RMVD: Robust Monocular VSLAM for Moving Robot in Dynamic Environment

• Published in: Cognitive Systems and Signal Processing Vol. 1006; pp. 454 - 464
• Main Authors: Li, Qile, Sun, Fuchun, Liu, Huaping
• Format: Book Chapter
• Language: English
• Published: Singapore Springer 2019; Springer Singapore
• Series: Communications in Computer and Information Science
• Subjects: Dynamic; Monocular slam; Moving robot
• ISBN: 9789811379857; 9811379858
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-981-13-7986-4_40

For moving robot, accurate localization and mapping is important for route planning. And for mower robot, it is hard to localizing and mapping because of the dynamic working environment of mower robot, and especially hard for robot which takes use of monocular visual sensor. In this paper, we introduced our work of an intelligent monocular visual slam system which has been applied on our mower robot platform. And the system works well in our mower robots testing environment, which is outdoors and dynamic. In our work, we combines method of traditional slam algorithm and method of deep learning to deal with dynamic outdoors environment. The slam system can work in real time for robots localization. Profiting from our robust and real-time slam system, we can implement accurate route planning which takes use of localization result of the mower robots slam system.