Visual Inertial Navigation for a Small UAV Using Sparse and Dense Optical Flow

Abstract

There has been a recent surge in interest regarding small Unmanned Aerial Vehicles (UAV) that operate in indoor environments. A crucial part of such platforms is the navigation system that allows them to navigate in Global Navigation Satellite System (GNSS) denied environments. The two objectives of this work were to develop an Optical Flow (OF) based method for navigation and a high-fidelity simulation for testing and verification. This paper presents the Visual-Inertial Navigation System (VINS) developed that utilises sparse and dense Optical Flow patterns within an Extended Kalman Filter (EKF) framework for autonomous navigation in GNSS denied environments. The principal novelty is the use of dense Optical Flow measurements to estimate the velocity of the UAV.