Aerospace and Electronic Systems - November 2018 - 4

Feature Article:

DOI. No. 10.1109/MAES.2018.170220

Nonvisible Satellite Estimation Algorithm for Improved
UAV Navigation in Mountainous Regions
Francesco De Vivo, Manuela Battipede, Piero Gili, Polytechnic of Turin, Torino, Italy

INTRODUCTION
Global Navigation Satellite System (GNSS) is one of the most
used technologies around the world for navigation. This technology is integrated in aircraft, ship, and submarine guidance systems;
it is available in cell phones and cars, and also used to monitor
Earth surface movements. The system accuracy spans from a range
of a few centimeters up to meters. This variation is related to different sources of error such as the receiver clock error, Earth's
ionospheric and tropospheric signal delay [1], or multipath effect
in semiobstructed areas. If the surrounding objects have highly reflective surfaces, the error magnitude might surge to several hundred meters due to multiple reflections that the signal experiences
before reaching the receiving antenna. A further source of error
is present when the same signal is received more than once. As
the present work is focused only on the effect of nonvisible satellites, multipath effect is not further investigated. In order to reduce
errors related to nondirect signal reception, a proposed solution
considers the exclusion of satellites that are not in line of sight
(NLOS), with the receiver and the use of the data related to the
excluded satellites for state estimation purposes. This data is also
used for different scopes, such as creating a time and space dependent safety bubble to improve obstacles safety clearance. Ignoring
NLOS satellites can lead to the improvement of the positioning accuracy [2]. However, excluding satellites, the dilution of precision
(DOP) associated with a bad geometrical repartition of the remaining satellites could increase, providing a less accurate position
estimation. In addition, if less than four satellites are accessible
for the localization the service becomes unavailable. In [3], NLOS
satellites are detected by projecting the satellite position on a virtual view from the vehicle position. The surrounding features act
as a mask. If satellite projections are occluded by the mask, they
are excluded. In [4], digital elevation maps and three-dimensional
(3D) data of the environment from cartography are used in order
to identify NLOS signals by ray-tracing technique [5]. In this case,
Authors' current address: Department of Mechanical and
Aerospace Engineering, Polytechnic of Turin, Corso Duca degli
Abruzzi, 24-10129 Torino, Italy, Email: (francesco.devivo@
polito.it).
Manuscript received December 17, 2017, revised March 20,
2018, May 28, 2018, and ready for publication June 4, 2018.
Review handled by G. Fasano.
0885/8985/18/$26.00 © 2018 IEEE
4

positioning of the receiver in the 3D model is needed. In [6], the
3D digital elevation model is not recorded. In this case, the Digital
Elevation Model (DEM) is built with Light Detection and Ranging
(LIDAR) data. The scanned model is used as a mask to determine
NLOS satellites. For fixed positions, excluding NLOS signals,
an accuracy of less than 1 m is achieved. In [7], two methods to
identify NLOS satellites are compared: the first is a vision-based
method, where the visible structures in roof-mounted fisheye images are processed [8], whereas the second is based on a carrierto-receiver noise (CN0) density threshold technique. Better results
have been achieved with the second method, namely by excluding NLOS signals with the CN0 threshold. In [9], a technique for
high-accuracy localization of moving vehicles that utilizes maps
of urban environments is proposed. Papers [10] and [11] introduce
a novel solution for autonomous navigation of a micro helicopter,
through a completely unknown environment, addressed by using
a single camera and inertial onboard sensors. The use of a monocular simultaneous localization and mapping (SLAM) framework to stabilize the vehicle in six degrees of freedom enables it
to overcome the problem of platform drift caused by a GNSS loss.
A similar approach, based on SLAM, is presented in [12] for indoor environments and urban canyon navigation. In [13] and [14]
a visual-inertial system is used to sustain prolonged real-world
GNSS-denied flights, both for indoor and outdoor GNSS-degraded
environments. A different approach, based on stereo camera and
laser rangefinder, is adopted in [15] to navigate a quadrotor helicopter in an unstructured and unknown GNSS-denied indoor environment. In [16], a novel sensor grid, using ultrasonic transmitters,
is presented for GNSS-denied navigation and 3D positioning applications. In [17], a novel information fusion algorithm based on
a particle filter is proposed to achieve lane level tracking accuracy
under a GNSS-denied environment. Reference [18] proposes a laser-based tracking system of moving objects using multiple mobile
robots as sensor nodes in GNSS-denied environments. In [19], the
problem of cooperative multirobot planning in unknown environments is investigated. This is done by actively developing belief
spaces planning approaches that account for the different sources
of uncertainty within planning and environment. In [20] and [21],
a 3D Lidar SLAM system is integrated with GNSS/ inertial navigation system by adapting the measurement noise covariance matrix
of an Extended Kalman Filter, to improve accuracy and reliability
of the position estimation in urban canyons. There is less literature dedicated to the optimization of the platform trajectory, starting from the knowledge of the regions of degraded GNSS signal

IEEE A&E SYSTEMS MAGAZINE

NOVEMBER 2018

Aerospace and Electronic Systems - November 2018

Table of Contents for the Digital Edition of Aerospace and Electronic Systems - November 2018