A new method to estimate the pose of an arbitrary 3D object without prerequisite knowledge: projection-based 3D perception

Yejun Kou, Yuichiro Toda, Mamoru Minami

Research output: Contribution to journalArticlepeer-review

Abstract

Recognizing a target object and measuring its pose are important functions of robot vision. Most recognition methods require prerequisite information about the target object to conduct the pose estimation, which limits the usability of the robot vision. To overcome this issue, the authors proposed a new approach to estimate an arbitrary target’s pose using stereo-vision, which was inspired by the parallactic character in human perception. The authors continued the previous research presented in AROB 2020 and expanded the ability of projection-based 3D perception (Pb3DP). Through tracking the trajectory of the target’s motion with a hand–eye robot, it has been confirmed that the Pb3DP method can provide a feasible result in the visual servoing of an unknown target object. In this paper, the authors introduce the methodology of the Pb3DP approach in detail and show the effectiveness of the method through the experimental results of visual servoing in 6 DoF using a stereo-vision hand–eye robot.

Original languageEnglish
Pages (from-to)149-158
Number of pages10
JournalArtificial Life and Robotics
Volume27
Issue number1
DOIs
Publication statusPublished - Feb 2022

Keywords

  • Arbitrary target object
  • Pose estimation
  • Projection-based

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'A new method to estimate the pose of an arbitrary 3D object without prerequisite knowledge: projection-based 3D perception'. Together they form a unique fingerprint.

Cite this