Hindawi Publishing Corporation

Journal of Control Science and Engineering

Volume 2013, Article ID 657182, 15 pages
http://dx.doi.org/10.1155/2013/657182

Research Article
Design of Attitude Control Systems for
CubeSat-Class Nanosatellite

Junquan Li, Mark Post, Thomas Wright, and Regina Lee

Department of Earth & Space Science and Engineering, York University, 4700 Keele Street, Toronto, ON, Canada M3J 1P3

Correspondence should be addressed to Junquan Li; [email protected]

Received 18 December 2012; Accepted 24 April 2013

Academic Editor: Sabri Cetinkunt

Copyright 2013 Junquan Li et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We present a satellite attitude control system design using low-cost hardware and software for a 1U CubeSat. The attitude control
system architecture is a crucial subsystem for any satellite mission since precise pointing is often required to meet mission objectives.
The accuracy and precision requirements are even more challenging for small satellites where limited volume, mass, and power are
available for the attitude control system hardware. In this proposed embedded attitude control system design for a 1U CubeSat,
pointing is obtained through a two-stage approach involving coarse and fine control modes. Fine control is achieved through the
use of three reaction wheels or three magnetorquers and one reaction wheel along the pitch axis. Significant design work has been
conducted to realize the proposed architecture. In this paper, we present an overview of the embedded attitude control system
design; the verification results from numerical simulation studies to demonstrate the performance of a CubeSat-class nanosatellite;
and a series of air-bearing verification tests on nanosatellite attitude control system hardware that compares the performance of the
proposed nonlinear controller with a proportional-integral-derivative controller.

1. Introduction suitable actuators like magnetorquers (torque coils or torque

rods) and small-sized reaction wheels [5, 6]. An overview of
The development of nanosatellites (with a mass of 110 kg) the proposed ACS design adopted for this study is shown in
is currently a significant trend in the area of space science Figure 2. Currently, commercial nanosatellite torque rods and
and engineering research. The development of CubeSat-class reaction wheels are too expensive for use in many research
nanosatellites started in 1999 as a collaborative effort between nanosatellite projects. The contributions of this research
California Polytechnic State University and Stanford Univer- are the development of ACS hardware from off-the-shelf
sity and has achieved great success as a way to efficiently components, complete simulation of the ACS system, and
construct and orbit small, inexpensive satellites using com- validation testing of the ACS system for attitude control in
mercial technology. CubeSat, in general, is described as a class the lab environment.
of nanosatellites ranging from 1 kg, 10 10 10 cm3 , and In this paper, we first briefly describe the ACS hard-
upwards in 10 cm increments of length. Currently more than ware proposed for CubeSat-class nanosatellite missions. We
50 research groups around the world are developing CubeSat- outline the hardware development of the ACS actuators in
class nanosatellites for technology demonstration and scien- Section 2. In particular, we describe the sizing and design of
tific and student training missions. A solid model of a typical the magnetorquers. More discussions on the ACS hardware
1U CubeSat with attitude control systems (ACS) is shown in selection, design, and characterization for CubeSat-class
Figure 1. nanosatellite missions, currently under development at York
Full-scale satellite attitude control systems are generally University, can be also found in [711]. In Sections 3 and 4,
too large or too expensive to be installed in CubeSat-class we describe the satellite system models and the results from
nanosatellites [1], so passive attitude control systems have the simulation study based on this design. In Section 5, we
usually been used for nanosatellites in the past [2, 3]. More show the ground testing results of the hardware and software
active attitude control subsystems [4] in CubeSat-class nano- system. Section 6 includes future work that is planned.
satellites have been implemented with the development of Section 7 concludes the paper.
2 Journal of Control Science and Engineering

Reaction wheel The power and mass constraints were applied using (1),
and a power value was estimated as the lesser of 0.2 W and the
power dissipation was achieved at the maximum current for
the wire
2
Magnetorquer 2 = ,

4 [ln ( / ) 1]
= 2
, (2)
( / ) 4 ln ( / )

