An efficient RANSAC hypothesis evaluation using sufficient statistics for RGB-D pose estimation

Ilankaikone Senthooran, Manzur Murshed, Jan Carlo Barca, Joarder Kamruzzaman, Hoam Chung

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Achieving autonomous flight in GPS-denied environments begins with pose estimation in three-dimensional space, and this is much more challenging in an MAV in a swarm robotic system due to limited computational resources. In vision-based pose estimation, outlier detection is the most time-consuming step. This usually involves a RANSAC procedure using the reprojection-error method for hypothesis evaluation. Realignment-based hypothesis evaluation method is observed to be more accurate, but the considerably slower speed makes it unsuitable for robots with limited resources. We use sufficient statistics of least-squares minimisation to speed up this process. The additive nature of these sufficient statistics makes it possible to compute pose estimates in each evaluation by reusing previously computed statistics. Thus estimates need not be calculated from scratch each time. The proposed method is tested on standard RANSAC, Preemptive RANSAC and R-RANSAC using benchmark datasets. The results show that the use of sufficient statistics speeds up the outlier detection process with realignment hypothesis evaluation for all RANSAC variants, achieving an execution speed of up to 6.72 times.

Original languageEnglish
Pages (from-to)1257-1270
Number of pages14
JournalAutonomous Robots
Volume43
Issue number5
DOIs
Publication statusPublished - Jun 2019

Keywords

  • Limited processing
  • MAV
  • Pose estimation
  • RANSAC
  • RGB-D
  • Visual odometry

Cite this

Senthooran, Ilankaikone ; Murshed, Manzur ; Barca, Jan Carlo ; Kamruzzaman, Joarder ; Chung, Hoam. / An efficient RANSAC hypothesis evaluation using sufficient statistics for RGB-D pose estimation. In: Autonomous Robots. 2019 ; Vol. 43, No. 5. pp. 1257-1270.
@article{c54d45517d0840d48cdfab88434d1847,
title = "An efficient RANSAC hypothesis evaluation using sufficient statistics for RGB-D pose estimation",
abstract = "Achieving autonomous flight in GPS-denied environments begins with pose estimation in three-dimensional space, and this is much more challenging in an MAV in a swarm robotic system due to limited computational resources. In vision-based pose estimation, outlier detection is the most time-consuming step. This usually involves a RANSAC procedure using the reprojection-error method for hypothesis evaluation. Realignment-based hypothesis evaluation method is observed to be more accurate, but the considerably slower speed makes it unsuitable for robots with limited resources. We use sufficient statistics of least-squares minimisation to speed up this process. The additive nature of these sufficient statistics makes it possible to compute pose estimates in each evaluation by reusing previously computed statistics. Thus estimates need not be calculated from scratch each time. The proposed method is tested on standard RANSAC, Preemptive RANSAC and R-RANSAC using benchmark datasets. The results show that the use of sufficient statistics speeds up the outlier detection process with realignment hypothesis evaluation for all RANSAC variants, achieving an execution speed of up to 6.72 times.",
keywords = "Limited processing, MAV, Pose estimation, RANSAC, RGB-D, Visual odometry",
author = "Ilankaikone Senthooran and Manzur Murshed and Barca, {Jan Carlo} and Joarder Kamruzzaman and Hoam Chung",
year = "2019",
month = "6",
doi = "10.1007/s10514-018-9801-y",
language = "English",
volume = "43",
pages = "1257--1270",
journal = "Autonomous Robots",
issn = "0929-5593",
publisher = "Springer-Verlag London Ltd.",
number = "5",

}

An efficient RANSAC hypothesis evaluation using sufficient statistics for RGB-D pose estimation. / Senthooran, Ilankaikone; Murshed, Manzur; Barca, Jan Carlo; Kamruzzaman, Joarder; Chung, Hoam.

In: Autonomous Robots, Vol. 43, No. 5, 06.2019, p. 1257-1270.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - An efficient RANSAC hypothesis evaluation using sufficient statistics for RGB-D pose estimation

AU - Senthooran, Ilankaikone

AU - Murshed, Manzur

AU - Barca, Jan Carlo

AU - Kamruzzaman, Joarder

AU - Chung, Hoam

PY - 2019/6

Y1 - 2019/6

N2 - Achieving autonomous flight in GPS-denied environments begins with pose estimation in three-dimensional space, and this is much more challenging in an MAV in a swarm robotic system due to limited computational resources. In vision-based pose estimation, outlier detection is the most time-consuming step. This usually involves a RANSAC procedure using the reprojection-error method for hypothesis evaluation. Realignment-based hypothesis evaluation method is observed to be more accurate, but the considerably slower speed makes it unsuitable for robots with limited resources. We use sufficient statistics of least-squares minimisation to speed up this process. The additive nature of these sufficient statistics makes it possible to compute pose estimates in each evaluation by reusing previously computed statistics. Thus estimates need not be calculated from scratch each time. The proposed method is tested on standard RANSAC, Preemptive RANSAC and R-RANSAC using benchmark datasets. The results show that the use of sufficient statistics speeds up the outlier detection process with realignment hypothesis evaluation for all RANSAC variants, achieving an execution speed of up to 6.72 times.

AB - Achieving autonomous flight in GPS-denied environments begins with pose estimation in three-dimensional space, and this is much more challenging in an MAV in a swarm robotic system due to limited computational resources. In vision-based pose estimation, outlier detection is the most time-consuming step. This usually involves a RANSAC procedure using the reprojection-error method for hypothesis evaluation. Realignment-based hypothesis evaluation method is observed to be more accurate, but the considerably slower speed makes it unsuitable for robots with limited resources. We use sufficient statistics of least-squares minimisation to speed up this process. The additive nature of these sufficient statistics makes it possible to compute pose estimates in each evaluation by reusing previously computed statistics. Thus estimates need not be calculated from scratch each time. The proposed method is tested on standard RANSAC, Preemptive RANSAC and R-RANSAC using benchmark datasets. The results show that the use of sufficient statistics speeds up the outlier detection process with realignment hypothesis evaluation for all RANSAC variants, achieving an execution speed of up to 6.72 times.

KW - Limited processing

KW - MAV

KW - Pose estimation

KW - RANSAC

KW - RGB-D

KW - Visual odometry

UR - http://www.scopus.com/inward/record.url?scp=85052559069&partnerID=8YFLogxK

U2 - 10.1007/s10514-018-9801-y

DO - 10.1007/s10514-018-9801-y

M3 - Article

VL - 43

SP - 1257

EP - 1270

JO - Autonomous Robots

JF - Autonomous Robots

SN - 0929-5593

IS - 5

ER -