|Abstract:||Micro air vehicles (MAVs) are promising mobile platforms for exploring tasks. They are able to quickly reach the place of interest and to obtain important and crucial information on the ongoing situation. Beyond metrical representation, semantic information of the observed environment enhances the ability of situational awareness. Thus, the aim of this paper is the semantic mapping. It discusses how to gain place labels and assign them to metrical space. Herby, fine-tuned Convolution Neural Networks are used for place classification. After that, the gained semantics are assigned to the metrical space by using a grid map update strategy. In this context, different metrics for weighting of the observed grid elements are considered. The overall approach is evaluated by experimental data within a typical office environment. The resulting semantic map is easy to interpret.|
Proceedings of the 2019 International Technical Meeting of The Institute of Navigation
January 28 - 31, 2019
Hyatt Regency Reston
|Pages:||862 - 869|
|Cite this article:||
Atman, Jamal, Trommer, Gert F., "Place Classification and Semantic Mapping for MAV Applications," Proceedings of the 2019 International Technical Meeting of The Institute of Navigation, Reston, Virginia, January 2019, pp. 862-869.
ION Members/Non-Members: 1 Download Credit