- •Contents
- •Introduction
- •Principle of operation
- •Math model
- •Equations of motion
- •Passivity properties of the system
- •Linearization of the system
- •Feedback linearization
- •Energy-based control design
- •Simulation results
- •Programming code for Arduino
- •Project plan
- •Project budget
- •Applicability
- •References
Math model
Equations of motion
The reaction wheel pendulum is a two-degree-of-freedom robot as shown in Figure 3.1. The pendulum constitutes the first link, while the rotating wheel is the second one. The angle of the pendulum is q1 and is measured clockwise from the vertical. The angle of the wheel is q2.
The parameters of the system are described in the following table.
: Mass
of the pendulum
: Mass
of the wheel
: Length
of the pendulum
: Distance
to the center of mass of the pendulum
: Moment
of inertia of the pendulum
: Moment
of inertia of the wheel
: Angle
that the pendulum makes with the vertical
: Angle
of the wheel
: Motor
torque input applied on the disk
We
introduce the parameter
,
which will be used later.
The
kinetic energy of the pendulum is
and the kinetic energy of the wheel is
.
F
igure
3.1: The reaction wheel pendulum
Therefore, the total kinetic energy is given by
(1.1)
The
potential energy of the system is
.
Finally, the Lagrangian function is given by
(1.2)
Using Euler-Lagrange's equations we therefore have
(1.3)
The dynamic equations of the system are finally given by
(1.4)
(1.5)
In compact form, the system can be rewritten as follows
(1.6)
where
is the vector of generalized coordinates,
is the vector of joint torques,
is the inertia matrix and is given by
(1.7)
and
(1.8)
Note that the matrix is constant and positive definite. The equations of motion are also given by
(1.9)
(1.10)
Passivity properties of the system
The total energy of the reaction wheel pendulum is
(1.11)
From (1.6) - (1.8) and (1.11), it follows that
(1.12)
As
a consequence, the system with
as input and
as output is passive. The reaction wheel pendulum with zero control
input
has an unstable equilibrium at
with energy
and a stable equilibrium at
with energy
.
The disk position
can be arbitrary, since the energy in (1.11) does not depend on
,
i.e.
is a cyclic variable. Hence, the equilibrium points described above
are not isolated equilibrium points in the four-dimensional state
space of the system.
The
control objective will be to control the pendulum position
,
the pendulum velocity
and the disk velocity
and to leave the disk position
unspecified.
Let
us consider the state vector
.
We will bring the state vector
to
.
Linearization of the system
In this section, we will study the controllability of the linearized system about the origin. The equations of motion are given by (see (1.9) and (1.10))
(1.13)
(1.14)
where
.
Let
us consider the vector state
.
Differentiating equations (1.13) and (1.14) with respect to the
states and evaluating them at the origin leads to the following
linear system
We then have
and
.
The linearized system is controllable. Therefore, a full state
feedback controller
with an appropriate gain vector
is able to successfully stabilize the system in a neighborhood of the
origin.
