Quality Evaluation of the Occupancy Grids without Ground Truth Maps

ICAART 2020: Proceedings of the 12th International Conference on Agents and Artificial Intelligence. Vol.1 2020
Ilze Andersone

Robot map merging is an important task in mobile multi-robot systems to facilitate cooperation and higher performance. Map merging has been extensively researched in recent years, but little attention has been paid to the merging of maps that have different quality levels. In this paper a method is proposed that allows the quality evaluation of occupancy grid maps without the need for ground truth maps. The method uses Convolutional Neural Network (CNN) for map fragment classification and can be used for overall map quality evaluation as well as for evaluation of map regions, which is especially useful for map merging purposes.

Keywords
Convolutional neural network, Multi-robot mapping, Occupancy grid maps, Robot map merging
DOI
10.5220/0009175503190326
Hyperlink
https://www.scitepress.org/Link.aspx?doi=10.5220/0009175503190326

Andersone, I. Quality Evaluation of the Occupancy Grids without Ground Truth Maps. In: ICAART 2020: Proceedings of the 12th International Conference on Agents and Artificial Intelligence. Vol.1, Malta, Valletta, 22-24 February, 2020. Setubal: SciTePress, 2020, pp.319-326. ISBN 978-989758395-7. Available from: doi:10.5220/0009175503190326

Publication language
English (en)