Dominant plane detection using a RGB-D camera for autonomous navigation

Abstract

Dominant plane estimation is an fundamental task not only for trajectory finding problems but also autonomous navigation of mobile robots and MAVs (Micro Air Vehicles). In this paper, we illustrate a novel dominant plane detection approach from a RGB-D camera image sequences. A plane fitting as region growing technique is used in this work, rather than implementing the original algorithm, we modified it and updated to a incremental version, and optimised the plane calculation and mean square error calculation, to improve the accuracy and efficiency. The preliminary experimental results in different scenarios are presented by implementing the algorithm.