DESIGN, DYNAMICS AND CONTROL OF A SIX DEGREES OF FREEDOM

SPACECRAFT SIMULATOR WITH INTERNAL ACTUATORS FOR

AUTONOMOUS RENDEZVOUS AND
PROXIMITY OPERATIONS
BY
SASI PRABHAKARAN VISWANATHAN, B.E.
A thesis submitted to the Graduate School
in partial fulllment of the requirements
for the degree
Master of Science in Mechanical Engineering
Advised by:
Dr. Amit K Sanyal
New Mexico State University
Las Cruces New Mexico
November 2012
Design, Dynamics and Control of a Six Degrees of Freedom Spacecraft Simulator
with Internal Actuators for Autonomous Rendezvous and Proximity Operations,
a thesis prepared by Sasi Prabhakaran Viswanathan in partial fulllment of the
requirements for the degree, Master of Science in Mechanical Engineering, has
been approved and accepted by the following:
Linda Lacey
Interim Dean of the Graduate School
Amit K. Sanyal
Chair of the Examining Committee
Date
Committee in charge:
Dr. Amit K. Sanyal, Chair
Dr. Ou Ma
Dr. Frederick Leve
Dr. Michael Engelhardt
ii
The Lord is the Beginning of all,
He is the Atom-within-the-atom;
Divide an atom within the atom,
Into parts one thousand,
They who can thus divide
That atom within the atom
May well near the Lord,
He, indeed, is the Atom-within-the-atom.
Tirumular (8th century AD)
Tantra Seven 28 : 2008 : 2
Tirumantiram
iii
ACKNOWLEDGMENTS
First and foremost I would like to thank my advisor and mentor Dr. Amit
Sanyal for his credibility towards me and giving me this opportunity. Dr. Sanyal
is always supportive, encouraging my ideas, giving lot of freedom and providing
enough room to make mistakes, then guiding me in the correct direction. He our-
ished my pragmatic interest in space system engineering, especially in spacecraft
attitude dynamics and taught the fundamentals strongly. Without his words of
encouragement and moral support, it wouldnt been possible for me to defend my
thesis. I would also like to thank my friends and family for their support.
iv
VITA
July 1, 1985 Born at Tiruchirapalli, Tamil Nadu, India.
August 2004May 2008 B.E., Government College of Technology,
Anna University, India.
July 2008October 2009 Research Associate, Aerospace Engineering,
Indian Institute of Science, Bangalore.
October 2009November 2010 Research Assistant, Mechanical Engineering,
Indian Institute of Technology, Mumbai.
January 2011December 2012 M.S., New Mexico State University,
Las Cruces, New Mexico.
PROFESSIONAL AND HONORARY SOCIETIES
American Institute of Aeronautics and Astronautics
PUBLICATIONS
1. Sasi Prabhakaran V, Amit Sanyal, and Lee Holguin, Dynamics and Control of
a Six Degrees of Freedom Ground Simulator for Autonomous Rendezvous
and Proximity Operation of Spacecraft, AIAA Guidance, Navigation, and
Control Conference, Minnesota, Minneapolis, 2012.
2. Lee Holguin, Sasi Prabhakaran V and Amit Sanyal, Guidance and Control for
Spacecraft Autonomous Chasing and Close Proximity Maneuvers, AIAA
Guidance, Navigation, and Control Conference, Minnesota, Minneapolis,
2012.
3. Amit Sanyal, Lee Holguin, and Sasi Prabhakaran, Guidance and Control for
Spacecraft Autonomous Chasing and Close Proximity Maneuvers, 7th IFAC
Symposium on Robust Control Design (ROCOND12), Aalborg, Denmark,
2012.
v
ABSTRACT
DESIGN, DYNAMICS AND CONTROL OF A SIX DEGREES OF FREEDOM
SPACECRAFT SIMULATOR WITH INTERNAL ACTUATORS FOR
AUTONOMOUS RENDEZVOUS AND PROXIMITY OPERATIONS
BY
SASI PRABHAKARAN VISWANATHAN, B.E.
Master of Science
New Mexico State University
Las Cruces, New Mexico, 2012
Dr. Amit K Sanyal, Chair
Simulating all degrees of freedom of spacecraft motion with internal actuation
in ground is a challenging problem. It is due to several features of the natural
dynamics in space that are dicult to reproduce on ground. Unlike terrestrial
(aerial, land or underwater) vehicles, space vehicles have an overwhelmingly large
percentage of their total energy in their translational motion. Dynamical coupling
between the translational and rotational degrees of freedom can signicantly aect
the overall motion of spacecraft. The attitude motion is particularly important for
a spacecraft tasked to autonomously rendezvous and capture or dock with a target
vi
object in space. Large spacecrafts attitude motion can be actuated using Con-
trol Moment Gyroscopes (CMG) due to the torque amplication characteristics
of CMG. Precise attitude maneuver can also be performed using CMG due to the
reaction-less nature of the device. Here we present a ground simulator design for
6 DOF simulation of spacecraft engaged in autonomous rendezvous and proximity
operation (ARPO) with an unaided target space object, internally actuated by
momentum exchange devices for the attitude control. The attitude dynamics of a
spacecraft with a Variable Speed Control Moment Gyroscopes (VSCMG), in the
presence of conservative external inputs, are derived using variational principles
and generalized to nVSCMG. A complete dynamics model, that relaxes some of
the assumptions made in prior literature on control moment gyroscopes, is ob-
tained. These dynamics equations show the complex nonlinear coupling between
the internal degrees of freedom associated with the CMG and the spacecraft base
bodys attitude degrees of freedom. General ideas on how this coupling can be
used to control the angular momentum of the base body of the spacecraft using
changes in the momentum variables of the VSCMG, are provided. Control of
spacecraft engaged in ARPO are very risky and dicult to carry out in space,
since the targets motion is not well known in advance. Ground simulation using
6 DOF motion simulation capabilities can help reduce the risk of actual on-orbit
ARPO missions. The novel simulator design, called the Autonomous Rendezvous
and Proximity Operation ground Simulator (ARPOS) presented here mimics all
the six DOFs of rigid spacecraft with high delity using internal actuators. The
vii
advantages of internal attitude actuators, linear and spherical air bearings are in-
corporated in ARPOS to reproduce the near frictionless environment of an actual
spacecraft in space with accurate attitude maneuvering capability.
viii
CONTENTS
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Overview of ARPO . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Phases of ARPO . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 A brief survey of OOS techniques . . . . . . . . . . . . . . . . . . 4
1.3 Spacecraft Simulators . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.1 Classication of Spacecraft Simulators . . . . . . . . . . . 6
1.4 Attitude Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.1 Reaction Wheel . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.2 Control Moment Gyroscope . . . . . . . . . . . . . . . . . 10
2 Dynamics of Spacecraft in ARPO . . . . . . . . . . . . . . . . . . 13
2.1 Dynamics of Chaser and Target (Tumbling) Spacecraft . . . . . . 13
2.2 Relative Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3 Autonomous Rendezvous and Proximity Operation Ground Simulator 18
3.1 Ground Simulator Design . . . . . . . . . . . . . . . . . . . . . . 19
3.1.1 Planar Translation Stage . . . . . . . . . . . . . . . . . . . 19
3.1.2 Vertical Translation Stage . . . . . . . . . . . . . . . . . . 21
3.1.3 Attitude Stage . . . . . . . . . . . . . . . . . . . . . . . . 21
ix
3.1.4 Spherical Air-Bearing Stage . . . . . . . . . . . . . . . . . 21
3.1.5 Interchangeable Spacecraft Model Stage . . . . . . . . . . 22
3.2 ARPOS Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.1 Attitude Dynamics of Simulator . . . . . . . . . . . . . . . 26
3.2.2 Translational Dynamics of Simulator . . . . . . . . . . . . 27
4 ARPOS Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1 Attitude Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2 Translational Control . . . . . . . . . . . . . . . . . . . . . . . . . 32
5 ARPOS Experimental Testbed . . . . . . . . . . . . . . . . . . . . 35
6 Spacecraft Attitude Actuators . . . . . . . . . . . . . . . . . . . . 37
6.1 Variable Speed Control Moment Gyroscope . . . . . . . . . . . . . 37
6.1.1 Rotational Kinematics of Spacecraft with VSCMG . . . . 38
6.1.2 Lagrangian Dynamics for a Spacecraft with a CMG . . . . 42
6.1.3 Application of Hamiltons Principle . . . . . . . . . . . . . 44
6.2 Formulation of the Equations of Motion of Spacecraft with VSCMG 49
6.2.1 Equations Expressing the Dynamics . . . . . . . . . . . . . 49
6.2.2 Dynamics Equations with CMG Control Torques . . . . . 51
6.3 Relations between Momentum Quantities . . . . . . . . . . . . . . 52
6.3.1 Relations between Total Angular Momentum and CMG Mo-
menta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.4 CMG Design parameters . . . . . . . . . . . . . . . . . . . . . . . 55
6.5 Generalization to Spacecraft with n VSCMGs . . . . . . . . . . . 58
x
7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
xi
LIST OF TABLES
1 Classication of air-bearing based spacecraft simulator . . . . . . 8
2 Comparison between RW and CMG . . . . . . . . . . . . . . . . . 12
3 Tracking errors for the chaser . . . . . . . . . . . . . . . . . . . . 18
xii
LIST OF FIGURES
1 Phases of ARPO (OOS) mission . . . . . . . . . . . . . . . . . . . 3
2 Spacecraft with 3-axis reaction wheel . . . . . . . . . . . . . . . . 11
3 Chaser spacecraft chasing a tumbling target spacecraft . . . . . . 14
4 Schematic rendering of ARPOS showing its stages . . . . . . . . . 19
5 Thruster conguration of the planar stage . . . . . . . . . . . . . 20
6 Vertical Translation Stage . . . . . . . . . . . . . . . . . . . . . . 22
7 Spherical air-bearing stage . . . . . . . . . . . . . . . . . . . . . . 23
8 Interchangeable Spacecraft Model Stage (a) Dumbbell shape model,
(b) Umbrella shape model and (c) Tabletop model. . . . . . . . . 24
9 Attitude Stage of ARPOS . . . . . . . . . . . . . . . . . . . . . . 25
10 ARPOS Experimental Testbed for orbit determination problem . 35
11 ARPOS Experimental Testbed for rendezvous problem . . . . . . 36
12 Schematic diagram of single-gimbal VCMG with an axially sym-
metric rotor with its center oset in the axial direction from the
center of mass of the gimbal. . . . . . . . . . . . . . . . . . . . . . 39
xiii
1 Introduction
1.1 Overview of ARPO
Various autonomous on-orbit operations such as assembly of large space struc-
tures, resupply of orbital platforms and stations, exchange of crew in orbital sta-
tions [1], re-orbiting an orbiting vehicle and On-Orbit Servicing [2] (OOS) can
be grouped under Autonomous Rendezvous and Proximity Operations (ARPO).
Typical ARPO requires pursuing and then engaging two or more independent
spacecraft in the space to satisfy the mission objectives. As early as 1965, NASAs
GEMINI program [3] attempted spacecraft rendezvous operations, followed by
SKYLAB program and ApolloSoyuz Test Project [4]. At present the Inter-
national Space Station is extensively using rendezvous and docking or berthing
(RVD/B) and has also established an onboard laboratory to study and test the
concept of Rendezvous and Formation Flight [5]. Several attempts have been made
to study and showcase the application of On-Orbit Servicing (OOS), which in-
cludes Demonstration of Autonomous Rendezvous Technology (DART) [6], DARPAs
Orbital Express [7], NASDAs (now known as JAXA) ETSVII (Engineering Test
Satellite VII) [8] and ESADRLs TECSAS [9]. These recent attempts have led
to technological advances in the eld of Autonomous Rendezvous and Proximity
Operation (ARPO) [5][6][8][9].
OOS can be performed on cooperative (designed to be serviced) spacecraft
and non-cooperative (not designed for OOS) tumbling spacecraft; in both cases,
1
OOS comprises of retrieval, reorienting and refueling of target spacecraft [1][2].
In particular, servicing a tumbling, unaided (with no communication between
ground or chaser spacecraft) target is very challenging. Here in this work, an
ARPO mission is considered for the case of OOS with a non-cooperative tumbling
target spacecraft.
1.1.1 Phases of ARPO
An OOS mission can be divided into a number of important phases: ren-
dezvous phase (launch, phasing, far range rendezvous, close range rendezvous) [1],
station-keeping, capture, securing the target spacecraft, servicing and release[9].
Out of all, closerange rendezvous is the most critical and important phase for
successful proximity operations. The objective of close range rendezvous is to
match the position and orientation of the chaser relative to the target along with
the reduction in the relative distance between the two spacecraft.
Before the capture phase, the chaser spacecraft is aligned in close proxim-
ity with the target spacecraft in a station-keeping mode. In a stationkeeping
phase, the chaser spacecraft is kept in a xed position wirth respect to the target
and without motion in any direction. Stationkeeping requires active control to
maintain the relative geometry of the chaser and target in the presence of pertur-
bation forces and torques. Both the close range rendezvous and station-keeping
operations have to be controlled using relative navigation and motion estimation.
2
Phasing
Reduction of
orbital phase
angle between
chaser and
target
spacecraft,
absolute
navigation
Launch
Injection into orbital
plane of target
achievement of stable
orbital conditions
Far range
rendezvous
Transfer from
phasing orbit
to target orbit
Close range rendezvous
