by Jiang, R, Klette, R and Wang, S
Abstract:
An intrinsic problem of visual odometry is its drift in long-range navigation. The drift is caused by error accumulation, as visual odometry is based on relative measurements. The paper reviews algorithms that adopt various methods to minimize this drift. However, as far as we know, no work has been done to statistically model and analyze the intrinsic properties of this drift. Moreover, the quantification of drift using offset ratio has its drawbacks. This paper models the drift as a combination of wide-band noise and a first-order Gauss-Markov process, and analyzes it using Allan variance. The model’s parameters are identified by a statistical method. A novel drift quantification method using Monte Carlo simulation is also provided. © 2011 Springer-Verlag Berlin Heidelberg.
Reference:
Statistical modeling of long-range drift in visual odometry (Jiang, R, Klette, R and Wang, S), In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), volume 6469 LNCS, 2011.
Bibtex Entry:
@inproceedings{jiang2011statisticalodometry,
author = "Jiang, R and Klette, R and Wang, S",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "214--224",
title = "Statistical modeling of long-range drift in visual odometry",
volume = "6469 LNCS",
year = "2011",
abstract = "An intrinsic problem of visual odometry is its drift in long-range navigation. The drift is caused by error accumulation, as visual odometry is based on relative measurements. The paper reviews algorithms that adopt various methods to minimize this drift. However, as far as we know, no work has been done to statistically model and analyze the intrinsic properties of this drift. Moreover, the quantification of drift using offset ratio has its drawbacks. This paper models the drift as a combination of wide-band noise and a first-order Gauss-Markov process, and analyzes it using Allan variance. The model's parameters are identified by a statistical method. A novel drift quantification method using Monte Carlo simulation is also provided. © 2011 Springer-Verlag Berlin Heidelberg.",
doi = "10.1007/978-3-642-22819-3_22",
isbn = "9783642228186",
issn = "0302-9743",
eissn = "1611-3349",
issue = "PART 2",
language = "eng",
}