An Error-Aware Incremental Planar Motion Estimation Method Using PairedVertical Lines for Small Robots inUrban Areas1
Chapter
Overview
Identity
Additional Document Info
Other
View All
Overview
abstract
2015 Zhejiang University Press Co., Ltd. Published by Elsevier Inc. All rights reserved. We report on our development of a monocular visual odometry system based on vertical lines such as vertical edges of buildings and poles in urban areas. Because vertical lines are easy to extract, insensitive to lighting conditions/shadows, and sensitive to robot movements on a ground plane, they are excellent landmarks. We derive an incremental visual odometry method based on vertical line pairs. We analyze how errors are introduced and propagated in the continuous odometry process by deriving the closed-form representation of a covariance matrix. We formulate the minimum variance ego-motion estimation problem and present two different algorithms. The two algorithms have been extensively tested in physical experiments. The error-aware odometry method has also been compared with two popular methods and consistently outperforms these two counterparts in robustness, speed, and accuracy. The relative errors of the odometry are less than 2% in physical experiments.