Towards Pose Estimation for Large UAV in Close Range

Ni Ou*, Junzheng Wang, Shangfei Liu, Jiehao Li

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)

Abstract

This paper deals with the problem of 4D pose estimation for large unmanned aerial vehicles (UAVs) in close range. A sensor system consisting of one single point laser range-finder and two cameras is designed and a novel pose estimation method based on vision fusion and point cloud registration is proposed. Our approach works on one-shot mode and only requires 10 samples with real poses for template construction. Through V-rep simulation environment, we generate two 200-sample datasets of different difficulty for evaluation. Error quantiles, 5cm5deg and 10cml0deg are three evaluation metrics used in our ablation experiments. It is illustrated that our method outperforms in robustness and precision due to proposed dimension extension modification and fusion of vision sensors.

Original languageEnglish
Title of host publicationProceedings - 2022 37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages56-61
Number of pages6
ISBN (Electronic)9781665465366
DOIs
Publication statusPublished - 2022
Event37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022 - Beijing, China
Duration: 19 Nov 202220 Nov 2022

Publication series

NameProceedings - 2022 37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022

Conference

Conference37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022
Country/TerritoryChina
CityBeijing
Period19/11/2220/11/22

Keywords

  • UAV
  • point cloud registration
  • pose estimation
  • sensor fusion

Fingerprint

Dive into the research topics of 'Towards Pose Estimation for Large UAV in Close Range'. Together they form a unique fingerprint.

Cite this