3D SLAM using Microsoft Kinect

Abstract

In this thesis, the Microsoft Kinect sensor has been considered for the purpose of simultaneous localization and mapping. A simple system for 6D pose estimation and 3D mapping is presented. The system uses the Kinect sensor, feature detection and recognition based on the speeded up robust features algorithm, estimation of movement based on the iterative closest point and random sample consensus algorithms and mapping with RGB colored 3D point clouds. The results indicate that the Kinect, which provides both 3D depth and RGB color images, is a suitable sensor for simultaneous localization and mapping systems. The system shows robustness for frame-to-frame rotation angles of 5 or less.