Aerospace and Electronic Systems Magazine March 2017 - 61

Moraes et al.

Figure 3.

UAV landing procedure with the landing marker.

An important point to note about the OF is related to the computational costs of image processing. Since the information collected by the lenses is numerous and obtained at high frequencies,
the computational processes must be sufficiently fast so that the
effectiveness is guaranteed to avoid that decision procedures were
performed with delayed position data. Therefore, the PX4FLOW
camera employs a low-resolution camera that permits the OF calculation but has a much lower resolution than conventional cameras.
Despite the low resolution, the PX4FLOW camera permits to
identify textures and shapes in objects important to measure the
OF, but not identify more accurate details and colors.

THE FUZZY CONTROLLER
The fuzzy controller acts on landing operation by tracking augmented reality (AR) markers through ROS package ar_track_alvar.
A Mamdani controller was used. This type of fuzzy logic controller
is based on associating the areas under the curves of the membership functions. The input variables are horizontal distances dX and
dY. The UAV linear velocities vX, vY, and vZ, relative to the body
frame, are the output variables.
The strategy implemented consists in moving the UAV horizontally until the center of the marker, decreasing dX and dY, is
as shown in Figure 3, and then move it downwards. The rules in
Table 1, where the "p" and "n" prefixes correspond to positive or
negative distance or movement, describe this behavior.

POSITIONING ERROR TESTS WITH GPS MODULE +
OPTICAL FLOW + FUZZY CONTROLLER
OF is the distribution of the apparent velocities of brightness patterns in an image. OF can arise from relative motion of objects
and viewer [4]. The use of the OF technique, through processes
MARCH 2017

of detection, integration, and differentiation performed on the pattern of motion of pixels between subsequent frames is capable of
transmitting information about the motion of an aircraft, such as
velocity, acceleration, and direction, to a data reception unit [5].
Weighting this data against the information collected by the GPS
module, it is possible to reduce the error in the position estimated
by the aircraft, improving the control dynamics for the aircraft
landing process. Also, in environments with poor or denied GPS
signal, indoor scenarios for instance, OF can be used to estimate
the position of a UAV [6].
The ideal OF system for Pixhawk is PX4Flow. It is formed
by a module capable of intercalating 250 images per second. It is
noteworthy that the use of IMUs was a solution in order to cancel
any motion ambiguities.
The OF resource is deployed through weighted averages
taken from the measurements collected by the module and the
GPS readings. In this work, the weights were chosen equal to 0.5,
in a scale of 0 to 1, for both sensors. The environmental conditions used were in open space, various weather conditions, such
as wind speed, and air humidity. However, for the purpose of
calibrating the PX4Flow unit, the use of GPS was interrupted and
readings arising exclusively from the former sensor were plotted
as graphs. Figure 4 shows the movement along the Y axis.
Tests of tracking landpoints were performed to compare the precision of the system using only GPS and GPS plus OF and fuzzy
controller.
Figure 5 shows the landing coordinates X and Y, in centimeters, for the quadrotor without the OF (x markers), and with OF
and fuzzy controller (o markers). The "star" marker indicates the
desired position for landing.
After analyzing the results, it was noted that the absolute positioning error was calculated with a mean value of 18 cm, against
the 41.3 cm from the same test using only the GPS module, an
improvement of around 56.4%, using the weight of 0.5 for each of
the sensors used, as mentioned before.

Table 1.

Fuzzy Rules
Input Variables
dX

Output Variables

pFar

nFar

pSlow

nSlow

null

pFar

pSlow

null

pFar

pSlow

null

nFar

nSlow

null

nFar

nSlow

null

nFar

pFar

nSlow

pSlow

null

nFar

null

nSlow

null

nSlow

pFar

null

pSlow

null

IEEE A&E SYSTEMS MAGAZINE

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine March 2017