Final approach to capture point,
achievement of capture condition
Closing: reduction of relative distance
and maintaining constant attitude
between target and spacecraft
Capture
Grappling of target with
robotic manipulator
Docking
Achievement of rigid
structural connection
Target Chaser Target Chaser
Figure 1: Phases of ARPO (OOS) mission
3
1.2 A brief survey of OOS techniques
A stationkeeping maneuver with a cooperative spacecraft is similar to the
task of formation ight. In formation ying, two spacecraft or a cluster of
spacecraft is stationed in independent orbits, such that the formation geome-
try is preserved with active communication between the distributed spacecraft.
In leadfollower type of formation ying [10], one spacecraft is intended to follow
the trail of another spacecraft, which is similar to the ARPO problem provided
the target is cooperative. Autonomous capture of an uncooperative tumbling tar-
get using a robotic spacecraft is a challenging guidance, navigation and control
problem. This problem may arise in practice during the autonomous capture of
impaired satellites prior to servicing or of orbital debris as part of debris mitigation
eorts. Few studies have been carried out on the problem of capturing a tum-
bling spacecraft. A brief survey of linearized terminal rendezvous studies using the
TschaunerHempel equations or the ClohessyWiltshire equations is given in [11].
Ma et al., studied optimal control of a spacecraft to rendezvous with a tumbling
satellite in close proximity using a feedforward controller for a planar case [12].
Sakawa studied the problem of controlling a single freelyying object to y from
one position and orientation to another in an optimal manner [13]. Matsumoto
et al., studied yby and optimal orbits for maneuvering to a rotating satellite
[14]. Nakasuka and Fujiwara proposed a method for matching angular velocities
between the chaser and target by changing the targets moments of inertia [15].
4
Fitz-Coy and Liu proposed a two phase navigation solution for rendezvous with
a tumbling satellite in 2D space[16]. Studying the terminal APRO problem with
nonlinearities in the equations of motion is of great importance in reducing the
control eort and fuel consumption.
In this thesis, the APRO problem is studied with a nonlinear dynamics model
of the relative motion between chaser and target spacecraft. We assume that the
motion states of the tumbling target are available to the chaser from measurement
of the relative motion.
1.3 Spacecraft Simulators
Simulating the rendezvous and capture of tumbling spacecraft on ground
prior to an actual unaided ARPO mission would help to understand realtime
problems in motion control, state estimation and guidance. Design and testing
of control algorithms for spacecraft through realistic simulation on the ground
is technically challenging. The challenges include the complexities of spacecraft
dynamics with several nonlinearly coupled degrees of freedom of motion and their
energy distribution, the diculty of obtaining global control schemes due to the
non-linear state space of motion and the presence of uncertainties in dynamics
and measurements.
In this thesis, we only consider realistic simulation of the dynamics when the
dynamics models of both the pursuer and the target spacecraft are known and the
measurements are assumed to be accurate. The pursuer is controlled, while the
5
target is uncontrolled. The degrees of freedom (DOFs) for a multi-body space-
craft consist of three translational degrees of freedom, three rotational or attitude
degrees of freedom and internal or shape degrees of freedom. Internal DOFs
could include conguration variables of internal actuators like reaction wheels,
control-moment gyroscopes and proof mass actuators, as well as structural vi-
bration modes. Ground simulators that are designed to simulate partial or full
range motion of spacecraft have been in existence for several years. An excellent
review of spacecraft simulators using air-bearing supports is given in [17]. Spher-
ical air-bearing testbeds, which provide a large range of torque-free 3D attitude
motion, have been used to test attitude maneuvers and attitude control design
[18, 19, 20, 21]. Air-bearing pads that allow frictionless planar translational mo-
tion, have also been used to demonstrate ARPO and formation ying maneuvers
in ground tests [22, 23, 24, 25, 26].
1.3.1 Classication of Spacecraft Simulators
Most of these simulators can be grouped into two categories:(1) those that can
only simulate attitude and internal motion since they are mounted on stationary
support; and (2) those that can only simulate planar motion since they constrain
the spacecraft model to move on a plane. However, the dynamics of a spacecraft
model on such ground simulators is very dierent from that of spacecraft in plan-
etary orbit. Even the equilibria are dierent in these two situations, as can be
seen by comparing the dumbbell spacecraft model in central gravity [27, 28] to
6
the dumbbell model in an attitude control simulator [29, 30]. The main problem
with air-bearing tables is that they are limited in simulating planar motion, which
is adequate only for simulating the motion of spacecraft in co-planar Keplerian
orbits. Moreover, the planar translational motion does not lead to the same dy-
namical coupling between translational and attitude motion as the 3D motion in
space, as pointed out in [22]. Agarwal et al. and Tsiotras et al. designed and
developed 3 DOF and 5 DOF ground simulators, respectively; to study spacecraft
attitude dynamics, but these simulators cannot simulate all the complex dynamic
coupling between the several DOFs of spacecraft. Robotic manipulators have also
been used for simulating spacecraft motion in formation ying, with a spacecraft
model xed to the manipulator arm, as in [31]. These robotic manipulators can
simulate limited ranges of 3D translational motion and 3D attitude motion, but
are susceptible to disturbances due to ground vibrations and vibrations in the ma-
nipulator arm. Classication of existing spacecraft simulator based on the degrees
of freedom, motion type and actuator type is given in Table 1.
For spacecraft dynamics, the translational DOF contain most of the total
energy of motion. However, the nonlinear dynamics of a spacecraft in central
gravity leads to complex dynamic coupling between its translational, attitude and
internal degrees of freedom. This dynamic coupling in turn leads to energy transfer
between these dierent degrees of freedom. High percentage of total energy in the
translational DOFs implies that translational motion is not substantially aected
by energy transfer from and to the attitude and shape DOFs.
7
Table 1: Classication of air-bearing based spacecraft simulator
DOF Motion type Mechanism
2 Translation (planar) Linear air bearing
3 Rotation Spherical air bearing
5 3D Rotation Linear
+ +
2D Translation Spherical air bearing
6 3D Planar+ 3D Rotation Linear
+
Spherical air bearing
+
Ball-screw mechanism
On the other hand, such energy transfer between the DOFs can signicantly
aect the overall motion. Changes in the translational energy can signicantly
aect both attitude and internal motion, as seen during atmospheric entry. More-
over, the spacecrafts attitude signicantly aects orbital (translational) maneu-
vers when thrusters are xed to the spacecraft body. Ground simulators designed
to realistically simulate spacecraft motion need to account for these facts. There
are existing spacecraft ground simulators that can simulate 6 DOF (translational
and attitude) motion of spacecraft with limited attitude and translation con-
straints, like the one presented in [32]. For simulation of ARPO between two
8
satellites in orbit, the relative attitude motion of the spacecraft is more important
than the relative translational motion. This is because the relative translational
motion is primarily dictated by the orbital motions of the two satellites, while
the relative attitude motion is controlled to achieve desired proximity maneuvers.
We present a novel design concept for a 6 DOF spacecraft ground simulator that
can simulate the complex dynamical interactions between the dierent degrees of
freedom and can be used to test spacecraft control algorithms.
1.4 Attitude Actuators
Spacecraft attitude control is the most complex part of a rendezvous problem.
Precise attitude matching between the chaser and target spacecraft is necessary for
a successful capture mission. Internal motion can signicantly aect the attitude
motion; a fact that is used by all internal actuators (reaction wheels, control-
moment gyroscopes, proof mass actuators, etc.,) used for active attitude control.
1.4.1 Reaction Wheel
Reaction wheels generate torques on a spacecraft by internal angular momen-
tum transfer from a rotor to the spacecraft bus (base body). Since the torque is
internal to the spacecraft, the reaction wheel cannot change the inertial angular
momentum of the spacecraft. Rather, it exchanges momentum between the space-
craft bus and the ywheel or rotor. In the presence of any external torque, the
spacecraft attitude can be reorientated by accelerating or decelerating the reaction
9
wheel rotor, changing the momentum magnitude. While in the case of a Control
Moment Gyroscope (CMG), the direction of the rotor angular momentum vector
of constant magnitude is reoriented using a gimbal mechanism. However, if there
is a constant inertial torque that the reaction wheel has to counteract, then the
wheel will spin up and eventually saturate [33]. In order to have full 3D attitude
and angular velocity control, a minimum of 3 noncoplanar reaction wheels are
needed as shown in the Figure 2. Table 2 gives the comparison between RW and
CMG, inferred from [34] for same conditions.
1.4.2 Control Moment Gyroscope
Control moment gyroscopes are momentum exchange devices that can be used
as actuators for controlling the attitude motion of spacecraft [35], including atti-
tude stabilization of agile spacecraft [36]. Control moment gyroscopes (CMGs) can
be categorized as single gimbal Control Moment Gyroscopes (SGCMGs) or double
gimbal Control Moment Gyroscopes (DGCMGs). Further, variable speed Control
Moment Gyroscopes (VSCMGs) combine the features of constant speed SGCMGs
and reaction wheels (RWs), which have variable angular speed. Dening features
and comparisons between SGCMGs and DGCMGs are given in [37, 38]. Theoret-
ical aspects of VCSMGs for attitude control are given in [39], while experiments
using VSCMGs for this purpose are reported in [40]. SGCMGs can produce larger
torques for the same actuator mass and peak power when compared to reaction
wheels [38]. However, due to higher mechanical and electrical complexity and
10
Figure 2: Spacecraft with 3-axis reaction wheel
the complex nonlinear mapping, with inherent geometric singularities, between
the input (gimbal) space and output (momentum) space, SGCMGs are not as
commonly applied as RWs. A detailed analysis of singularities and steering laws
for SGCMGs, obtained for dierent geometrical congurations, is to be found in
[41]. Most of the available steering laws that locally produce instantaneous control
torques, have diculties avoiding singular states in minimally redundant SGCMG
systems [42]. An earlier treatment of singular congurations of SGCMG clusters
11
and singularity avoidance is provided in [43]. A hybrid approach, that takes into
account the form of singularities by avoiding hyperbolic singularities and escap-
ing elliptic singularities for a four-SGCMG rooftop arrangement, is provided in
reference [44].
Table 2: Comparison between RW and CMG
Parameters Reaction Wheel CMG
DOF 1 2 - SGCMG; 3 - DGCMG
Rotating parts Rotor Gimbal and Rotor
Rotor rate Variable speed Constant [except VSCMG]
Cluster mass High High
Peak power consumption High Low
Max. torque Low High
Control Simple Complex
12
2 Dynamics of Spacecraft in ARPO
Spacecraft in orbit is subjected to lot of perturbations which leads to uncer-
tainties in modeling them for ground simulations. In order to build a spacecraft
simulator which mimic the exact dynamics as that of actual spacecraft in or-
bit, dynamics of actual spacecraft in rendezvous with tumbling target object is
studied.
2.1 Dynamics of Chaser and Target (Tumbling) Spacecraft
The target spacecraft is assumed to be non-cooperative with the chaser and
moves with unconstrained and uncontrolled motion in an orbit. The chaser is
tasked to rendezvous with the target as shown in Figure 3. At the terminal stage,
the chaser approaches the target in close proximity without any relative motion
between the two.
The global conguration of the chaser spacecraft and target object is given by
their respective position vectors and orientations with respect to an inertial (refer-
ence) coordinate frame. Orientation is described by the rotation matrix transfor-
mation from a body-xed coordinate frame to the inertial coordinate frame; this
gives a global and unique representation of the attitude motion. The body coor-
dinate frames for both the chaser and the target have origins at their respective
centers of mass.
The translational and attitude motion kinematics for the two spacecraft are
13
Figure 3: Chaser spacecraft chasing a tumbling target spacecraft
14
described as follows:

