66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.

IAC-15-C1.7.6

OPTIMIZATION OF THREE DIMENSIONAL TRAJECTORY USING LEGENDRE

PSEUDO-SPECTRAL METHOD FOR LUNAR SOFT LANDING
S. Mathavaraj
ISRO Satellite Centre, Bangalore, India, mathan.hce@gmail.com
R. Pandiyan
ISRO Satellite Centre, Bangalore, India
Ritu Karidhal
ISRO Satellite Centre, Bangalore, India

ABSTRACT
The dynamics of the Lunar lander trajectory in three dimensions is formulated and optimal trajectory design
studied in an optimal control setup. The cost function considered is minimum fuel satisfying all mission constraints
envisaged. A Legendre Pseudo-spectral method is used to discretize the trajectory optimization as a nonlinear
programming problem. The method guarantees enforcing terminal mission constraints as part of the formulation. The
mission constraints considered in studying this problem are: reaching the desired landing site with desired latitude
and longitude along with desired altitude, attitude and, velocity from a specified perilune height. At touchdown, the
lander orientation should be vertically up favouring landing leg of the Lander to touch the Moons surface as desired.
The commanded acceleration should be within the limit of the main engine thruster capability and saturation limit
which enforces maximum and minimum thrust constraints. In order to generate the precise trajectories to land at
desired site, the co-ordinates of the landing site must be considered as terminal constraint and three dimensional
approach is a must since landing site is determined by latitude and longitude of the Moon. Moons angular rotational
velocity is also accounted in the model because the landing site moves relative to an inertial frame.
Legendre PS method was applied to approximate the state as well as control variables and the state
differential equations as algebraic equations using a PS differentiation matrix. After converting the optimization
problem into NLP, Sequential Quadratic Programming is used to solve for optimal solutions. Control variables in
the formulation are thrust and its direction (right ascension and declination). Optimal solutions of the above problem
are those control variables which dynamically satisfying all mission constraints. Because the initial longitude,
latitude and transverse velocity at optimal perilune altitude affect the design of the optimal trajectory for lunar
landing for a desired landing site, these values are derived from the optimization results. Based on these information,
de-boost can be designed to reach these initial conditions at perilune. The fmincon in MATLAB (R) is exploited in
solving for the feasible solution. The simulation results that are presented clearly establish that the designed
trajectory meets all the constraints enforced.

I. INTRODUCTION was recently confirmed by the impact of NASAs Lunar

Precise landing of lunar modules to Moon has
been an interesting research topic ever since Man landed
on the Moon and has realized that as a base, Moon
provides great many opportunities to reach outer planets
with ease. Finally, Moon has been recognized as the
important destination for space science and exploration.
In order to achieve this requirement, it is necessary to
develop skills to soft land the supplies and sensitive
instruments on to the Moon precisely and accurately.
The major technological objectives of a soft landing
lunar mission are to assure precise landing, soft touch
down and retargeting for hazard avoidance.
Further to that, there are several reasons for the
interest in Moon exploration. The moon has an
abundant supply of Helium-3 which is a potential
energy source in the future for use in nuclear fusion.
The Moon has also been base station to reach other
planets. In addition, the existence of water on the Moon
crater observation and Sensing satellite and the
confirmation of presence of water is expected to provide
enormous assistance when a lunar base is constructed on
Moons surface.
During atmospheric entry and landing, there
has been unique features that are to be encountered and
conquered for efficient landing. Entry into planets
with atmosphere can be done by techniques like aero
braking, aero capture, ballistic entry, skip entry, gliding
entry etc. However, when entry is attempted into
celestial bodies without atmosphere, not only the
advantage one gets in reducing the velocity using the
drag force is lost but also entirely the reduction should
take place using the fuel one carries onboard which is
planned and limited. Therefore, a careful plan and
execution of the burns is very important factor in
landing the modules in planets without atmosphere.