1
= [1 + ].
2 1 + ( 1)
Figure 1: ACS (CubeSat-class nanosatellite with three reaction
wheels and three torque rods). Using (2) from the well-known solenoid equation and
the relations derived in [12], where is the demagnetizing
factor, a parametric analysis of the effect of core length and
2. CubeSat ACS Hardware core radius on magnetic dipole moment, , was used to
determine the optimal length and radius of the core material,
2.1. Attitude Sensors. Attitude magnetic sensor hardware in given the wire thickness and corresponding length to satisfy
the current study consists of a Honeywell HMC5883L three- the power and dipole moment requirements. Figure 3 illus-
axis MEMS magnetometer for magnetic field measurements. trates the effect of core sizing on generated magnetic moment.
Angular rate information is obtained in three axes from three The number of turns is implicitly determined, as the dipole
orthogonally mounted Analog Devices ADXRS614 MEMS moment of the torque rod is independent of the number of
gyroscopes. The attitude control system is managed by an turns of wire, and the core radius is also constrained to sizes
AT91SAM9260 32-bit ARM9 microcontroller that runs that are commercially available.
embedded Linux, with 32 MB SRAM, and 256 MB NAND To maximize the field generated with the dimensions
Flash attached for volatile and nonvolatile storage. All pro- while satisfying the design constraints, a prototype magnetor-
gramming of control algorithms is accomplished in the C quer was designed with 70 mm long permalloy core and
language using the GNU C compiler for the ARM processor. 36 AWG wire to provide a maximum load power of 200 mW
The system is designed for power-efficient operation because at 4.2 V. The design parameters of the constructed torque rods
there is typically less than 3 W generated from the photovol- are shown in Table 1.
taics on a typical 1U CubeSat in low-earth orbit, and a battery In order to precisely wind 36 AWG wires around a core,
must be used during eclipse periods. a coil winding machine was designed and assembled using
stepper motors and L298 H-bridge drivers controlled by an
2.2. Magnetorquer Design. Magnetic torque coils, also ATMega644P microcontroller. A permalloy core is rotated by
referred to as magnetorquers, in CubeSat-class nanosatellites one motor, while another positions a plastic feeder guide from
provide baseline control in many small satellites. They are the wire spool. The winder, shown with a completed torque
commercially available in two typical configurations: in loose rod in Figure 4, allows a torque rod to be automatically wound
coils of flat-wound wire and in tightly wound coils around by setting the number of turns required, length of the core,
a permalloy rod. The rod configuration is often preferred and thickness of the wire, which determines the ratio of wind-
because of its compactness and rigidity and the use of high- ing speed to feeder speed.
permeability, , materials for the core. To meet the mass and
power requirements of a nanosatellite, a maximum mass, , 2.3. Reaction Wheel Design. Pointing and slew maneuvering
of 30 g and a conservative maximum power draw, , of 0.2 W of satellites are often accomplished by a motorized rotating
were set, and a typical maximum supply voltage, , between mass such as a reaction wheel and momentum wheel, which
3.7 V and 4.2 V was assumed to avoid having to step up provide maneuvering torque and momentum storage [8].
voltage on power components. After combining the power Reaction wheels can provide a high degree of attitude control
and mass equations (1), where is resistance, is the accuracy with the limitation that the wheel may reach satura-
resistivity of the wire, and is the density, and solving for the tion after continued use, requiring an additional momentum
core radius, , and length, , control method such as magnetorquers to desaturate the
wheel in a process known as momentum dumping.
2 Each of the reaction wheels in the proposed ACS system
= , consists of a steel cylinder that is press-fitted to the shaft of a

Faulhaber brushless flat micromotor. Design choices for the
(1) motor were limited to inexpensive commercial motors with
= ,
low-power consumption, and the reaction wheels were sized
to provide maximum momentum storage given the mass and
= 2 + 2 . volume constraints of a 1U CubeSat. Three reaction wheels
Journal of Control Science and Engineering 3

Reaction wheel
drivers

Torque rod
Magnetometer Microcontroller
drivers
(HMC5883L) (AT91SAM9G20)
(BD6212)

Coarse sun
Rate sensors sensors
(ADXRS614) ADC
(TSL1402)

Figure 2: Overview of proposed ACS design for a CubeSat-class nanosatellite.

Moment for wire radius of 0.11 mm Table 2: Reaction wheel parameters.

4 Parameter Value Unit

Magnetic moment

Rotor mass 0.214 kg

3 Moment of inertia (axial) 9.41 105 Kg m2
2 Moment of inertia (transverse) 5.02 105 kg m2
: 0.0075 Motor shaft Torque 6.0 104 Nm
1 : 0.11
: 1.302 Maximum speed 1539 rad/s
0 Supply voltage 3.74.2 V
0.2

0.15
Co

ACS testing. The board makes a HMC5883 three-axis mag-

re

0.1
l

netometer, an ADXL345 3-axis accelerometer, and an ITG-

en

8 3
gth

7
0.05 4 5 6 10 3200 3-axis MEMS rate gyroscope available on the OBC I2 C
2 3 d ius
Core ra bus. Primary rate sensing for attitude control is accomplished
Figure 3: Magnetorquer sizing surface. by three independent ADXRS614 rate gyro units oriented
on orthogonal axes by means of right-angle IC sockets and
connected to the first three ADC channels on the OBC. This
Table 1: Magnetorquer parameters. allows accurate high-speed sampling of rotation rates for use
by the attitude controller. To drive the magnetorquers, three
Parameter Value Unit
BD6212 integrated H-bridges are used, controlled by three
Maximum dipole moment 0.37 A m2 PWM channels from the OBC and three general purpose IO
Total mass 28 g pins for current direction control. To allow one PWM signal
Number of turns 6063
and one direction pin to control each H-bridge, the inputs are
Core diameter 5.7 mm
demultiplexed by a SN74LVC1G8 tristate output demulti-
Wire diameter 0.127 mm
plexer and pull-up resistors. In full operation, the board
Wire resistance 121
draws up to 100 mW of power on average, though compo-
Maximum current 34.7 mA
nents can be shut down as needed to conserve power if not in
use.
can be used in the ACS if maximum control authority is
required. Table 2 shows the design parameters of the reaction 3. System Models
wheels used on the proposed ACS [13]. A completed reaction
wheel assembly is shown in Figure 5. 3.1. Attitude Equations of Motions. In this section, the satellite
is modelled as a rigid body with actuators that provide tor-
ques about three mutually perpendicular axes that defines a
2.4. Electronic Integration of ACS Components. To control the body-fixed frame. The equations of motion [14, 15] are given
reaction wheel and magnetorquers hardware and house the by
attitude sensors and actuator drivers, a printed circuit board
(PCB) was fabricated, shown in Figure 6. The board stacks = ( + ) + + + , (3)
with existing PC/104 sized on-board computer (OBC) hard-
ware and provides both I/O breakout and power supplies where = (1 2 3 ) is the angular velocity of the satellite
for the ACS hardware. It contains 3.3 V and 5 V switching expressed in the body frame. is the inertia matrix of the
supplies for the ACS sensors, as well as external interfaces for satellite. is the inertia matrix of the reaction wheel and
a battery and radio to be used specifically for air-bearing = . is the layout matrix of the reaction
4 Journal of Control Science and Engineering