b = R,

R = R

, (1)

b
0
= R
0

0
,

R
0
= R
0

. (2)
In the above equations, ()

denotes the skew-symmetric matrix corresponding to

the linear cross-product operator for 3D vectors; i.e., v

1
v
2
= v
1
v
2
for v
1
, v
2
R
3
and superscript ()
0
denotes quantities for target object. Let m and J denote the
mass and inertia tensor of the chaser spacecraft respectively, and let m
0
and J
0
denote the mass and inertia tensor of the target object. The inertia tensors of
both bodies are expressed in their respective body frames. The dynamics of the
chaser spacecraft is expressed as:
m = m + F
g
(b, R) +
c
+
d
, (3)
J

= J + M
g
(b, R) +
c
+
d
, (4)
where
c
,
c
are the control force and control moment respectively, and
d
,
d
are the non-conservative force and moment (respectively) acting on the chaser
spacecraft. The gravity force is denoted by F
g
(b, R) and the gravity gradient
moment is given by M
g
(b, R). The dynamics of the target object (in its body
coordinate frame) is expressed as:
m
0

0
= m
0

0
+ F
0
g
(b
0
, R
0
) +
0
g
, (5)
J
0

0
= J
0

0
+ M
0
g
(b
0
, R
0
) +
0
g
, (6)
where
0
d
and
0
d
are the non-conservative force and moment on the target object.
15
F
0
g
(b
0
, R
0
) and M
0
g
(b
0
, R
0
) are the gravity force and gravity gradient moment,
respectively, acting on the target. The external force and moment include the
eects of gravity, atmospheric, solar, and magnetospheric eects. Note that all
forces and moments acting on these two bodies are resolved in their respective
body coordinate frames. It is assumed that there are no control inputs to the
targets motion and the target is not communicating with the chaser.
2.2 Relative Dynamics
Prior to autonomous rendezvous, the two spacecraft are assumed to be in
proximity such that the relative attitude of the target can be resolved in the
chasers body frame. For ease in design of the control and estimation schemes,
the relative motion between the chaser and target spacecraft are resolved in the
body frame of the chaser. To achieve the objectives of tracking the target objects
motion, we assume that we have a guidance algorithm for the chaser that directs
the chaser to close in on the target and maintain a constant relative pose (posi-
tion and orientation) with respect to the target after a period of time from the
start of the maneuver. This guidance algorithm provides a desired or reference
state trajectory for the chaser spacecraft based on remote measurements of the
instantaneous states of motion of the target. For example, the guidance scheme
given in [45], creates a reference trajectory which the chaser tracks to reach the
desired end state relative to the target. In this paper, the desired end state is
considered to be the targets state after a given time period from the start of the
16
maneuver. The reference states of the chaser satellite is represented using the
following representations of SE(3) and se(3):
g
r
=
_
R
r
b
r
0 1
_
SE(3) and
r
=
_

r

r
0 0
_
se(3), (7)
where the lower-left entry 0 on both matrices above denotes a row vector of three
zeros. The above representation is similar to the so-called Denavit-Hartenberg
representation of transformations between links in a chain of links in robots [46].
A subscript ()
r
is used to denote a reference trajectory value for the chaser.
Therefore, the kinematics of the reference trajectory on SE(3) can be obtained as:
g
r
= g
r

r
(8)
We assume that there are guidance and feedback tracking control schemes
implemented on the chaser spacecraft, that enable it to close in on the target
object.
With the denitions of the tracking errors given in Table 3, we obtain a left-
invariant error kinematics on SE(3) as given by:
q = q, where q =
_
Q x
0 1
_
and =
_

0 0
_
. (9)
These are trajectory tracking errors between the chasers desired (reference) state
trajectory and its actual states.
17
Table 3: Tracking errors for the chaser
a(t) b(t) b
r
(t) error in inertial position
x(t) R
T
r
(t)a(t) position error given in
reference body frame
Q(t) R
T
r
(t)R(t) error in body attitude (orientation)
(t) (t) Q
T
(t)(
r
(t) +
r
(t)

x(t)) error in body translational velocity

(t) (t) Q
T
(t)
r
(t) error in body angular velocity
where
c
r

r
+

r
x.
3 Autonomous Rendezvous and Proximity Operation Ground Simu-
lator
To simulate the complex dynamical interactions between the dierent de-
grees of freedom of motion in a spacecraft with high delity, the proposed ground
simulator, Autonomous Rendezvous and Proximity Operation ground Simula-
tor (ARPOS) presented here is composed of three stages, namely (A) Planar
Motion Stage, (B) Vertical Translation Stage (translational only), and (C) 3D
Attitude Stage as shown in Figure 4. This design combines salient features of
spherical air-bearings and linear air-bearing tables to provide 3D translational
and 3D rotational (attitude) motion within hardware limits in limited laboratory
oor space. This design is to facilitate: (a) the realistic simulation of relative
spacecraft motion during ARPO using ground experiments, and (b) the design
and testing of spacecraft control and navigation algorithms for ARPO through
18
Figure 4: Schematic rendering of ARPOS showing its stages
such ground simulations. To meet the overall objectives, we ensure in our design
that the complex dynamical interactions between the dierent degrees of freedom
of motion in a spacecraft in orbit is closely emulated in our ground simulator
design.
3.1 Ground Simulator Design
3.1.1 Planar Translation Stage
The planar motion stage serves as the mobile base and physically forms the
lower part of ARPOS. The remaining systems and sub-systems are built on this
19
base platform. The entire system sits on the planar stage and is allowed to oat
over a thin lm of air on a smooth granite surface. Three linear air bearing pads
mounted rigidly to the base platform in a triangular conguration are powered by
compressed air from three air tanks connected directly to respective bearing pad,
providing near frictionless smooth motion of ARPOS. The air tanks are secured
and harnessed to each of the three linear actuators. This argument is intended to
maintain a constant position of the CoM of the planar translational stage despite
the air discharge from the air tanks. This CoM location is maintained at the ge-
ometric center of the planar translational stage. The rotational and translational
motion of ARPOS lower stages (planar and vertical traversing stages) is achieved
using three fan thrusters xed to the base platform in a triangular conguration
as shown in the Figure 5. Planar stage can be independently controlled for pla-
nar translational and rotational motion using a simple control scheme, which is
described in section 4.2.
Figure 5: Thruster conguration of the planar stage
20
3.1.2 Vertical Translation Stage
The vertical traversing stage of ARPOS is designed to provide vertical trans-
lation of the suspended body along the vertical axis. Three motorized linear
actuators are rigidly xed to the base platform in a triangular conguration as
shown in Figure 6, to take the load of the attitude stage (spherical air-bearing and
supported spacecraft model). The advantage of having linear actuation mecha-
nism instead of pneumatic mechanism is because of its ner resolution in the
vertical translation motion and weight reduction. The vertical translation stage
of ARPOS does not need any gravitational force balancing mechanism as in [32],
since uniform gravity moment is counteracted by the active control scheme. The
vertical motion of all the three linear actuators is synchronized so that the vertical
motion of the attitude stage with its supported body is balanced.
3.1.3 Attitude Stage
The attitude stage of ARPOS is further sub-divided into:
3.1.4 Spherical Air-Bearing Stage
The spherical air-bearing would be mounted on vibration dampers as shown in
Figure 7 to overcome any external disturbances caused by the motors of the linear
actuator. These dampers are xed to the moving part of the linear actuator so that
the spacecraft model suspended on the spherical air bearing translates vertically.
The spherical air bearing stage is fueled by pressurized air from one tank located
21
Figure 6: Vertical Translation Stage
at the mobile platform, symmetric about the planar stage rotational axis.
3.1.5 Interchangeable Spacecraft Model Stage
The interchangeable spacecraft model stage in the ground simulator can be
further categorized into dumbbell, tabletop, and umbrella [17] as shown in Figure
8. All these types of spacecraft models can be accommodated on ARPOS. A
22
Figure 7: Spherical air-bearing stage
dumbbell-shaped spacecraft model, shown in Figure 8.a, has a sphere mounted in
the center with two side plates attached to the extreme ends of the rod aligned
along the center axis of the sphere. The side plates have provision for attaching
some spacecraft subsystems and actuators. Dumbbell shape has the advantage of
unrestricted yaw and roll motion with limited pitch motion and greater stability,
as the center of mass (CoM) of the spacecraft model can be made close to the
center of rotation of the spherical air-bearing. Figure 8.b and 8.c depict the
umbrella and tabletop shaped spacecraft models respectively. Tabletop model uses
a hemispherical bearing, whereas the umbrella model has a much larger portion
of the spherical ball of the air-bearing. Both these models are restricted in roll
23
and pitch motion, with unrestricted yaw motion. However, the pitch motion of
tabletop model is the most restricted among the other types of model. Dumbbell-
Figure 8: Interchangeable Spacecraft Model Stage (a) Dumbbell shape model, (b)
Umbrella shape model and (c) Tabletop model.
shaped spacecraft model found in ARPOSs attitude stage as in the Figure 9, can
simulate 3D attitude motion and internal motion. The spherical air-bearing which
supports the interchangeable dumbbell-style spacecraft model allows full freedom
in yaw and roll motion, and a limited range of pitch motion. The spacecraft
model stage can have RWs or VSCMGs to produce the required attitude with its
24
Figure 9: Attitude Stage of ARPOS
control unit mounted at one of its side plate. As the spacecraft model has all the
6 DOFs of attitude (with limited pitch) and translational motion, the ARPOS
simulator can mimic the dynamics of a spacecraft in planetary orbit [47]. Mass
and inertia properties of ARPOSs spacecraft model can be set to desired values
by adjusting the weights on the side plate. So, the same simulator can be used for
testing dierent congurations. The spacecraft model is balanced by adjusting the
counter weights on either side of the side plates, which ensures stable equilibrium
when the center of gravity lies below the center of rotation [48]. Gravity moment
is taken into account in the active control scheme applied to the spacecraft model.
25
3.2 ARPOS Dynamics
3.2.1 Attitude Dynamics of Simulator
Simulating the attitude dynamics of a spacecraft in orbit through ground sim-
ulations is more challenging than simulating the relative translational dynamics
for two spacecraft with a small relative velocity (as expected during rendezvous
and docking), since the attitude motion has strong nonlinearities. Here we briey
describe how the proposed spacecraft simulator would simulate the attitude dy-
namics of a spacecraft in orbit. For simplicity, we assume that the spacecraft
model mounted on the simulator can be modeled as a rigid body. Let m
b
> 0 and
J
b
R
33
> 0 denote the mass and moment of inertia matrix for this spacecraft
model. The conguration of the mobile base of the simulator, which executes pla-
nar motion on the ground, can be expressed in terms of the inertial position of its
center of mass and its attitude angle in its plane of motion. Let R
g
denote the
attitude of the supported spacecraft model, given by the rotation matrix from a
body-xed coordinate frame to the mobile base coordinate frame, and let
g
R
3
is the angular velocity of the spacecraft model with respect to the mobile base of
ARPOS. The attitude kinematics of the supported model is then given by

