Jiwu Wang, Shunkai Zheng, Yao Du, Sugisaka Masanori
Available Online 1 December 2015.
SLAM; ORB; Monocular Vision, Mobile Robot
In order to reduce the accumulative errors in our monocular SLAM (simultaneous localization and mapping), the loop closing (detection + correction) method based on PTAM (Parallel Tracking and Mapping) is applied. Here natural environment features are necessary to be extracted efficiently, so the ORB (Oriented FAST and Rotated BRIEF (Binary Robust Independent Elementary Features)) algorithm is used for the feature extraction and matching. The experiment results show that there is strong feature recognition power in ORB so that it can realize environment recognition under the conditions of severe viewpoint change. Moreover, it is so fast to extract and match (without using multi-threading or GPU (Graphics Processing Unit) acceleration) that it can accurately track and map in real time. Its capability of fast and efficiency is verified with outdoor scene experiments.
© 2013, the Authors. Published by ALife Robotics Corp. Ltd
This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).