Figure 4: Magnetorquer winding apparatus and completed torque rod.

Coil drive bridges

Magnetometer

MEMS rate gyros

3.3 V/5 V supplies

Figure 5: Reaction wheel assembly.

wheels whose columns represent the influence of each wheel Figure 6: ACS sensor and actuator board.
on the angular acceleration of the satellite. is the velocity of
a reaction wheel, is the torque control provided by the reac- In terms of Euler angles, we can also express the satellite
tion wheel, is the torque control provided by the magne- attitude as
torquers, and is the bounded external disturbance, which 1 sin () tan () cos () tan ()
is a sum of the gravity gradient gravity , aerodynamic aero , and [ ]
[ sin () ]
solar radiation pressure solar disturbances. [ ] = [0 cos () ] , (5)
The gravity gradient disturbance is gravity = [ ]
[ cos () ]
2 [ ] 0 sin ( )
3 /3 , where is the gravitational parameter of the [ cos () cos () ]
Earth, is the semimajor axis of the orbit, and is the
where is the roll angle about the -axis, is the pitch angle
direction cosine matrix in terms of quaternions.
about the -axis, and is the yaw about the -axis.
The aerodynamic disturbance is aero = /2V2 ,
where is the coefficient of drag for a flat plate, is the
cross-sectional area causing aerodynamic drag, V is the satel- 3.3. Sensor Models. Magnetic field vectors are obtained in the
lite velocity, is the distance between the centre of pressure orbit reference frame
and the centre of gravity, and is the atmospheric density
related to the altitude. 1 =
03
The solar radiation pressure disturbance is solar =
/ (1 + ), where is the solar constant at the Earths [cos (0 ) (cos () sin () sin () cos () cos ( ))
orbital distance from the Sun, is the speed of light, is the
illuminated surface area, and is the surface reflectance. sin (0 ) sin () sin ( )] ,

2 = [(cos () cos () + sin () sin () cos ( ))] ,
3.2. Attitude Kinematics. The satellite attitude kinematics is 03
represented using quaternions 3
3 = 3
0
1 4 33 + 1
= ( ) () , (4) [sin (0 ) (cos () sin () sin () cos () cos ( ))
2 2
2 sin (0 ) sin () sin ( )] ,
(6)
where = ( , 4 ) = (1 , 2 , 3 , 4 ) .
Journal of Control Science and Engineering 5

where 0 is the angular velocity of the orbit with respect to the the combined actuators has been compared with three reac-
inertial frame, 0 is the distance from the center of the satellite tion wheels-based attitude control in the references [18, 19].
to the center of the Earth, is the orbit inclination, is the Classical sliding mode control has also been used for mag-
magnetic dipole tilt, is the spin rate of the Earth, and is netic actuated spacecraft [20, 21]. However, the proposed
the magnetic dipole moment of the Earth. The magnetometer nonlinear adaptive fuzzy sliding mode control law has never
model is been used in magnetic attitude control.
To address the attitude tracking problem, the attitude
1
= [2 ] + + , (7) tracking error = ( , 4 ) is defined as the relative orienta-
[3 ]
tion between the body frame and the desired frame with

orientation = ( , 4 ) . In order to apply the proposed
where is the direction cosine matrix in terms of quater-
nonlinear controller, the equations of motion are rewritten as
nions, is the zero mean Gaussian white noise of the
magnetometer, is the vector formed with the components of = ( , ) + + + . (10)
the Earths magnetic field in the body frame of the reference,
1
and = [ 2 ]. is the magnetometer bias. The adaptive fuzzy sliding mode magnetic control law is
3 given by
The angular velocity is measured from three rate gyro-
scopes. The model is given by 3
= 1 2 tanh ( ), (11)
= + + ,
(8) = , (12)
= + ,
ap = 2 , (13)