R
g
= R
g

g
, (10)
The attitude dynamics equations of motion for the supported model are:

=
g
where = R
T
g
e
3
,

g
=
g
(
g
+

) + m
b
g +
u
,
(11)
26
where
g
= J
b
(
g
+

) is the angular momentum of the supported body in the
body frame, e
3
= [0 0 1]
T
R
3
is the unit vector denoting the direction of gravity
in inertial frame, g is the acceleration due to gravity, is the body vector from
the center of support to the center of mass, and
u
R
3
is the control moment
applied to the supported spacecraft model. The vector is the direction of gravity
expressed in the body frame. Note that we do not assume that the spacecraft
model is mass-balanced on the spherical air-bearing support such that the center
of mass coincides with the center of support. This is a common assumption in
spherical air-bearings used for spacecraft attitude motion simulation, but is known
to be dicult to achieve in practice [49]. The proposed ARPOS simulator design
does not impose this constraint on the spacecraft model being supported, since
the moment generated by uniform gravity is counteracted by active control during
ground simulations.
3.2.2 Translational Dynamics of Simulator
A brief description of how two ARPOS simulators would simulate the relative
translational motion between two spacecraft, with one simulating a chaser space-
craft and the other simulating a target, is provided here. Let b
g
R
3
and b
0
g
R
3
denote the position vectors of the centers of support of the two supported bodies
in their corresponding simulator supports in a lab-xed inertial frame, where b
0
g
is the position of the body representing the target and b
g
is the position of the
body representing the chaser. We assume that the chaser obtains feedback of the
27
target bodys position and velocity relative to itself, as is the case in an unaided
docking or capture maneuver. Let the orientation of the mobile base of the chaser
body in the lab-xed inertial frame be
R

=
_
_
cos sin 0
sin cos 0
0 0 1
_
_
,
R = R

R
g
Attitude of spacecraft model with respect to inertial frame
= R
T
e
3
= R
T
g
R

e
3
= R
T
g
e
3
Since,
R

e
3
= e
3
where is measured counter-clockwise from the inertial x-axis. Let
g
R
3
and
v
g
R
3
denote the translational velocity of the supported body simulating the
chaser, represented in the body frame of the simulator base and in the inertial
frame respectively. Then we have
v
g
= R

g
, or
g
= R
T

v
g
. (12)
If x
g
= b
g
b
0
g
denotes the relative position between the ARPOS supported bodies
in the inertial frame, then the relative velocity is:
x
g
=

b
g


b
0
g
= R

g
v
0
g
, (13)
where v
0
g
=

b
0
g
is the translational velocity of the target body represented in the
inertial frame, which can be obtained from the measured (or estimated) relative
velocity of target body with respect to the chaser body, and the known velocity
28
of the chaser body. Note that the velocities v
g
and v
0
g
have z-components that are
separately actuated by the vertical traversing stages of the two ARPOS simulators.
29
4 ARPOS Controls
4.1 Attitude Control
In contrast to the ARPOS attitude dynamics (11), the attitude dynamics of
a rigid spacecraft is given by equation (4) if the spacecraft is controlled (chaser
spacecraft) or equation (6) if the spacecraft is not controlled (target spacecraft).
In general, the attitude dynamics of a rigid spacecraft can be written as

s
=
s

s
+ M
g
(b, R) +
c
+
d
, (14)
where
s
= J
s

s
, J
s
is the spacecraft moment of inertia matrix,
s
is the to-
tal angular velocity of the spacecraft in its body frame, M
g
(b, R) is the gravity
gradient moment,
d
is the sum of other external moments, and
c
is the control
torque applied to the spacecraft. Here we assume that we know these quantities;
in particular, the control torque may be obtained from a known control law. Note
that
s
contains coupling terms from the translational (orbital) motion of the
spacecraft due to the orbital angular velocity. The idea behind simulating the
attitude dynamics with the proposed simulator is to ensure that the total angular
velocity of the supported spacecraft model approaches the angular velocity of the
orbiting spacecraft being simulated, i.e.,
g
+


s
.
Since the attitude of the planar motion stage can be independently controlled,

and
u
can be controlled so that the attitude dynamics (11) approaches the
desired attitude dynamics (14). The following result gives a control scheme for
the attitude dynamics of an ARPOS simulator tracking the attitude dynamics of
30
a controlled chaser spacecraft.
Theorem 1. Control law

u
= L(
s

g


) + J
b
(

g

s
) + M
g
(R) +
d
+
c
m
b
g

, (15)

= 0, (16)
ensure that the angular velocity
g
+

of the supported body in ARPOS tracks

s
asymptotically.
Proof: Dene the total angular velocity of the supported body on the simulator
as

g
=
g
+

. (17)
We use the candidate Lyapunov function
V =
1
2
(
g

s
)
T
J
b
(
g

s
). (18)
Taking a time derivative of this function and substituting the attitude dynamics
equations (11) and (14), as well as the control law (15), we get

V = (
g

s
)
T
L(
g

s
) +

T
_
J
b
(
g

s
)
s
_
. (19)
Equation (19) assumes that J
b
= J
s
Substituting the angular velocity control law (16) in the expression (19), we get

V = (
g

s
)
T
L(
g

s
), (20)
which is negative denite in the angular velocity error. This proves the result.
31
According to the control law (16), the planar motion stage (base) can have
a constant planar attitude (zero turn rate

). Note that this design eliminates
dynamical coupling from the attitude motion of the supported body to the trans-
lating base and not vice-versa, dynamical coupling is retained from the translation
to the attitude motion of the supported body.
4.2 Translational Control
Physical constraints of ground experiment on limited lab oorspace, imposes
constraints on the translational motion that can be achieved. Therefore, a simple
control scheme for the translational motion of ground simulator, that can be
carried out on limited oorspace is given here. For translational control of the
ARPOS ground simulators, we consider a maneuver in which the chaser positions
itself at a constant relative position from the target as a precursor to docking or
capture maneuvers. Let this desired constant relative position vector be denoted
p. A simple velocity control law to achieve this positioning is given by:
x
g
= v
g
v
0
g
= (x
g
p) v
g
= v
0
g
(x
g
p), (21)
where > 0 is a constant scalar gain. This can then be converted to a force
control law in the body frame of the chaser simulators mobile base, as follows:
f = mR
T

v
g
= mR
T

( v
0
g
x
g
), (22)
32
which requires knowledge of the acceleration v
g
0
of the target. Substituting for x
g
from (21), this control law can be expressed as
m v
g
= R

f = m( v
0
g
+
2
(x
g
p))
v
g
v
0
g
=
2
(x
g
p)
x
g
=
2
(x
g
p)
Which needs,
x
g
+ c x
g
=
2
(x
g
p) v
g
v
0
g
+ c(v
g
v
0
g
) =
2
(x
g
p) (23)
f = mR
T

v
0
g
c(v
g
v
0
g
) k(x
g
p) (24)
for asymptotic stability.
The acceleration v
0
g
can be estimated from the observed velocity of the target
simulator when simulating unaided (uncooperative) capture. The force control
law (24) gives the x- and y-components force (body-frame), on the planar motion
stage of the simulator. The z-component of f is the force provided to the vertical
traversing stage for the vertical motion of the attitude stage and supported body.
For the simple case that v
0
g
= 0, we get
f = mR
T

c(v
g
v
0
g
) k(x
g
p) (25)
It is assumed that the translational control law of the simulator is in conso-
nance with the attitude control law for the spacecraft model stage. In the special
case that both v
0
g
= 0 and v
0
g
= 0, we get
f = mR
T

cv
g
k(x
g
p) (26)
33
The attitude control (15), (16) and translational force control (24) are com-
patible with each other. As the chaser ARPOS approaches the target utilizing the
force control law given in equation (24), the relative velocity between the chaser
and target remains constant. The attitude control law equation (15) and equation
(16) which imposes

= 0, controls the chaser ARPOSs attitude stage to track
the targets attitude and ensures
g
=
g

