Visual-Inertial State-Estimation Using Ground Station for UAV

This paper presents the controller design and visual-inertial state estimation using IMU and camera for the quadrotor's flight. To accomplish this, first, the mathematical modeling of the quadrotor is presented based on Newtonian mechanics. Then a controller is designed that ensures the autonomous flight of the quadrotor. To achieve a high-level practical autonomy, state estimation is the first step. In this regard, vision-only estimation topologies are conventionally used. However, in a few very recent studies visual-inertial combined state estimation methodologies are introduced. In this research, the monocular visual-inertial system (VINS-mono) is utilized to achieve the autonomy of the quadrotor. The approach to employ this algorithm is different from existing work as the authors used offline ground computations that simplify the implementation. The results are presented to solidify the utility of the proposed method.

I. Introduction

Unmanned aerial vehicles (UAVs) have been around for more than a decade. Due to increased processing power capabilities and advancements in technologies, the research on autonomous aerial systems especially quadrotors has increased in the past decade. Due to their simple architecture and maneuverability, they have been used extensively for military purposes. The autonomy of these aerial vehicles makes them useful for a variety of applications, including monitoring, controlling disasters, and for other civil purposes as discussed in [1], [2] and [3]. For full autonomy in the flight of a UAV, state estimation plays a driving role and is one of the fundamentals required for a high level of autonomy.