Aerospace and Electronic Systems Magazine October 2017 - 41

Changey et al.

Figure 3.

WiFi module, XBee Pro module and microSD card on the motherboard.

supply (5 V) of the servomotors. In addition to the management
role of actuators, there is an access, through this microcontroller, to
the current and voltage probes that can monitor the battery status,
regardless of the Gumstix system.
Two sensorless brushless motors are used to control the two
coaxial contrarotating rotors of the GLMAV. They require sophisticated control strategies and have dedicated electronic boards
known as the electronic speed controller (ESC), which estimates
the motor angular position and generates the appropriate waveforms on the motor phases accordingly. A tight control of the motors is needed during the deployment phase of the GLMAV: the
speed of the GLMAV is high when the rotors are deployed (more
than 50 m/s), and it has to slow down, stabilize, and reach a hover
flight within a few seconds. For this reason, specific strategies
have been developed for motor control, resulting in a custom ESC
firmware, which includes motor braking and closed-loop speed
control (this study is detailed in [13] and [14]).
The main onboard sensors are
C

An inertial measurement unit (IMU), which communicates
via the RS232 protocol with the Gumstix system.
A digital camera Point Grey Chameleon, connected to the
motherboard via Universal Serial Bus (USB 2.0). It is powered by 5 V and consumes up to 2 W.

A G-switch (Pewatron), which detects an acceleration higher
than the threshold of 300 g. It has a default open contact that
closes when this acceleration threshold is reached.
Additionally, the motherboard has control circuits for monitoring the current consumed and the voltage of the battery. This
information is accessible to the onboard programs via the analogdigital converters of the Gumstix computer, which alerts the user
of an anomaly detection in the event of excessive consumption or
low battery or both, under the voltage level necessary to control a
quick landing.
The motherboard has connectors ready for the use of additional
sensors. These sensors are not integrated in the present version of the
GLMAV, but some were used on intermediate models during the development and testing, and they can be easily integrated in the vehicle
if needed. An ultrasonic range finder was used (SRF08 reference) for
measuring a distance between 3 cm and 6 m, with a detection angle
of 55°; it is, therefore, very useful to approach phases for which the
accuracy of the barometer is too low. A Pitot tube was used for measurements of the projectile speed during the ballistic phase.
C

OCTOBER 2017

MOTHERBOARD OF THE GLMAV, SOFTWARE PART
The main onboard computer is a COM (Gumstix Overo). It integrates a microprocessor OMAP3530 from Texas Instruments
which has an ARM Cortex-A8 core (named general purpose processor [GPP]) associated with a digital signal processor (DSP) of
C64x+ type. It is able to reach more than 1200 Dhrystone million
instructions per second (DMIPS) and has 512 MB of RAM and
512 MB of flash memory (in addition to the memory card).
ARM processors are now the ones that offer the best relative
performance in terms of MIPS to consumed power. The GPP Overo module is designed to support a Linux operating system. The
Angstrom Linux distribution is installed and modified to give it
hard real-time functionalities, thanks to a Xenomai core.
The DSP C64x+ is a 32-bit fixed-point signal processor. The
special architecture of the operating system allows the DSP to be
devoted to heavy matrix calculations (optical flow and image processing) and allows the use of the GPP for other tasks.
Figure 4 shows the schematic diagram of the communication
between the processors. The main program running on the GPP is
articulated around a set of real-time tasks. Each task is prioritized
over the other tasks (Figure 5). At the program launching, after a
number of needed initializations, the following tasks are executed: IMU signal acquisition, range finder signal acquisition, video
server operation, Pitot tube acquisition, optical flow acquisition,
communication with ground station, battery pack checking, and
communication with the remote control.
Sensor data are processed by a set of tasks and functions designed for the acquisition, filtering, fusion, and formatting of the
data. Figure 6 illustrates this operation. Data acquisition is done
at fixed frequencies given in Figure 5. The IMU data frame is received by UART, and other sensor values are read on the I2C bus
(range finder and Pitot tube). Digital filtering is necessary on data
from multiple sensors, in particular, data from the inertial unit.
Indeed, the setting operation of the motors generates significant
vibrations that affect the sensors. Figure 7 shows, for example, the
rotational speed measurement from the gyrometer placed along the
x-axis (perpendicular to the body). The blue unfiltered measurement (radians per second) and the red filtered measurement using
a digital infinite impulse response are depicted. In the lower part of
the figure, the velocity of the lower rotor (revolutions per minute)
is plotted to show the correlation with the presence of vibrations.
The IMU is responsible for the necessary data fusion between
accelerometers, magnetometers, gyrometers, and global position-

IEEE A&E SYSTEMS MAGAZINE

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