where is the output of a gyroscope and is the real angular
rate of the gyro. and are Gaussian white noise. is the
= ap 2 . (14)
random drift of the gyro and is the drift constant.
Here, = are the torques generated by the magne-
3.4. Actuator Models. Reaction wheels are widely used to per-
torquers, is the vector of magnetic dipoles for the three
form precise satellite attitude maneuvers because they allow
magnetorquers, and is the vector formed with the compo-
continuous and smooth control of internal torques. Torques
nents of the Earths magnetic field in the body frame of the
are produced on the satellite by accelerating or decelerating
reference. = + is the sliding surface. and are the
the reaction wheels. Let the torque demanded by the satellite
adaptive parameters and fuzzy weight functions generated
be denoted as , where = ( + ).
The input voltage
by the fuzzy logic systems [22], and , 1 , 2 , , , are
required to control the actuator dynamics of the reaction
positive constants used for tuning the control response.
wheel can be written as
= 1 ( ) , (9) Remark 1. AFSMC controller design details can be found
in the authors previous papers [11]. The design includes: (1)
where is the motor torque constant, is the back-EMF Sliding surface design [22] and (2) fuzzy logic system design
constant, is the armature resistance, and friction in the [22].
reaction wheels is ignored. The maximum voltage of the reac-
tion wheel is 4.2 V, and a dead zone for the reaction wheel is
estimated to be below 1 V. is 0.0082, is 0.007, is 0.5, and 4.2. Simulation Results. The attitude detumbling and attitude
the moment of inertia of the reaction wheel is 0.0001 kg m2 . stabilization phases are considered in the ACS simulation.
The B-dot algorithm, PD magnetic control law, and adaptive
fuzzy sliding mode magnetic control law are used for the
4. Control Law Design and Simulation Results two phases, respectively. We note that the orbit used for the
present simulation study is a 500 km circular orbit with 45
4.1. Satellite Attitude Control Laws. Magnetic control has inclination. At this altitude, the total disturbance torque for
been used over many years [16, 17] for small spacecraft atti- 1U CubeSats is estimated to be on the order of 5 107 Nm.
tude control. The main drawback of magnetic control is that This is intended to be a slight overestimation to include a
magnetic torque is two-dimensional and it is only present in safety margin.
the plane perpendicular to the magnetic field vector [18]. The
accuracy of satellite attitude control systems (ACS) using only 4.2.1. Detumbling Mode and Stabilization Mode
magnetic actuators is known to be accurate on the order of
0.40.5 degree [18]. The satellite cannot be controlled pre- Scenario 1. In the initial stage of ACS control, the angular
cisely in three-dimensional space using only magnetorquers velocities of the satellite are assumed to be 0.169 rad/s as a
[18], but the combination of magnetorquers with one reaction result of separation from the launch vehicle. The ACS damps
wheel expands the two-dimensional control torque possi- the angular rate by controlling three magnetorquers. The
bilities to be three-dimensional. The attitude accuracy of control logic generally used for detumbling is called B-dot
6 Journal of Control Science and Engineering

Magnetic dipole: 0.1 Am2 B-dot control Magnetic dipole: 0.1 Am2 B-dot control
200 106
200
2
150 100
0 1
100

Magnetic torque (Nm)

100
0
50 200
0 0.1 0.2 106
0 1 2

50 2 0

3 2
100
4
150 4
0 0.1 0.2
200 5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (orbits) Time (orbits)

Roll (deg) 1
Pitch (deg) 2
Yaw (deg) 3

(a) (b)
Magnetic dipole: 0.1 Am2 B-dot control
0.1
2
Magnetic dipole: 0.1 Am B-dot control
0.18 0.08
0.16 0.06
Angular velocity tracking error (rad/s)

0.14 0.2 0.04

Magnetic dipole (Am2 )

0.12
0.15 0.02
0.1
0.1 0
0.08 0.1
0.05 0.02
0.06 0.05
0 0.04 0
0.04
0.02 0.05 0.06 0.05
0 0.05 0.1 0.15 0.2
0 0.08 0.1
0 0.1 0.2
0.02 0.1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.5 1 1.5 2
Time (orbits) Time (orbits)

(c) (d)

Figure 7: Scenario 1 detumbling control results.

control [1], as it makes use of the derivative of the magnetic after 1 orbit, the angular velocities are reduced to the required
field . For a CubeSat with moment of inertia = rates before continuing with other ACS tasks.
diag(0.002, 0.002, 0.002) kg m2 , we include the external dis-
turbances (aerodynamic, gravity gradient, and solar pres- Scenario 2. Now, we consider a CubeSat with the same
sure), set the desired quaternion to be (0, 0, 0, 1), set the moment of inertia and orbit and assume the magnetic dipole
initial quaternion to be (0.1, 0.1, 0.1, 0.9849), and assume maximum of the magnetorquer to be 0.4 Am2 with a magne-
the magnetic dipole maximum of the rods to be 0.1 Am2 . The tometer sensor bias calculated by 20 104 rand(1). Pro-
Euler angle tracking errors, angular velocity tracking errors, portional-derivative (PD) magnetic control [23] laws (shown
and magnetic torquers, magnetic dipoles results are shown in in (15)) are used in this simulation and the results over 6 orbits
Figure 7. The satellite starts at the selected tip-off rate, and, are shown in Figure 8. During the first three orbits, three
Journal of Control Science and Engineering 7

