You are on page 1of 2

Final Year Project & Thesis

DESIGN OF ADAPTIVE CONTROLLER FOR UAV


(Unmanned Aerial Vehicle)

ABSTRACT

Although PID controllers work well on Unmanned Air Vehicles (UAVs), they require
tuning for each UAV. Also, they quickly lose performance in the presence of actuator
failures or changes in the UAV dynamics. Adaptive control algorithms that self tune to
each UAV and compensate for changes in the UAV during flight are explored. However,
because the autopilots on UAVs are small, many of the adaptive control algorithms like
those that employ least squares estimation may take too much code space, memory,
and/or computing power. In this project thesis I develop Lyapunov-based model reference
adaptive control (MRAC) scheme that is both simple and efficient with the UAV
autopilot resources. The scheme controls both roll and pitch through aileron and elevator
commands. The result of this Adaptive control is efficient. Results for open and closed
loop are also compared.

My thesis has two main contributions. They are the application of two varieties of MRAC
to UAVs. The first type of MRAC is a Lyapunov based MRAC. Lyapunov based MRACs
are not new, but do not seem to have been applied to UAVs. Adaptive scheme is derived
for MRAC. The results of scheme will be compared and contrasted.

Fahad Amin
Final Year Project & Thesis

Adaptive Control

Adaptive control is a term used to describe a large class of control systems. Simply put,
adaptive control algorithms adapt to unknown parameters in a plant. This does not mean
that an adaptive controller is the optimal controller for a system. Instead, adaptive
controllers are able to control plants with parameters that are unknown or changing. To
illustrate this idea look at traditional PID control. It can control a large set of plants, and
its intent is to drive the error between a desired reference signal and the output of the
plant to zero. It does this by operating on that error and passing the result to the plant's
input. The proportional part amplifies the error; it is used to drive the error to zero. The
integral part amplifies the integral of the error; it is used to eliminate steady-state error.
And the derivative part amplifies the derivative of the error; it is used to reduce
oscillations caused by the previous two parts. These three signals are added together and
passed to the plant's input. PID gain controllers are manually tuned for each system to
satisfaction of the operator. However, as the plant's parameters change, the PID controller
may need to be retuned. Parameter variation can be caused by changes in environmental
conditions, state changes (i.e. airplane dynamics changing as a result of airspeed, angle of
attach, and sideslip angle), time progression, etc. For UAVs, this means that a PID
controller that is tuned in one right regime may not work as well or become unstable
under another right regime, thus requiring retuning. Adaptive control typically does not
suffer from this problem.

The goal of adaptive control is to adjust to unknown or changing plant parameters. This is
accomplished by either changing parameters in the controller to minimize error, or using
plant parameter estimates to change the control signal. Therefore, there are many
different approaches to adaptive control.

Fahad Amin

You might also like