Sensitivity Study of Fiducial-Aided Navigation of Unmanned Aerial Vehicles

TL;DR

This paper investigates how camera-based line-of-sight measurements with fiducials can improve UAV navigation accuracy in urban environments where GNSS signals are unreliable, using an EKF to fuse IMU and LOS data.

Contribution

It introduces a sensitivity analysis of UAV navigation accuracy using fiducial-aided measurements and develops an EKF framework for integrating IMU and LOS data.

Findings

Fiducial placement significantly affects navigation accuracy.

Higher IMU quality improves state estimation.

Increased image processing frequency enhances system performance.

Abstract

The possible applications and benefits of autonomous Unmanned Aerial Vehicle (UAV) use in urban areas are gaining considerable attention. Before these possibilities can be realized, it is essential that UAVs be able to navigate reliably and precisely in urban environments. The most common means of determining the location of a UAV is to utilize position measurements from Global Navigation Satellite Systems (GNSS). In urban environments, however, GNSS measurements are significantly degraded due to occlusions and multipath. This research analyzes the use of camera Line-of-Sight (LOS) measurements to self-describing fiducials as a replacement for conventional GNSS measurements. An extended Kalman filter (EKF) is developed and validated for the purpose of combining continuous measurements from an Inertial Measurement Unit (IMU) with the discrete LOS measurements to accurately estimate the states of a UAV. The sensitivity of the estimation error covariance to various system parameters is assessed, including IMU grade, fiducial placement, vehicle altitude, and image processing frequency.