Roll-pitch-yaw Angular velocity

200 0.18
0.02

Angular velocity max (0.169 rad/s = 10 deg/s)

150 0.16 0.2
0.01
0.14
100 0 0.1
0.01 0.12
50 0 3-axis pure magnetic control
0.02 0.1 0 0.005 0.01
4 5 6
0 0.08
0.01
Detumbling mode: B-dot
50 0.06 0
3-axis pure magnetic control Detumbling mode: B-dot
0.04 0.01
100
0.02 0.02
3 3.02 3.04
150
0
200 0.02
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (orbits) Time (orbits)

(a) (b)
4
Magnetic moment 10 Body frame
0.4 3

0.3 Detumbling mode: B-dot

2.5 3-axis pure magnetic control
Magnetic field in body frame (T)

0.2
Magnetic dipole (Am2 )

Detumbling mode: B-dot 2

0.1
3-axis pure magnetic control
1.5
0
0.4
104
0.1 0.2 1
2
0
0.2 1 0.5
0.2 0 1
0.3 0.4 1 0
0 0.005 0.01 5.6 5.8 6
0.4 0.5
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (orbits) Time (orbits)
1
2
3
(c) (d)

Figure 8: Scenario 2 detumbling mode and stable mode control results.

magnetorquers are used for the detumbling mode. In the slew angle target for limb and nadir pointing. The configura-
second three orbits, three magnetorquers and one reaction tion with three magnetorquers and one flywheel [24] has been
wheel are used for the stable mode. The attitude control used for many years. In a real nanosatellite mission, hardware
accuracy is less than 0.02 degree while using the PD magnetic failures of the reaction wheels are very common [19]. When
control laws there are two wheel failures, the ACS can be switched from
using three reaction wheels to three magnetorquers and one
= , reaction wheel, and attitude control accuracy maintained
= 1 , (15) using this method. The adaptive fuzzy sliding mode magnetic
control laws are shown in (11)(14). The attitude control
= 1 + 2 .
accuracy using the nonlinear adaptive fuzzy sliding mode
control law will be more robust to the external disturbances
4.2.2. Attitude Stabilization Mode: Nadir and Limb Pointing than using the PD magnetic control law in (15).
with Three Magnetorquers and One Pitch Reaction Wheel.
Next, we consider the second stage of nanosatellite control Scenario 3. An ACS using one reaction wheel and three
with a low initial angular velocity after detumbling and large magnetorquers as actuators for limb pointing with the desired
8 Journal of Control Science and Engineering

Roll-pitch-yaw tracking errors Magnetic moment x-axis

80 0.3
70 0.25
60
0.2
50
0.15

1 (Am2 )
40
30 0.1
20 0.05
10
0
0
10 0.05
20 0.1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (orbits) Time (orbits)

(a) (b)
Magnetic moment y-axis Magnetic moment z-axis
0.3 0.05

0.2 0

0.1 0.05
3 (Am2 )
0.1
2 (Am2 )

0.1 0.15

0.2 0.2

0.3 0.25

0.4 0.3
0 2 4 6 8 10 12 0 0.5 1 1.5 2 2.5 3 3.5
Time (orbits) Time (orbits)

(c) (d)
100 0.8
0.6
80
0.4
Wheel speed (rad/s)

60
Wheel voltage (V)

0.2
40 0

20 0.2
0.4
0
0.6
20 0.8
40 1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (orbits) Time (orbits)

(e) (f)

Figure 9: Scenario 3 limb pointing control results using one reaction wheel and three magnetorquers.

quaternion set to (0, 0.5736, 0, 0.8192) is examined using magnetorquers, attitude tracking errors, wheel speed, and
AFSMC over 10 orbits. The wheel dead zone is not consid- wheel voltage are shown in Figure 9.
ered here. The initial quaternion is (0.1, 0.1, 0.1, 0.9849)
and initial angular velocity is (0.0169, 0.0169, 0.0169) rad/s. Scenario 4. An ACS using one reaction wheel and three mag-
Considering only the pitch reaction wheel is available, the netorquers as actuators for nadir pointing with the desired
numerical simulations demonstrate that the proposed tech- quaternion that is set to (0, 0, 0, 1) is examined using AFSMC
nique achieves a high pointing accuracy (<0.09 degree) over 10 orbits. The wheel dead zone is assumed to be 1.0 V.
for small satellites. The magnetic dipoles from the three The initial quaternion is (0.1, 0.1, 0.1, 0.9849) and initial
Journal of Control Science and Engineering 9

Roll-pitch-yaw tracking errors Magnetic moment x-axis

40 0.3

30 0.25

0.2
20

1 (Am2 )
0.15
10
0.1
0
0.05
10 0

20 0.05
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (orbits) Time (orbits)

(a) (b)
Magnetic moment y-axis Magnetic moment z-axis
0.3 0.05

0.2 0

0.1 0.05
2 (Am2 )