0
g
, where
0
g
is the target ARPOS
angular velocity. Equation (26) demonstrates the condition of the chaser ARPOS
approaching the stationary target. Once after establishing a constant relative po-
sition with respect to the target, the attitude control starts to track the targets
attitude.
34
5 ARPOS Experimental Testbed
ARPOS experimental testbed is designed to accommodate variety of problems like
autonomous rendezvous, proximity operations, formation ying and orbit deter-
mination etc. The testbed is made of at circular granite disk, leveled parallel to
the ground, vibration isolated and mounted rigidly. The circular testbed provides
enough surface area to perform the ARPO experiments in laboratory environment.
Figure 10: ARPOS Experimental Testbed for orbit determination problem
Each experiment requires certain setup modication, for example an orbit
determination or a trajectory generation experiment demands a track arrangement
as in the Figure 10, in which a target spacecraft model is allowed to move in a pre-
35
dened orbit. The chaser ARPOS, which oats on the smooth surface is tested
for orbit trajectory determination, generation and optimization algorithms.
Figure 11: ARPOS Experimental Testbed for rendezvous problem
Ground simulation of ARPO can be performed using ARPOS in a sophisticated
experimental testbed as in the Figure 11. Two non-identical ARPOS simulator
(dierentiated only by colors), are allowed to oat on the granite surface, con-
sidering one as the target and the other as chaser. Target ARPOS can be set to
follow feasible motion on the granite surface to any desired attitude and transla-
tional parameters. Chaser ARPOS is allowed to navigate the target based on the
tracking trajectory given by the guidance scheme to rendezvous and maintain a
relative position with respect to the target.
36
6 Spacecraft Attitude Actuators
6.1 Variable Speed Control Moment Gyroscope
The formulation of the equations of SGCMGs that have been given other
names such as gyrodynes, have been presented in Newton-Euler form under the
assumptions: a CMG with an axially symmetric ywheel, where its center of mass
(CM) is along its gimbal axis; both the gimbal and rotor (ywheel) are rigid; and
the input torque into the gimbal is negligible compared to the gyroscopic output
torque from the CMG, in references [50, 51, 39, 52]. In this paper, we do away with
some of these assumptions and study the dynamics and control of a spacecraft
with one VSCMG, from the perspective of variational mechanics on the nonlinear
state space of this system. A brief treatment of the generalization to a spacecraft
with a nite number of VSCMGs is also provided. For the VSCMG, we consider
the more general situation where the input torque to the gimbal is not negligible
and the center of mass of the rotor is not along the gimbal axis. To best of the
authors knowledge, there has been no complete formulation of these equations of
motion in a variational setting, considering a rotor (ywheel) with its center of
mass oset from the gimbal axis.
In order to better understand the relations between CMG inputs and base
body rotational motion, the dynamics of a spacecraft with one VSCMG has been
obtained here using a variational approach. Since the conguration space of ro-
tational motion of the spacecraft with internal actuators is a nonlinear manifold,
37
the global dynamics of this system is treated using the formulation of geometric
mechanics [53, 54]. The treatment of the dynamics and control of a rigid body
(spacecraft) with an internal actuated rotor (reaction wheel) has been provided
in [54]. However, to the best of the authors knowledge, the treatment of the
dynamics of a spacecraft with CMGs in the framework of geometric mechanics
has not been carried out before. We therefore use the global representation of
the attitude of the spacecraft provided by the rotation matrix from the spacecraft
base body-xed coordinate frame to the inertial coordinate frame. This approach
is powerful and general enough to globally treat the dynamics of a spacecraft with
one CMG.
6.1.1 Rotational Kinematics of Spacecraft with VSCMG
Consider a coordinate frame xed to the center of mass of the base body
of the spacecraft. Assume that the CMG consists of an axially symmetric rotor
rotating about its symmetry axis, which in turn is rotating about a gimbal axis
xed in the spacecraft base body coordinate frame. Let (t) S
2
denote the
instantaneous symmetry axis of the rotor at time instant t, and let g S
2
denote
the xed axis of rotation of the gimbal, both expressed as unit vectors in the
base body coordinate frame. We assume that (t) is always normal to the xed
axis g. Dene a CMG gimbal-xed coordinate frame with its rst coordinate axis
along g, its second axis along (t) and its third axis along g (t), to form a
right-handed orthonormal triad of basis vectors. This VSCMG conguration is
38
represented pictorially in Figure 12.
g
(t)
g (t)

Figure 12: Schematic diagram of single-gimbal VCMG with an axially symmetric

rotor with its center oset in the axial direction from the center of mass of the
gimbal.
Let R
g
denote the rotation matrix from the CMG gimbal-xed coordinate
frame to the base body coordinate frame, which can be expressed as
R
g
(t) =
_
g (t) g (t)

. (27)
Let (t) represent the rotation angle of the rotor axis about the gimbal axis g.
Then the time rate of change of the rotor axis (t) is given by
(t) = (t)g

(t). (28)
39
The kinematics of the CMG is then given by

R
g
(t) =
_
0 (t)g (t) g (t)

=
_
0 (t)g (t) g
_
(t)g (t)
_
=
_
0 (t)g (t) (t)(t),

(29)
using the vector triple product identity a (b c) = (a
T
c)b (a
T
b)c in the last
step above. This equation can also be expressed as

R
g
(t) = R
g
(t)
_
0 e
3
e
2

= R
g
(t) (t)e

1
, (30)
where e
1
= [1 0 0]
T
, e
2
= [0 1 0]
T
and e
3
= [0 0 1]
T
are the standard basis vectors
(as column vectors) of R
3
. Therefore, the angular velocity of the gimbal-xed
frame with respect to the base body, expressed in the base body frame, is

g
(t) = (t)g. (31)
Let (t) denote the instantaneous angle of rotation of the rotor about its symmetry
axis (t); this is the angle between the rst axis of a CMG rotor-xed coordinate
frame and the gimbal axis, where the rotor axis forms the second coordinate axis
of this coordinate frame. Therefore, the rotation matrix from this rotor-xed
coordinate frame to the base body coordinate frame is
R
r
(t) = R
g
(t) exp
_
(t)e

2
_
. (32)
If the angular speed of the rotor

(t) is time-varying, this corresponds to a
VSCMG. The time rate of change of the rotation matrix R
r
(t) is obtained as
40
follows:

R
r
(t) =

R
g
(t) exp
_
(t)e

2
_
+ R
g
(t) exp
_
(t)e

2
_

(t)e

2
=R
g
(t) (t)e

1
exp
_
(t)e

2
_
+ R
r
(t)

(t)e

2
= (t)R
r
(t)
_
exp
_
(t)e

2
_
e
1
_

+ R
r
(t)

(t)e

2
,
using the fact that R
T
a

R = (R
T
a)

for R SO(3) and a R

3
[55]. It can be
veried that the following holds for any S
1
:
exp
_
e

2
_
e
1
= (cos )e
1
(sin )e
3
.
The time derivative of R
r
(t) can therefore be expressed as

R
r
(t) = R
r
(t)
_
(t)
_
(cos )e
1
+ (sin )e
3
_
+

e
2
_

.
Therefore, the angular velocity of the CMG rotor with respect to the base body,
expressed in the base body frame, is

r
(t) = R
r
(t)
_
(t)
_
(cos )e
1
+ (sin )e
3
_
+

e
2
_
= (t)R
g
(t)e
1
+

(t)R
g
(t)e
2
= (t)g +

(t)(t). (33)
Let R(t) denote the rotation matrix from the base body-xed coordinate frame
to an inertial coordinate frame. If is the angular velocity of the base body
with respect to the inertial frame and expressed in the base body frame, then the
attitude kinematics of the base body is given by

R(t) = R(t)(t)

. (34)
41
Let
g
denote the position vector of the center of mass of the gimbal from the
center of mass of the base body, assuming to lie at normal to the gimbal axis, and
expressed in the base body frame. Let be the distance of the center of mass of
the rotor from the center of mass of the gimbal; this distance is assumed to be
along the vector (t). The center of mass position vector of the rotor is therefore
given by
r
(t) =
g
+ (t) in the base body frame. Therefore, these vectors can
be expressed in the inertial frame as
r
g
(t) = R(t)
g
and r
r
(t) = R(t)
r
(t).
Thus, the inertial velocities of these centers of mass are given by
r
g
(t) = R(t)(t)

g
and (35)
r
r
(t) = R(t)
_
(t)

r
(t) + (t)g

(t)
_
. (36)
The spacecraft with a CMG has ve degrees of freedom, which is the dimension
of its conguration space described (locally) by the variables , and R. The
conguration space Q of this system has the structure of a principal (ber) bundle
with base space B = T
2
= S
1
S
1
and ber G = SO(3) [56].
6.1.2 Lagrangian Dynamics for a Spacecraft with a CMG
The dynamics model for a spacecraft with a VSCMG is developed in this
section. Denote the inertia matrix of the gimbal only as J
g
and the inertia matrix
of the rotor only as J
r
, both resolved in the base body coordinate frame. Let the
mass of the gimbal only be m
g
and the mass of the rotor only be m
r
. Therefore,
42
the total rotational kinetic energy of the gimbal only is
T
g
(t) =
1
2
_
(t) +
g
(t)
_
T
J
g
_
(t) +
g
(t)
_
+
1
2
m
g
r
g
(t)
T
r
g
(t)
=
1
2
_
(t) + (t)g
_
T
J
g
_
(t) + (t)g
_

1
2
m
g
(t)
T
_

g
_
2
(t). (37)
The total rotational kinetic energy of the rotor (ywheel) is
T
r
(t) =
1
2
_
(t) +
r
(t)
_
T
J
r
_
(t) +
r
(t)
_
+
1
2
m
r
r
r
(t)
T
r
r
(t)
=
1
2
_
(t) +

(t)(t)
_
T
J
r
_
(t) +

(t)(t)
_
+
1
2
m
r
_
(t)

g
+ (t)

(t)
_
T
_
(t)

g
+ (t)

(t)
_
, (38)
where
(t) = (t) + (t)g. (39)
Let J
b
denote the inertia matrix of the base body expressed in the base body
frame. The rotational kinetic energy of the base body of the spacecraft is
T
b
(t) =
1
2
(t)
T
J
b
(t). (40)
Therefore, the total rotational kinetic energy of the system is
T(t) = T
g
(t) + T
r
(t) + T
b
(t). (41)
43
Let the gravity potential for the spacecraft be dependent on its attitude only, i.e.,
V (t) = V (R(t)). The Lagrangian for the spacecraft with a CMG is given by
/(t) = T(t) V (t). (42)
6.1.3 Application of Hamiltons Principle
The action functional over a time interval [0, T] is dened
S =
_
T
0
/(t)dt, (43)
which depends on the time evolution of the states R, , , , and

in this
interval. Note that , are uniquely specied by , and the constant vector g.
In the absence of non-conservative external torques on the system consisting of
the spacecraft with a CMG, Hamiltons principle states that the time trajectory of
the states is given by the critical (stationary) point of the action functional with
xed terminal states. Before we obtain the rst variation of the action functional,
we express the rst variations of the variables R SO(3), R
3
, S
2
, and
T

S
2
as follows:
R(t) = R(t)(t)

