with one-shot deep pose estimation networks. A video of the experiments is available as Supplementary Material. |
| first_indexed | 2024-12-16T11:38:18Z |
| format | Article |
| id | doaj.art-b9c3c1d57a4f434b8b795326a9fa5fda |
| institution | Directory Open Access Journal |
| issn | 2296-9144 |
| language | English |
| last_indexed | 2024-12-16T11:38:18Z |
| publishDate | 2021-03-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Robotics and AI |
| spelling | doaj.art-b9c3c1d57a4f434b8b795326a9fa5fda2022-12-21T22:33:00ZengFrontiers Media S.A.Frontiers in Robotics and AI2296-91442021-03-01810.3389/frobt.2021.594583594583MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity TrackingNicola A. Piga0Nicola A. Piga1Fabrizio Bottarel2Fabrizio Bottarel3Claudio Fantacci4Giulia Vezzani5Ugo Pattacini6Lorenzo Natale7Humanoid Sensing and Perception, Istituto Italiano di Tecnologia, Genova, ItalyDipartimento di Informatica, Bioingegneria, Robotica e Ingegneria dei Sistemi, Università di Genova, Genova, ItalyHumanoid Sensing and Perception, Istituto Italiano di Tecnologia, Genova, ItalyDipartimento di Informatica, Bioingegneria, Robotica e Ingegneria dei Sistemi, Università di Genova, Genova, ItalyHumanoid Sensing and Perception, Istituto Italiano di Tecnologia, Genova, ItalyHumanoid Sensing and Perception, Istituto Italiano di Tecnologia, Genova, ItalyiCub Tech, Istituto Italiano di Tecnologia, Genova, ItalyHumanoid Sensing and Perception, Istituto Italiano di Tecnologia, Genova, ItalyTracking the 6D pose and velocity of objects represents a fundamental requirement for modern robotics manipulation tasks. 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 the velocity of an object in real-time. MaskUKF achieves and in most cases surpasses state-of-the-art performance on the YCB-Video pose estimation benchmark without the need for expensive ground truth pose annotations at training time. Closed loop control experiments on the iCub humanoid platform in simulation show that joint pose and velocity tracking helps achieving higher precision and reliability than with one-shot deep pose estimation networks. A video of the experiments is available as Supplementary Material.https://www.frontiersin.org/articles/10.3389/frobt.2021.594583/full6D object pose trackingobject velocity trackingunscented Kalman filteringdeep learning-aided filteringclosed loop manipulationhumanoid robotics |
| spellingShingle | Nicola A. Piga Nicola A. Piga Fabrizio Bottarel Fabrizio Bottarel Claudio Fantacci Giulia Vezzani Ugo Pattacini Lorenzo Natale MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking Frontiers in Robotics and AI 6D object pose tracking object velocity tracking unscented Kalman filtering deep learning-aided filtering closed loop manipulation humanoid robotics |
| title | MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking |
| title_full | MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking |
| title_fullStr | MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking |
| title_full_unstemmed | MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking |
| title_short | MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking |
| title_sort | maskukf an instance segmentation aided unscented kalman filter for 6d object pose and velocity tracking |
| topic | 6D object pose tracking object velocity tracking unscented Kalman filtering deep learning-aided filtering closed loop manipulation humanoid robotics |
| url | https://www.frontiersin.org/articles/10.3389/frobt.2021.594583/full |
| work_keys_str_mv | AT nicolaapiga maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT nicolaapiga maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT fabriziobottarel maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT fabriziobottarel maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT claudiofantacci maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT giuliavezzani maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT ugopattacini maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking AT lorenzonatale maskukfaninstancesegmentationaidedunscentedkalmanfilterfor6dobjectposeandvelocitytracking |