Hardware and Software System
for UAV Control
Petru Emanuel Stingu
PhD candidate,
ACS group
pestingu[.at.]arri.uta.edu
This project aims to develop the electronics and
the software for controlling a Joker 2 electric helicopter in order to allow
autonomous flight without human intervention. To avoid expensive crashes, the
model helicopter is kept onto a test stand that allows all the degrees of
freedom.
Current research is done on sensor fusion using
a Kalman filter, system identification and Neuro-Fuzzy control. In order to
implement these techniques, a low-power and low weight electronic system is
being designed.
|
 |
Directional control of the quad-rotor by
tilting the remote control. Unsuccessful trial to get a step response to
pilot commands. A more complex altitude controller was
implemented. It has better performance than the proportional controller,
but at the expense of small oscillations. |
|
|
Directional control of the quad-rotor by
tilting the remote control.
A proportional altitude controller was
added. It uses an ultrasound range sensor for the vertical position. |
|
|
Directional control of the quad-rotor by
tilting the remote control.
The direction is relative to the pilot
and not to the quad-rotor body frame, thus allowing an untrained person
to fly it easily. |
 |
The attitude is estimated using the gyroscopes, the accelerometer and
the magnetic field. The gyroscope readings are rotated from body to
Earth axes and integrated to give a low-noise signal. The drift due to
biases is eliminated using feedback from the noisy absolute attitude
readings from the QUEST algorithm (that uses the accelerometer and the
magnetic field sensors). |
|
 |
A (complex) nonlinear model for a helicopter was developed in Simulink
after Mark Dreier's book "Introduction
to Helicopter and Tiltrotor Flight Simulation". It will be used to
develop and test controllers for the real helicopters that we have and
to help in System Identification.
Here it is "flown" using a radio remote control and its outputs
displayed using the open-source
FlightGear
flight simulator. |
|
 |
A simple proportional controller that tries to keep the quad-rotor
horizontal. The platform is still unbalanced and the drift can not be
avoided. |
|
 |
The first test of the quad-rotor. The weight is unbalanced and there is
no controller implemented yet. All is done from the remote control now. |
|
 |
We build all the sensors and most of the on-board electronics. Here are
some pictures of a few PCBs. |
 |
The new Joker 2 electric helicopters have
arrived and have been assembled.
|
 |
Experiment to verify the correct functionality
of the PI controller for the combustion motor.
A feed-forward term needs to be included for
better performance to sudden changes in the collective angle.
The
Ground Vehicle provides an easy way to transport the equipment in the field. |
|
 |
Experiment to identify the longitudinal axis parameters for the nonlinear
model.
At
one point, the motor was turned off suddenly and the tail rotor has broken,
probably because of the metal fatigue of a screw and because the tail rotor
doesn't yet have an autorotation capability (will be upgraded soon). |
 |
Accident when thee helicopter was on the helicopter stand.
The
tail rotor flew away from the shaft and hit the main rotor, which in turn
has hit the tail boom. |
|
|
Preliminary identification of
the combustion motor model using a Neuro-Fuzzy algorithm (LOLIMOT) |
|
|
|
|
|
Other movies and pictures from our lab can
be found here:
http://arri.uta.edu/acs/mcmurrough/ |