0.1
3 (Am2 )
0

0.1 0.15

0.2 0.2

0.3 0.25

0.4 0.3
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (orbits) Time (orbits)

(c) (d)
0 0.1
0.2
10
0.3
Wheel speed (rad/s)

Wheel voltage (V)

20 0.4
0.5
30
0.6
40 0.7
0.8
50
0.9
60 1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (orbits) Time (orbits)

(e) (f)

Figure 10: Scenario 4 nadir pointing control results using one reaction wheel and three magnetorquers.

angular velocity is (0.0169, 0.0169, 0.0169) rad/s. The pitch 4.2.3. Attitude Stabilization Mode: Nadir Pointing with
reaction wheel and three magnetorquers are used. These sim- Three Reaction Wheels
ulations demonstrate that the proposed technique can also
achieve high accuracy (<0.09 degree with all disturbances) Scenario 5. An ACS using three reaction wheels as actua-
pointing control for small satellites. The magnetic dipoles tors for nadir pointing with the desired quaternion set to
of the three magnetorquers, attitude tracking errors, wheel (0, 0, 0, 1) is examined using AFSMC over 0.1 orbits. The
speed, and wheel voltage are shown in Figure 10. It takes about wheel dead zone is assumed to be 1.0 V. The initial quater-
40 orbits for the wheel speed to increase from 0 to 2000 rpm nion is (0.1, 0.1, 0.1, 0.9849) and initial angular velocity
(209 rad/s). is (0.0169, 0.0169, 0.0169) rad/s. This configuration can also
10 Journal of Control Science and Engineering

Roll-pitch-yaw tracking error with three reaction wheels

15

15
10
10 5
0
5
10
5
15
0 0.2 0.4 0.6 0.8 1 3
10
Angle (deg)

0
103
8
6
5 4
2
0
10 2
0.13 0.132 0.134 0.136 0.138 0.14

15
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (orbits)

Figure 11: Scenario 5 nadir pointing control results using three wheels: Euler angle tracking errors.

103 Control input with AFSMC 5. Satellite Attitude Control System

8 Hardware Testing
6
5.1. Spherical Air-Bearing Testbed for Satellite Attitude Control
4 Systems. The ground testing of the proposed CubeSat ACS
design was performed at York University using a nanosatellite
Controller (Nm)

2 attitude control testbed. This facility consists of a spherical

0
air-bearing platform [25, 26] suspended upon a thin layer of
air providing a full three degrees of freedom with negligible
2 friction for ACS testing. The platform includes a manual
balancing system and platform electronics that include an
4 on-board computer (OBC), wireless transceiver for telemetry,
reference inertial measurement unit (IMU), power distribu-
6
tion board, and batteries. The air-bearing system used for 1U
8 CubeSat testing is shown in Figure 15.
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (orbits)
5.2. Ground Test Results: ACS Testing Results with Three Reac-
Figure 12: Scenario 5 nadir pointing control results using three tion Wheel Actuators. Nonlinear attitude control has been
wheels: control input. explored widely in theory [14, 27, 28]. In real satellite appli-
cations, nonlinear controllers are usually not selected due to
their complexity of design. We have tested nonlinear control
algorithms [11] with our spherical air-bearing system. We now
achieve high attitude control accuracy (<0.008 degree with all test the proposed control method (from (11) and (12)) on this
disturbances). The reaction wheel attitude control laws use system using the sensors and ACS board described above and
(11) and (12). The settling time is shorter and the pointing three-axis reaction wheel actuation. A PID control law is also
accuracy is higher than that achieved using only three mag- used for comparison on the same hardware. The control laws
netorquers and one reaction wheel as actuators. However, are programmed in the C language, and the OBC runs the
the power consumption is higher than using three magnetor- Linux operating system. More details of the implementation
quers and one reaction wheel. The Euler angle tracking errors, can be also found in [11]. Here, we will show air-bearing
control inputs, wheel speeds, and wheel voltages are shown in system test results with a 1U CubeSat. The design parameters
Figures 11, 12, 13, and 14. of the control laws used are given in Tables 3 and 4.
Journal of Control Science and Engineering 11

Wheel speed with AFSMC

600

400

Wheel speed (rad/s)

200

200

400

600
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (orbits)

Figure 13: Scenario 5 nadir pointing control results using three wheels: wheel speed.

Wheel voltages with AFSMC

4

1
Wheel voltage (V)

5
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (orbits)

Figure 14: Scenario 5 nadir pointing control results using three wheels: wheel voltage.

Figure 15: Air-bearing ACS ground testing system.

12 Journal of Control Science and Engineering

AFSMC tracking error AFSMC control input

100 2

80 1

Roll-pitch-yaw-control
0
60
Error (deg)

1
40
2
20
3

0
4

20 5
0 50 100 150 200 0 50 100 150 200
Time (s) Time (s)

Roll Roll
Pitch Pitch
Yaw Yaw
(a) (b)
AFSMC wheel voltage AFSMC wheel speed
3 1500

2 1000

500
1
Wheel speed (rad/s)

0
Wheel voltage (V)

0
500
1
1000
2
1500
3 2000
4 2500

5 3000
0 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 160 180 200
Time (s) Time (s)

Roll Roll
Pitch Pitch
Yaw Yaw
(c) (d)

Figure 16: Air-bearing AFSMC controller results.

