Challenges in Monocular Visual Odometry: Photometric Calibration, Motion Bias and Rolling Shutter Effect
share
Monocular visual odometry (VO) has seen tremendous improvements in accuracy, robustness and efficiency, and has gained exponential popularity over recent years. Nevertheless, no comprehensive evaluations have been performed to reveal the influences of the three easily overlooked, yet very influential aspects: photometric calibration, motion bias and rolling shutter effect. In this work, we evaluate these three aspects quantitatively on the state of the art of direct, feature-based and semi-direct methods, providing the community with useful practical knowledge both for better applying existing methods and developing new algorithms of VO and SLAM. Conclusions (some of which are counterintuitive) are drawn with insightful technical and empirical analyses to all of our experiments. Possible improvements on existing methods are directed or proposed, such as a sub-pixel accuracy refinement of ORB-SLAM which boosts its performance.
READ FULL TEXT