, (t) =

(t) + (t) (t), (44)
(t) = (t)g

(t), (t) = (t)g

(t) + (t)g

(t), (45)
where (t)

so(3) gives a variation vector eld on SO(3); thus, R is tangent

to SO(3) at the point R [53, 54]. Using equation (28), the vector triple product
identity (a

c = (a
T
c)b (a
T
b)c), and the fact that g and (t) are orthogonal
44
vectors, the second of equations (45) can be expressed as
(t) = (t)g

(t) (t)(t)(t). (46)

Therefore, the rst variation of the Lagrangian at time t is given by
/(t) = T
g
(t) + T
r
(t) + T
b
(t) V (t). (47)
Dene the trace inner product on R
33
(the linear space of 3 3 matrices) as
follows
A, B = trA
T
B. (48)
The rst variation of the potential V (which depends only on R) is given in terms
of this inner product as follows:
V (t) =
_
V (t)
R(t)
, R(t)
_
where
_
V (t)
R(t)
_
ij
=
V (t)
R(t)
ij
for i, j 1, 2, 3. Using the rst of equations (44) and the fact that symmetric
and skew-symmetric matrices are orthogonal in the trace inner product (48), this
rst variation in the potential can be expressed as
V (t) =
1
2
|(R(t)), (t)

where
|(R(t)) = R(t)
T
_
V (t)
R(t)
_

_
V (t)
R(t)
_
T
R(t). (49)
If ()

: so(3) R
3
denotes the inverse of the ()

map, then
/
g
(R(t)) =
_
|(R(t))
_

45
is the gravity gradient moment on the spacecraft, expressed in the base body
frame. The rst variation of T
g
(t) is obtained as follows:
T
g
(t) =
_
(t) + (t)g
_
T
J
g
_
(t) + (t)g
_
m
g
(t)
T
_

g
_
2
(t), (50)
where (t) is of the form given by the second of equations (44). The rst variation
of T
r
(t) is:
T
r
(t) =
_
(t) +

(t)(t)
_
T
J
r
_
(t) +

(t)(t)
+

(t)(t)
_
+ m
r
_
(t)

g
+ (t)

(t)
_
T
_
(t)

g
+ (t)

(t) + (t)

(t)
_
, (51)
where (t) is given by the rst of equations (45), and
(t) = (t) + (t)g.
The rst variation of T
b
(t) is
T
b
(t) = (t)
T
J
b
(t). (52)
The terms dependent on (t) in /(t) observing equations (51), (50), and
(52), are given by
(t)
T
(t) where (t) = J
b
(t) + (J
g
+ J
r
)(t)
(m
g
+ m
r
)
_

g
_
2
(t) + J
r

(t)(t) (53)
m
r

g
(t)

(t) + (t)

g
(t) + (t)

(t)
_
46
Substituting for (t) from equation (39) in terms of (t) and (t), we can express
(t) as follows:
(t) = (J
b
+ J
g
+ J
r
)(t) (m
g
+ m
r
)
_

g
_
2
(t)
m
r

g
(t)

+ (t)

g
+ ((t)

)
2
_
(t)
m
r

g
(t)

+ ((t)

)
2
_
(t)g
+ (J
g
+ J
r
) (t)g + J
r

(t)(t). (54)
The terms dependent on (t) in the expression for /(t) are given by
p

(t)

(t) where p

(t) = g
T
_
(J
g
+ J
r
)(t) + J
r

(t)(t)
_
+ m
r

g
(t) + (t)

(t)
_
T
(t)

g. (55)
The terms dependent on

(t) in the expression for /(t) are given by
p

(t)

(t) where p

(t) = (t)
T
J
r
_
(t) +

(t)(t)
_
(56)
The term dependent on R(t) is V (t), as given by equation (49). The terms
dependent on (t) in the expression for /(t) are given by
q

(t)(t) where q

(t) =
_
(t) +

(t)(t)
_
T
J
r

(t)g

(t)
+ m
r

_
(t)

g
+ (t)

(t)
_
T
_
(t)

(t)
_
. (57)
Consider the last term in the expression (57) for q

. Substituting for (t) from

47
equation (39) in this term, we get:
_
(t)

(t)
_
T
_
(t)

(t)
_
=
_
(t)

(t) + (t)g

(t)
_
T
_
(t)

(t)
_
=
_
(t)

(t)
_
T
_
(t)

(t)
_
=
_
(t)

(t)
_
T
_
g

(t)

(t)
_
=
_
(t)

(t)
_
T
_
g

(t)

((t) + (t)g)
_
. (58)
The above expression contains the vector triple product g

(t)

g, which is eval-
uated to be
g

(t)

g = (g
T
g)(t) (g
T
(t))g = (t),
since g and (t) are orthogonal and both are unit vectors. Substituting this in the
last line of (58) above, we obtain
_
(t)

(t)
_
T
_
(t)

(t)
_
=
_
(t)

(t)
_
T
_
g

(t)

(t) + (t)(t)
_
=
_
(t)

(t)
_
T
_
g

(t)

(t)
_
= 0. (59)
Substituting (59) in the expression (57), we get a simplied expression for q

as
follows:
q

(t) =
_
(t) +

(t)(t)
_
T
J
r

(t)g

(t)
+ m
r

_
(t)

g
_
T
_
(t)

(t)
_
. (60)
There are no terms dependent on (t) in the expression for /(t) because the
rotor is axially symmetric (e.g., a perfect cylinder) and its center of mass lies along
48
the rotor axis. Applying Hamiltons principle, the state trajectory that gives the
critical value for the action functional is given by
S =
_
T
0
/(t)dt = 0

_
T
0
_
(t)
T
_

(t) + (t) (t)

_
+ p

(t)

(t)
+ p

(t)

(t) +
1
2
|(R(t)), (t)

+ q

(t)(t)
_
dt = 0

_
T
0
_
(t)
T
_

(t) + (t) (t)

_
+ p

(t)

(t)
+ p

(t)

(t) +/
g
(R(t))
T
(t) + q

(t)(t)
_
dt = 0, (61)
where we substituted for (t) from equation (44) and used the fact that
1
2
a

, b

=
a
T
b.
6.2 Formulation of the Equations of Motion of Spacecraft with VSCMG
6.2.1 Equations Expressing the Dynamics
The dynamics equations for the spacecraft with a CMG are obtained from
the expression (61) by carrying out integration by parts, taking into account that
the xed endpoint variations imply
(0) = (T) = 0, (0) = (T) = 0,
and (0) = (T) = 0.
This integration by parts leads to
S =
_
T
0
_
_
/
g
(R(t))
d(t)
dt
(t)

(t)
_
T
(t)
+
_
q

(t)
dp

(t)
dt
_
(t)
dp

(t)
dt
(t)
_
dt = 0. (62)
49
The variations (t), (t) and (t) are continuous and vanish at the terminal
points (t = 0 and t = T), but are otherwise arbitrary in the interval [0, T].
Therefore, the expression (62) leads to the following dynamics equations for the
spacecraft with a CMG:
d(t)
dt
=(t) (t) +/
g
(R(t)), (63)
dp

(t)
dt
=q

(t), (64)
dp

(t)
dt
=0. (65)
The total angular momentum of the spacecraft with CMG, given by (t), can also
be expressed as:
(t) = (J
T
+ I
T
(t))(t) + (J
c
+ I
c
(t)) (t)g + J
r

(t)(t)
where J
c
= J
g
+ J
r
, J
T
= J
b
+ J
c
(m
g
+ m
r
)(

g
)
2
,
I
c
(t) = m
r

g
(t)

+ ((t)

)
2
_
,
I
T
(t) = I
c
(t) m
r
(t)

g
. (66)
Here J
T
denotes the total constant part of inertia in the base body frame, I
T
(t)
is the total time-varying part of inertia in the base body frame, J
c
is the constant
inertia of the CMG in the base body frame, and I
c
(t) is the time-varying part of
inertia of the CMG in the base body frame. The dynamics of the base body is
therefore aected by the rotation rates (t) and

(t) of the CMG gimbal and rotor
respectively, but is independent of the rotation angle (t) of the rotor which is a
cyclic variable for this system. Therefore, the conjugate momentum corresponding
50
to (t), i.e. p

(t), is conserved if there is no torque input acting on the rotor

axis. Note that the angular speed of the rotor

(t) is not constant in this case
due to the dynamical coupling between the dierent degrees of freedom of this
spacecraft system, as can be seen from equations (56) and (65). Therefore, to keep
a constant rotor angular speed in the base spacecraft body frame as in constant
speed SGCMGs, a control torque has to be applied to the rotor.
6.2.2 Dynamics Equations with CMG Control Torques
In the presence of a (scalar) control torque
g
(t) for the gimbal axis, the
rotational dynamics of the gimbal axis is given by
dp

(t)
dt
= q

(t) +
g
(t), (67)
which replaces equation (64) in the system dynamics. In addition, if there is an
input torque
r
(t) on the rotor, then the rotational dynamics of the rotor is given
by
dp

(t)
dt
=
r
(t). (68)
This torque input could be applied, for example, to maintain a constant angular
rate

(t) for the rotor, as is the case for constant speed control moment gyroscopes.
Note that the considerable dynamical coupling between the three rotational de-
grees of freedom of the base body and the two rotational degrees of freedom of
the CMG can be used to control the base bodys attitude motion. This system
can be categorized as an under-actuated multibody system. It should be noted
51
that when internal friction torques along the gimbal and rotor are added to the
formulation of the equations of motion, they cannot be presented as external to
the system, and therefore are best handled as a set of cascaded (inner-loop) dier-
ential equations. The cascading of the dierential equations from internal friction
such as
dp

(t)
dt
= q

(t) +
g
(t)
f
(t), (69)
with friction torque
f
(t), preserves the contribution of torque, from friction, in-
ternal to the system. This can be seen by noticing friction changes (t), which in
turn changes spacecraft body response (t), because the eect from the contribu-
tion of moment exchange from (t) is directly sent to the spacecraft bus through
(t). In the absence of the external gravity gradient torque /
g
(R(t)) in equa-
tion (63), these internal eects would not change the systems inertial angular
momentum, i.e, R(t)(t) would remain constant along the systems dynamics.
Note that in our dynamics model for the base body (spacecraft bus) dynamics,
given by (63), the inertia of the base body is time-varying. The time-varying
terms, given by equation (66), are due to the oset between the gimbal axis and
the rotors center of mass. This leads to interesting and useful properties for the
control of the spacecrafts rotational motion with such VSCMGs.
6.3 Relations between Momentum Quantities
Here we study the relations between the total angular momentum (t) of
the spacecraft system and the momentum variables associated with the internal
52
states of the VSCMG. The expression for the total angular momentum is given by
equation (66). We rst express and

in terms of the CMG momentum variables
p

and p

(thereby eliminating and

as variables).
6.3.1 Relations between Total Angular Momentum and CMG Mo-
menta
The expressions for p

and p

given by equations (55) and (56) respectively, can

be simplied as follows:
p

(t) =
T
_
J
c
m
r

g
(t)

+ ((t)

)
2
_
_
g
+ g
T
_
J
c
(t)g m
r

2
(t)
_
(t)