Table 3: PID controller parameters for air bearing testing. Table 4: AFSMC controller parameters for air bearing testing.

Name Values Name Values

Proportional parameters 0.03 Sliding surface gain 0.00001, 100
Integral parameters 0.0001 Fuzzy membership function 1, 1
Derivative parameters 0.11 Adaptive gains , , , 0.1, 0.36, 0.01, 2
AFSMC parameter 1 0.004
AFSMC parameter 2 0.0025
Figures 16 and 17 show the attitude tracking errors,
AFSMC/PID control output signals, reaction wheel voltages,
and reaction wheel velocities for a 90 degree yaw slew of the and pitch. Due to the difficulty of perfectly balancing the
system about the -axis while maintaining 0 degrees of roll air-bearing system, a gravitational disturbance, larger than
Journal of Control Science and Engineering 13

PID tracking error PID control input

100 1

80 0.5

60 0

Torque fraction
Error (deg)

40 0.5

20 1

0 1.5

20 2
0 50 100 150 200 0 50 100 150 200
Time (s) Time (s)

Roll Roll
Pitch Pitch
Yaw Yaw
(a) (b)
PID wheel voltage PID wheel speed
2 500

1 0

0
Wheel voltage (V)

500
Wheel speed (rad/s)

1
1000
2
1500
3

4 2000

5 2500
0 50 100 150 0 50 100 150
Time (s) Time (s)

Roll Roll
Pitch Pitch
Yaw Yaw
(c) (d)

Figure 17: Air-bearing PID controller results.

normal in size, is considered to be present about the and to predict the performance of pure and hybrid magnetic
rotational axes. Compared to the PID controller, the AFSMC control as the Earths magnetic field present in the test facility
controller uses nearly the same gain and has much better is constant and invariant with many sources of additional
tracking performance under these conditions. magnetic noise. The next step in the development of a
complete ACS test facility is the addition of a Helmholtz cage
for magnetic control tests. The magnetic cage and the air-
6. Future Work bearing system in the lab are shown in Figure 18 and are
presently being prepared for air-bearing magnetic control
While the ground test demonstrates the effective and promis- testing. Future work will demonstrate the effectiveness of
ing control accuracy of the proposed ACS design, it is difficult magnetic control in further ground testing.
14 Journal of Control Science and Engineering

