Aerospace Instrument-Making Annotation << Back APPLICATION OF LIE GROUPS FOR VISUAL ODOMETRY IN THE CONTEXT OF DEEP LEARNING A.M. Andreev, A.V. Babichenko, F.A. Konokhov The article is considered the possibility of applying various methods of representing Lie groups as applied to the problem of visual odometry. Evaluations of the possibilities of using Lie groups in robotics and computer vision are given. We considered different representations of Lie groups, such as SO(3) and SE(3). The main problems of the rotation-translation presentation for visual odometry are also considered. Tests were carried out and a pivot table was constructed. With help of this table an estimation of the conversion accuracy and error accumulation was given for various representations of Lie groups. Key words: Lie groups, quaternions, visual odometry, deep learning. DOI: 10.25791/aviakosmos.10.2019.930 Contacts: - Pp. 44-51.