Authors:
Anderson Souza
1
;
Leonardo Souto
2
;
Fabio Fonseca de Oliveira
2
;
Biswa Nath Datta
3
and
Luiz M. G. Gonçalves
2
Affiliations:
1
University of the State of Rio Grande do Norte, Brazil
;
2
Federal University of the Rio Grande do Norte, Brazil
;
3
Northern Illinois University, United States
Keyword(s):
Visual Odometry, Polynomial Eigenvalue Problem, Motion Estimation.
Related
Ontology
Subjects/Areas/Topics:
Autonomous Agents
;
Image Processing
;
Informatics in Control, Automation and Robotics
;
Robotics and Automation
;
Vision, Recognition and Reconstruction
Abstract:
Visual Odometry (VO) is the process of calculating the motion of an agent (such as, robot and vehicle), using
images captured by a single or multiple cameras embedded to it. VO is an important process to supplement
autonomous navigation systems, since VO can provide accurate trajectory estimates. However, algorithms
of VO work with several steps of hard numerical computation which generate numerical errors and demand
considerable processing time. In this paper, we propose the use of a mathematical framework for monocular
VO process based on Polynomial Eigenvalue Problem (PEP) modeling in order to achieve both more accurate
motion estimation and to decrease the processing time of the VO process. Some previous experiments are
shown in order to validate the proposed computation accuracy.