“Comparison of Indoor & Outdoor Visual Odometry using Tracking Camera T265”

Abstract

Most of the monocular visual odometry approaches are based on feature algorithms.
However, not all the feature point is necessarily favorable for the matching process. In this paper we
address novel technique of ORB as feature point detection to achieve real time necessity. We propose a
method which is precise and real time which generate multiple views with spatial angles. We generate a
dataset using Intel T265 camera and implements Monocular visual odometry with the help of various
feature detection/matching algorithms to estimate the performance in indoor and outdoor environment,
respectively. We demonstrate that how various image matching and detection algorithms works for
different environment.