_
2
g
_
+ g
T
J
r

(t)(t)
= (t)g
T
_
J
c
m
r
_
(t)

_
2
_
g
+
T
_
J
c
+ I
c
(t)
_
g + g
T
J
r

(t)(t), (70)
p

(t) =(t)
T
_
J
r
(t)g + J
r
(t)
_
+

(t)(t)
T
J
r
(t). (71)
Re-arranging expressions (70) and (71), we can express and

in terms of p

and p

as follows:
_
g
T
_
J
c
+ I
r
(t)
_
g g
T
J
r
(t)
(t)
T
J
r
g (t)
T
J
r
(t)
_
_
(t)

(t)
_
=
_
p

(t) g
T
(J
c
+ I
c
(t))(t)
p

(t) (t)
T
J
r
(t)
_
, (72)
53
where
I
r
(t) = m
r
_
(t)

_
2
.
Note that the co-ecient matrix on the left side of equation (72) is positive
denite and hence invertible. Let us denote this coecient matrix by J
gr
, i.e.,
J
gr
(t) =
_
g
T
_
J
c
+ I
r
(t)
_
g g
T
J
r
(t)
(t)
T
J
r
g (t)
T
J
r
(t)
_
. (73)
We also dene the following quantities:
(t) =
_
(t)
(t)
_
, p(t) =
_
p

(t)
p

(t)
_
R
2
,
and B(t) =
_
g
T
(J
c
+ I
c
(t))
(t)
T
J
r
_
R
23
. (74)
Therefore, we can express equation (72) more compactly as:
J
gr
(t) (t) = p(t) B(t)(t). (75)
Using the notation dened in (73)-(74), we can also express the total angular
momentum of the spacecraft as:
(t) = (J
T
+ I
T
(t))(t) + B(t)
T
(t)
= J
T
+ I
T
(t) B(t)
T
J
1
gr
(t)B(t)(t)
+ B(t)
T
J
1
gr
(t)p(t)
= K
T
(t)(t) + A(t)p(t), (76)
where
K
T
(t) = J
T
+ I
T
(t) B(t)
T
J
1
gr
(t)B(t)
and A(t) = B(t)
T
J
1
gr
(t). (77)
54
The quantity A(t) in equation (77) denes the Ehresmann connection [54] on the
conguration space of the spacecraft with a CMG. As described earlier, this con-
guration space has the structure of a ber bundle with its base space coordinates
given by the CMG coordinates and and its ber space described by the atti-
tude R of the base body. In this case, the Ehresmann connection is a principal
connection on the principal bundle constituting the conguration space Q. Note
that the inertia matrix of this system, expressed in base body coordinate frame,
is given by
M(t) =
_
J
T
+ I
T
(t) B(t)
T
B(t) J
gr
(t)
_
, such that
_
(t)
p(t)
_
= M(t)
_
(t)
(t)
_
. (78)
6.4 CMG Design parameters
There are certain design parameters or specications that aect the qualitative
and quantitative performance of a CMG. Overall dimensions, structural materi-
als, mass moment of inertias, rotor and gimbal frequencies etc., can be considered
as mechanical design parameters. This section elaborates on some critical design
parameters, which are important in determining conditions, such that thegimbal
and rotor quantities does not aect the total angular velocity (and angular mo-
mentum) of the spacecraft bus. From equation (76), we see that the contribution
from the gimbal and rotor angular rates to the total angular momentum of the
spacecraft (t) can be nullied, if B(t)
T
(t) = 0, where B(t) is given by (74).
On the same line, if A(t)p(t) = 0, then the total spacecraft momenta (t) is not
aected by VSCMG quantities, directly. As mentioned in the previous subsection,
55
this implies that B(t) has a null space and (t) lies in this null space. Therefore,
the two rows of B(t) must be linearly dependent. Then,
[J
c
+ I
c
(t)]g = [J
c
+ m
r

2
]g m
r

g
(t)

g (79)
If the rows of B(t) are linearly dependent, equation (74) can be written using
(79) as,
[J
c
+ m
r

2
]g m
r

g
(t)

g = (t)J
r
(t) (80)
where (t) is scalar
Again,

g
((t)

g) =
g
((t) g) = (
T
g
g)(t) (
T
g
(t))g
Using the above identity, (80) can be written as
[J
r
+ J
g
+ m
r

2
+ m
r
(
T
g
(t))]g [m
r

T
g
+ (t)J
r
](t) = 0 (81)
The above equation can be written as,
C(t)g D(t)(t) = 0
where C(t) and D(t) are 33 symmetric matrices and therefore the column space
of C(t) and D(t) have a non-trivial intersection. Equation (81) can be expressed
as,
(J
r
+ J
g
+ m
r

2
)g (t)J
r
(t) = m
r
(
T
g
g)(t) m
r
(
T
g
g)g (82)
56
Multiplying equation (82) by g
T
and (t)
T
on both sides, successively gives
g
T
(J
r
+

J
g
)g (t)g
T
J
r
(t) = m
r
(
T
g
(t)) (83)
(t)
T
(J
r
+

J
g
)g (t)(t)
T
J
r
(t) = m
r
(
T
g
g) (84)
respectively, where

J
g
= J
g
+ m
r

2
The above equations (83)-(84) can be stacked as,
_
g
T
(J
r
+

J
g
)g g
T
J
r
(t)
(t)
T
(J
r
+

J
g
)g (t)
T
J
r
(t)
_
_
1
(t)
_
= m
r

T
g
(t)

T
g
g
_

_
Therefore, the null space of B(t) is one-dimensional and spanned by
_
_
1
(t)
_
_
= m
r

_
g
T
(J
r
+

J
g
)g g
T
J
r
(t)
(t)
T
(J
r
+

J
g
)g (t)
T
J
r
(t)
_
1
_

T
g
(t)

T
g
g
_

_
(85)
Also note from equation (82), that if (t) 0 then,
(J
r
+ J
g
+ m
r

2
)g = m
r
(
T
g
g)(t) m
r
(
T
g
g)g
However, the constant vector on the left hand side of the above equation cannot
have a constant coecient along the time-varying (t). Besides, it cannot be guar-
anteed to lie on the g(t) plane. Therefore, the second component of equation
(85) is not equals to zero and since ,= 0, we have
(t)
T
(J
r
+

J
g
)g
T
g
(t) + g
T
(J
r
+

J
g
)
T
g
g ,= 0 (86)
This is a condition that needs to be satised by
g
.
57
6.5 Generalization to Spacecraft with n VSCMGs
Here we look at the generalization of the relationship between internal (CMG)
momenta and the base bodys overall angular momentum for a spacecraft with n
VSCMGs of the type depicted in Figure 12, where n > 1. For such a spacecraft,
the conguration space Q of the dynamics is a principal bundle with base space
B = T
2
T
2
T
2
(n-fold product) or B = (T
2
)
n
, and ber space F = SO(3).
For this system, we expect the mass matrix to be of the form
M(t) =
_

_
J
T
+

n
i=1
I
i
T
(t) B
1
(t)
T
. . . B
n
(t)
T
B
1
(t) J
1
gr
(t) . . . 0
.
.
. 0 . . . 0
B
n
(t) 0 . . . J
n
gr
(t)
_

_
, (87)
which is of size (3 + 2n) (3 + 2n). This mass matrix indicates that the con-
tributions to the total angular momentum of the spacecraft from the individual
VSCMGs are given by the B
i
(t), which in turn, depend on the gimbal axis di-
rections and rotor axis directions in the base body coordinate frame. The total
angular momentum is given by
(t) =
_
J
T
+
n

i=1
I
i
T
(t)
_
(t) +
n

i=1
B
i
(t)
T

i
(t), (88)
where
i
= [
i

i
]
T
T
2
denotes the vector of angles of the ith VSCMG. The
angular momentum variables of the ith VSCMG, denoted p
i
, are given by
p
i
(t) = B
i
(t)(t) + J
i
gr
(t)
i
(t). (89)
The angular rates
i
can be uniquely determined in terms of p
i
and from
equation (89), since the J
i
gr
R
22
are positive denite. Substituting for these
58
angular rates in equation (88), the total angular momentum can be expressed in
terms of base spacecraft angular velocity and CMG momenta p
i
as follows:
(t) = K
T
(t)(t) +
n