[4] G. P. Candini, F. Piergentili, and F. Santoni, Miniaturized atti-

tude control system for nanosatellites, Acta Astronautica, vol.
81, pp. 325334, 2005.
[5] T. Xiang, T. Meng, H. Wang, K. Han, and Z. H. Jin, Design and
on-orbit performance of the attitude determination and control
system for the ZDPS-1A pico-satellite, Acta Astronautica, vol.
77, pp. 82196, 2012.
[6] M. Abdekrahman and S. Y. Park, Integrated attitude determi-
nation and control system via magnetic measurements and
actuation, Acta Astronautica, vol. 69, no. 3-4, pp. 168185, 2011.
[7] M. Cannata, Development of a sun vector determination algo-
rithm for Cubesat-Class spacecraft [M.S. thesis], Dept. of Earth
and Space Science and Engineering, York University, Toronto,
Canada, 2010.
[8] I. Proper, Reaction wheel design, construction and qualification
testing [M.S. thesis], Dept. of Earth and Space Science and
Engineering, York University, Toronto, Canada, 2010.
[9] J. Li, M. A. Post, and R. Lee, Real time fault tolerant nonlinear
Figure 18: Helmholtz cage for magnetic field generation. attitude control system for nanosatellite application, in AIAA
Infotech@Aerospace 2012 Conference, pp. 1921, Garden Grove,
Calif, USA, 2012.
[10] J. Li, M. A. Post, and R. Lee, Nanosatellite air bearing tests
7. Conclusion of fault-tolerant sliding-mode attitude control with Unscented
Kalman Filter, in AIAA Guidance, Navigation, and Control
In this paper, the proposed attitude control system design will Conference, Minnesota, Minneapolis, Minn, USA, 2012.
allow inexpensive and capable satellites to be developed for
[11] J. Li, M. A. Post, and R. Lee, Nanosatellite attitude air bearing
academic use. The proposed attitude control system uses
system using variable structure control, in IEEE 25th Annual
three reaction wheels as actuators or one wheel with three Canadian Conference on Electrical and Computer Engineering,
magnetorquers as actuators. The hardware designs of the Montreal, Canada, May 2012.
actuators and embedded attitude control systems are also
[12] M. F. Mehrjardi and M. Mirshams, Design and manufacturing
described for prototype development. Five different scenarios
of a research magnetic torquer Rod, Contemporary Engineering
of numerical simulation results for a 1U CubeSat-class nano- Sciences, vol. 3, no. 5, pp. 227236, 2010.
satellite show the effectiveness of the proposed attitude con-
[13] T. Ustrzycki, Spherical air bearing testbed for nanosatellite atti-
trol design for detumbling and attitude pointing purposes.
tude control development [M.S. thesis], Dept. of Earth and Space
The ground testing results in this paper provide a promising
Science and Engineering, York University,, Toronto, Canada,
comparison of an adaptive fuzzy sliding mode controller with 2011.
a conventional proportional-integral-derivative controller for
[14] C. H. Won, Comparative study of various control methods for
a 1U CubeSat-class nanosatellite. The nanosatellite testing
attitude control of a LEO satellite, Aerospace Science and Tech-
hardware can be also extended for use in 2U or 3U CubeSat- nology, vol. 5, pp. 323333, 1999.
class nanosatellites in the future.
[15] J. Y. Lin, S. Ko, and C. K. Ryoo, Fault tolerant control of
satellites with four reaction wheels, Control Engineering Prac-
Acknowledgments tice, vol. 16, no. 10, pp. 12501258, 2008.
[16] E. Silani and M. Lovera, Magnetic spacecraft attitude control: a
The authors gratefully acknowledge the support provided by survey and some new results, Control Engineering Practice, vol.
COM DEV Ltd., NSERC, MITACS, and OCE. The authors 13, no. 3, pp. 357371, 2005.
would also like to acknowledge the work of M. Cannata, I. [17] C. J. Dameran, Hybrid magnetic attitude control gain selec-
Proper, T. Ustrzycki, G. Benari, and H. Hakima in this paper. tion, Proceedings of the Institution of Mechanical Engineers G,
vol. 223, no. 8, pp. 10411047, 2009.
References [18] Z. Q. Zhou, Spacecraft attitude tracking and maneuver using
combined magnetic actuators, in AIAA Guidance, Navigation,
[1] Y. W. Jan and J. C. Chiou, Attitude control system for ROCSAT- and Control Conference, Toronto, Ontario, Canada, August
3 microsatellite: a conceptual design, Acta Astronautica, vol. 56, 2010.
no. 4, pp. 439452, 2005. [19] J. R. Forbes and C. J. Damaren, Geometric approach to space-
[2] M. Ovchinnikov, V. Penkov, O. Norberg, and S. Barabash, craft attitude control using magnetic and mechanical actuation,
Attitude control system for the first Swedish nanosatellite Journal of Guidance, Control, and Dynamics, vol. 33, no. 2, pp.
MUNIN, Acta Astronautica, vol. 46, no. 2, pp. 319326, 2000. 590595, 2010.
[3] M. I. Martinelli and R. S. S. Pena, Passive 3 axis attitude control [20] R. Wisniewski, Satellite attitude control using magnetic actuation
of MSU-1 pico-satellite, Acta Astronautica, vol. 56, no. 5, pp. only [Ph.D. thesis], Dissertation, Dept. of Control Engineering,
507517, 2005. Aalborg University, Denmark, 1996.
Journal of Control Science and Engineering 15

[21] P. Wang, Y. B. Shtessel, and Y. Q. Wang, Satellite attitude control

using only magnetorquers, in Proceeding of the 13th Southeast-
ern Symposium on System Theory, West Virginia University,
Morgantown, WVa, USA, March 1998.
[22] J. Li and K. D. Kumar, Fault tolerant attitude synchronization
control during formation flying, Journal of Aerospace Engineer-
ing, vol. 24, no. 3, pp. 251263, 2011.
[23] D. V. Guerrant, Design and analysis of fully magnetic control
for picosatellite stabilization [M.S. thesis], California Polytechnic
State University, 2005.
[24] M. Chen, S. J. Zhang, F. R. Liu, and Y. C. Zhang, Combined atti-
tude control of small satellite using One flywheel and magnetic
torquers, in Proceedings of the 2nd International Sympo-
sium on Systems and Control in Aerospace and Astronautics
(ISSCAA 08), Shenzhen, China, December 2008.
[25] J. Prado, G. Bisiacchi, L. Reyes et al., Three-axis air-bearing
based platform for small satellite attitude determination and
control simulation, Journal of Applied Research and Technology,
vol. 3, no. 3, pp. 222237, 2005.
[26] J. J. Kim and B. N. Agrawal, Automatic mass balancing of air-
bearing-based three-axis rotational spacecraft simulator, AIAA
Journal of Guidance, Control, and Dynamics, vol. 32, no. 3, pp.
10051017, 2009.
[27] K. S. Kim and Y. Kim, Robust backstepping control for slew
maneuver using nonlinear tracking function, IEEE Transac-
tions on Control Systems Technology, vol. 11, no. 6, pp. 822829,
2003.
[28] S. C. Lo and Y. P. Chen, Smooth sliding mode control for
spacecraft attitude tracking maneuvers, Journal of Guidance
Control and Dynamics, vol. 18, no. 6, pp. 13451349, 1995.
 International Journal of

Rotating
Machinery

International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of

Hindawi Publishing Corporation

World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design
Advances in
OptoElectronics
International Journal of

International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014

International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014