i=1
A
i
(t)p
i
(t), (90)
where
K
T
(t) = J
T
+
n

i=1
_
I
i
T
(t) B
i
(t)
T
(J
i
gr
)
1
(t)B
i
(t)
_
and A
i
(t) = B
i
(t)
T
_
J
i
gr
(t)
_
1
. (91)
The vectors p
i
can be concatenated to form a vector P R
2n
and the A
i
R
32
can be concatenated to form a matrix / R
32n
, so that equation (90) can be
more compactly represented as
(t) = K
T
(t)(t) +/(t)P(t). (92)
From equation (92), and given desired angular velocity , one can then obtain
a relation between the vector of VSCMG momenta P(t) and the total angular
momentum (t) of the spacecraft.
59
7 Conclusion
This thesis describes a ground simulator design and simulator control laws for
simulating autonomous rendezvous and proximity operations (ARPO) of space-
craft. The simulator and control laws can be used to realistically simulate six
degrees of freedom motion of rigid spacecraft engaged in ARPO during ground
simulations. Control laws for controlling relative translational motion between a
target and a chaser spacecraft, and the attitude motion of the spacecraft, ensure
that the spacecraft models supported by these simulators can accurately mimic
the motion of spacecraft engaged in ARPO. Future extensions of this work will
be to supplement these control laws with guidance and navigation schemes for
closed-loop control of these simulators during ground simulations of spacecraft
ARPO.
Attitude control and stabilization of spacecraft using momentum exchange de-
vices are studied in great detail. A variational set of equations of motion for a
spacecraft containing a single VSCMG have been explored. From this variational
setting, some of the most complicated attributes of this system were presented
without resorting to rst principles. In addition, this formulation provided intu-
ition into the eect of cyclic coordinates appearing due to symmetries present in
the system (e.g., rotor angle is cyclic if rotor is axially symmetric and its center
of mass lies along its axis).
Future work will explore the addition of multiple VSCMGs to the system and
60
relaxed assumptions on the axial symmetry of the rotors, position and orientation
of the rotor axis with respect to the gimbal axis, non-conservative eects due to
friction, as well as eects due to exibility of spacecraft body. Additional future
work will explore the most general of these variational equations in their eect on
geometric singularities associated with the nonlinear mapping of the gimbal and
ywheel states to the spacecraft states.
61
REFERENCES
[1] W. Fehse, Automated Rendezvous and Docking of Spacecraft [Cambridge
Aerospace Series]. Cambridge University Press, 2003.
[2] D. King, Space servicing: Past, present and future, in Proceeding of the 6th
International Symposium on Articial Intelligence and Robotics Automation
in Space. Canada: AIAA, 2001.
[3] B. J.R and H. W.E, Gemini rendezvous, Journal of Spacecraft and Rockets,
vol. 3, no. 1, pp. 145147, 1966.
[4] S. D. K, Apollosoyuz test project, Society of Experimental Test Pilots,
Technical Review 2, 1974.
[5] S.-O. A. Miller D.W., Kong E.M.C, Overview of the spheres autonomous
rendezvous and docking laboratory on the international space station, Ad-
vances in the Astronautical Sciences on Guidance and Control, vol. 113, no. 1,
pp. 155169, 2003.
[6] T. E. Rumford, Demonstration of autonomous rendezvous technology (dart)
project summary, in Proceeding of SPIE. PSIE, 2003.
[7] S. B. W. I. David A. Whelan, E. A. Adler and G. M. R. Jr., Space robotic
experiments on nasdas ets vii satellite, in Proceeding of SPIE. PSIE, 2000.
[8] M. Oda, Darpa orbital express program: eecting a revolution in space-
based systems, in Proceeding of IEEE International Conference on Robotics
and Automation. Michigan: IEEE, 1999.
[9] R. G. L. R. P. K. Rekleitis I., Martin E. and D. E., Autonomous capture of
a tumbling satellite, Journal of Field Robotics, vol. 24, p. 275296, 2007.
[10] K. T. Alfriend and HanspeterSchaub, Autonomous capture of a tumbling
satellite, Journal of the Astronautical Sciences, vol. 48, pp. 249267, 2000.
[11] C. E. T., State transition matrices for terminal rendezvous studies: Brief
survey and new example, Journal of guidance, control and dynamics, vol. 21,
pp. 148156, 1998.
[12] O. M. Zhanhua Ma and B. N. Shashikanth, Optimal control for spacecraft
to rendezvous with a tumbling satellite in a close range, in Proceeding of
IEEE/RSJ, Internation Conference on Robots and systems. China: IEEE,
2006.
[13] Y. Sakawa, Trajectory planning of a free-ying robot by using the optimal
control, Optimal Control Applications and Methods, vol. 20, pp. 235248,
1999.
62
[14] S. Matsumoto, S. Jacobsen, S. Dubowsky, and Y. Ohkami, Approach plan-
ning and guidance for uncontrolled rotation satellite capture considering
avoidance, in Proceeding of 7th International Symposium on Articial Intel-
ligence and Automation in Space, Japan, 2003.
[15] S. Nakasuka and T. Fujiwara, New method of capturing tumbling object in
space and its control aspects, in Proceeding of International Conference on
Robotics and Automation, 1999, pp. 973978.
[16] N. Fitz-Coy and M. Liu, Modied proportional navigation scheme for ren-
dezvous and docking with tumbling targets: the planar case, in Proceeding
of Flight Mechanics/Estimation Theory, NASA, 1995, pp. 243252.
[17] J. Schwartz, M. Peck, and C. Hall, Historical review of air-bearing spacecraft
simulators, AIAA Journal of Guidance, Control and Dynamics, vol. 26,
no. 4, pp. 833842, 2003.
[18] J. Shen, N. H. McClamroch, and A. M. Bloch, Local equilibrium controlla-
bility of the Triaxial Attitude Control Testbed, 2002, pp. 528533.
[19] J. Shen, A. K. Sanyal, and N. H. McClamroch, Asymptotic stability of
multibody attitude systems, in Stability and Control of Dynamical Systems
with Applications, Boston, 2003, d. Liu and P. J. Antsaklis, Eds., Birkhauser.
[20] B. Agrawal and R. Rasmussen, Air-bearing based satellite attitude dynam-
ics simulator for control software research and development, in Proc. SPIE
Conference on Technologies for Synthetic Environments, 2004, pp. 204214.
[21] B.-M. Kim, E. Velenis, P. Kriengsiri, and P. Tsiotras, Designing a low-cost
spacecraft simulator, IEEE Control Systems Magazine, 2003.
[22] F. Tasker and C. G. Henshaw, Managing contact dynamics for orbital
robotic servicing missions, in AIAA SPACE 2008 Conference and Expo-
sition, San Diego, CA, Sep. 2008.
[23] K. Yoshida, Engineering test satellite VII ight experiment for space robot
dynamics and control: Theories on laboratory test beds ten years ago, now
in orbit, The Int. Journal of Robotics Research, vol. 22, no. 5, pp. 321325,
2003.
[24] H. Choset and D. Kortenkamp, Path planning and control for free-ying
inspection robot in space, Journal of Aerospace Engineering, vol. 12, no. 2,
pp. 7481, 1999.
[25] S. Matsunaga, K. Yoshihara, T. Takahashi, S. Tsurumi, and K. Ui, Ground
experiment systems for dual-manipulator-based capture of damaged satel-
lites, in Proc. IEEE Int. Conference on Intelligent Robots and Systems, Pis-
cataway, 2000, pp. 18471852.
63
[26] J. S. Hall and M. Romano, Novel robotic spacecraft simulator with mini-
control moment gyroscopes and rotating thrusters, in Proc. IEEE/ASME
Int. Coference on Advanced Intelligent Mechatronics, San Antonio, TX, Sep.
2007, pp. 16.
[27] A. Sanyal, J. Shen, and N. McClamroch, Dynamics and control of an elastic
dumbell spacecraft in a central gravitational eld, in Proceeding of 42nd
IEEE conference on Decision and Control, Maui, Hawaii, 2003, pp. 2798
2803.
[28] A. K. Sanyal, J. Shen, A. M. Bloch, and N. H. McClamroch, Stability
and stabilization of relative equilibria of dumbbell bodies in central grav-
ity, AIAA Journal of Guidance, Control and Dynamics, vol. 28, no. 5, pp.
833842, 2005.
[29] J. Shen, N. H. McClamroch, and A. M. Bloch, Local equilibrium controlla-
bility of the triaxial attitude control testbed, in Proc. 41st IEEE Conference
on Decision and Control, 2002, pp. 528533.
[30] J. Shen, A. K. Sanyal, and N. H. McClamroch, Asymptotic stability of
multibody attitude systems, in Stability and Control of Dynamical Systems
with Applications, Boston, 2003.
[31] F. Aghili, A robotic testbed for zero-g emulation of spacecraft, in Proc.
2005 IEEE/RSJ International Conference on Intelligent Robots and Systems,
2005, pp. 10331040.
[32] D. Gallardo, R. Bevilacqua, and R. Rasmussen, Advances on a 6 degrees of
freedom testbed for autonomous satellites operations, in Proceeding of AIAA
Guidance, Navigation, and Control Conference, Portland, Oregon, 2011.
[33] I. Sta of Princeton Satellite Systems, Attitude and Orbit Control Using the
Spacecraft Control Toolbox V4.0. Princeton Satellite Systems, Inc., 2000.
[34] R. Votel and D. Sinclair, Comparison of control moment gyros and reaction
wheels for small earth-observing satellites, in Proceeding of 26
th
Annual
AIAA/USU Conference on Small Satellites, Logan, UT, 2012.
[35] V. Lappas, W. Steyn, and C. Underwood, Torque amplication of control
moment gyros, Electronics Letters, vol. 38, no. 15, pp. 837 839, jul 2002.
[36] A. Defendini, P. Faucheux, P. Guay, K. Bangert, H. Heimel, M. Privat, and
R. Seiler, Control Moment GYRO CMG 15-45 S: a compact CMG product
for agile satellites in the one ton class, in 10th European Space Mechanisms
and Tribology Symposium, ser. ESA Special Publication, R. A. Harris, Ed.,
vol. 524, Sep. 2003, pp. 2731.
64
[37] X. Roser and M. Sghedoni, Control moment gyroscopes (cmgs) and their
application in future scientic missions, in Spacecraft Guidance, Navigation
and Control Systems, Proceedings of the 3rd ESA International Conference
held 26-29 November, 1996 at ESTEC, Noordwijk, the Netherlands. Edited by
Brigitte Kaldeich-Schuermann. ESA SP-381, European Space Agency, 1997,
p. 523.
[38] H. Schaub and J. L. Junkins, Analytical Mechanics of Space Systems, 2nd ed.
Reston, VA: AIAA Education Series, October 2009.
[39] J. McMahon and H. Schaub, Simplied singularity anoidance using variable
speed control moment gyroscope nullmotion, AIAA Journal of Guidance,
Control, and Dynamics, vol. 32, no. 6, pp. 19381943, Nov. Dec. 2009.
[40] H. Yoon and P. Tsiotras, Spacecraft adaptive attitude and power track-
ing with variable speed control moment gyroscopes, Journal of Guidance,
Control and Dynamics, vol. 25, pp. 10811090, 2002.
[41] H. Kurokawa, A geometric study of single gimbal control moment gyros (
singularity problems and steering law ), Agency of Industrial Technology
and Science, Japan, Tech. Rep. 175, 1998.
[42] J. Paradiso, A search-based approach to steering single gimbaled CMGs,
Draper Laboratory Technical Report, Cambridge, Massachusetts, Tech. Rep.
CSDL-R-2261, August 1991.
[43] W. . Schiehlen, Two dierent approaches for a control law of single gimbal
control moment gyros, NASA-Marshall Space Flight Center, Alabama, Tech.
Rep. NASA-TM-X-64693, 1972.
[44] F. Leve and N. Fitz-Coy, Hybrid steering logic for single-gimbal control
moment gyroscopes, Journal of guidance, control, and dynamics, vol. 33,
no. 4, pp. 12021212, 2010.
[45] A. Sanyal, L. Holguin, and S. P. Viswanathan, Guidance and control for
spacecraft autonomous chasing and close proximity maneuvers, in 7th IFAC
Symposium on Robust Control Design, Denmark, 2012, submitted.
[46] J. Craig, Introduction to Robotics Mechanics and Control, 2005.
[47] A. K. Sanyal and R. Baghaei, Realistic ground simulation of spacecraft
motion using spacecraft simulators, in Proceedings of IEEE Conference on
Control Applications, Sep. 2008.
[48] J. J. Kim and B. Agarwal, Automatic mass balancing of air-bearing-based
three-axis rotational spacecraft simulator, Journal of Guidance, Control,
And Dynamics, vol. 32, no. 3, 2009.
65
[49] J. L. Schwartz, M. A. Peck, and C. D. Hall, Historical review of air-bearing
spacecraft simulators, AIAA Journal of Guidance, Control and Dynamics,
vol. 26, no. 4, pp. 833842, 2003.
[50] P. Hughes, Spacecraft attitude dynamics. J. Wiley, 1986.
[51] B. Wie, Space Vehicle Dynamics and Control, 2nd ed. Reston, VA: American
Institute of Aeronautics and Astronautics, 2008.
[52] E. Tokar and V. Platonov, Singular surfaces in unsupported gyrodyne sys-
tems, Cosmic Research, vol. 16, pp. 547555, 1979.
[53] J. E. Marsden and T. S. Ratiu, Introduction to Mechanics and Symmetry,
2nd ed., 1999.
[54] A. M. Bloch, Nonholonomic Mechanics and Control, ser. Interdisciplinary
Texts in Mathematics, 2003, no. 24.
[55] F. Bullo and A. D. Lewis, Geometric Control of Mechanical Systems, ser.
Texts in Applied Mathematics, 2004, no. 49.
[56] G. E. Bredon, Topology and Geometry, ser. Graduate Texts in Mathematics,
1993, no. 139.
66