IAC-15-C1.7.6 Page 1 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.
The space probe orbiting the Moon need to
land on the surface satisfying all mission constraints
while complying vehicle limitation and capabilities. For
this, guidance trajectory is generated by treating space
probe as a point mass and the entire problem as
optimization problem. After the success of
Chandrayaan-I, Moon has been seriously considered as
a base for extended missions. For reducing the energy
of the vehicle orbiting, one has to depend on the
onboard propulsion system to de-orbit, descend and soft
land on the Moon. Therefore, optimum guidance
algorithm is needed for efficient use of onboard fuel for
achieving the goal. To add to that woes, out of plane
maneuver of the vehicle will also be very costly and
therefore precise and careful de-orbiting manoeuvre is a
must to achieve landing to specific sites on the planet.
The various phases for a lunar module to land on the
Moon are provided in the next section followed by a
literature review of some of the published works.
I.I Phases in Moon Landing
There are three distinct phases that need to be
performed in order to land the lunar module on to the
surface of Moon. They are: (i) De-orbit Maneuver
phase (ii) Transfer orbit phase or coasting phase and
(iii) Powered descent phase. The parking orbit of the
lunar spacecraft is planned to be 100 Km circular for
this study. Otherwise, the method developed is general
in nature and can be applied to other types of orbit as
well.
Deorbit maneuver: This de-orbit phase is essential to
bring the vehicle from the parking orbit of 100 km to a
suitable perigee height, say 18 Km. Thruster firing is
executed based on the principles of Hohmann transfer
between the parking orbit and the descent orbit.
Transfer orbit: This phase consists of transition of lunar
module from the parking orbit to descent orbit. It is to
be noted that the this manoeuvre is carried out in such a
way that the desired initial condition of the powered
descent phase has been achieved in terms of position,
velocity, and attitude. This is one of the main
requirements in order to land the module precisely to a
landing site; otherwise retargeting and out of plane
manoeuvres will result in precious fuel expenditure.
Powered Descent: This phase starts from the desired
terminal condition of the transfer orbit. The main
thruster is used to absorb the orbital energy to achieve
soft landing and also rigid body dynamics and attitude
control algorithm is implemented suitably to configure
and land stably. In this phase, goal of the guidance
algorithm is to safe land the lunar module at desired
landing site by computing the acceleration which has to
be met by attitude control system.
IAC-15-C1.7.6
In this paper, powered descent phase is
analyzed assuming that de-orbit manoeuvre phase,
transfer orbit phase is carried out resulting in assumed
initial condition for descent phase and also assuming
that attitude control system is capable of meeting the
guidance demands along the entire trajectory.
I.II. Literature Review
Vijith et al. [1] proposed optimal design of soft
landing trajectories of a probe from a lunar parking orbit
with minimum fuel consumption and hence maximizing
the landing mass of the probe. The problem is
formulated as an optimal control problem with thrust
direction as control variable. Using the maximum
principle of Pontryagin, the problem is converted into a
two-point boundary value problem (TPBVP). A
sequential quadratic programming (SQP) algorithm in a
suitable form is used to solve the TPBVP. For liquid
propulsion, simulations for different thrust levels are
conducted and its effect on the optimal soft landing
trajectory is studied.
Alisa [2] presented a detailed investigation of
the constrained trajectory optimization of the Moon
landing problem. Assumptions include a spherical,
homogeneous, and non-rotating Moon, a vehicle with
one throttlable fixed main engine with a constant
exhaust velocity, and a purely Newtonian gravity
model. The trajectory was examined from an initial
circular parking orbit of 40 Km and descended to a soft-
landing on the surface. Constraints were progressively
added to mitigate the operationally undesirable
characteristics of the obtained solutions. The resulting
trajectories were analyzed based on state and control
histories, effect on fuel and V usages, operational
feasibility.
Alexander I. Kozynchenko [3] addresses a
problem of expanding the attainable downrange area for
a lunar landing module when guiding to a specified
point. Both the conventional boundary conditions and
the restrictions on fuel consumption and throttling
coefficient are taken into account. More specifically, the
purpose of algorithm development is to provide the
equality of the available thrust margins over the
maximal remaining part of a descent trajectory at the
braking phase. Three kinds of predictive algorithms are
considered: the first one having a constant thrust over a
trajectory and not providing the equal thrust margins,
the second and third have a stepped thrust profiles and a
jump distance as a control. The first and second, so
called planning algorithms are able to generate the
complete control profiles at the beginning of the braking
phase, whereas the third one, gradual or spreading-
over-path, computes the control profile in real-time
fashion gradually when moving over the trajectory.
Chao et al. [4] presented line-of-sight homing
technology to achieve more precise lunar landing in
Page 2 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.
their paper. Simulation results show that the presented
guidance and control laws are effective. Hardware
realization concept of precise landing mission is
discussed in detail and by making use of the lunar
gravity, thruster fire times can be reduced which will
contribute to save precious fuel. It is claimed that with
the development of electronics and photo-electricity,
precision lunar landing could be realized by micro, low
cost, low power hardware.
Dong et al. [5] in their paper studied the
perilune altitude of the intermediate orbit with optimal
thrust programming for minimal energy. The perilune
altitude and optimal thrust programming can be
expressed by a function of the radius of a parking orbit
by using continuation method and co-state estimator.
For designing optimal lunar landing trajectories,
perilune altitude turned out to be important in reducing
the total energy for the lunar landing. Therefore, the
initial state of descent phase is free in this optimal
control problem. Using shooting method and the
continuation approach, the optimal solution under
various parking orbit conditions has been found.
Rekha et al. [6] focuses on the navigation
system required for the Lunar Soft Lander which is a
complete, integrated, autonomous system with multiple
sensor information for precise and soft landing. The
challenges in the navigation system design for safe,
precise, soft and autonomous landing on the Moons
surface are discussed. The Navigation Guidance Control
(NGC) system has to perform several functions that are
critical to the soft landing. The NGC configuration
during various phases of the mission is also described.
The mission profile including the navigation plan from
the time of lift off till landing on moons surface is
briefed. The NGC sensor suite was selected catering to
the functional requirements of the mission. The
selection of specific sensors during various phases is
also described.
A logical development of a novel, minimum-
complexity guidance system for precise and soft lunar
landing is presented by Vaeth et al., [7] and they
provide an evaluation of predominant error sensitivities.
A new trajectory control technique presented in this
paper allows excellent compromise between sensor
requirements (e.g., Doppler and beacon tracking radar
gimbaling), control system complexity and total fuel
usage.
Badak et al. [8] gave an overview of the timing
and V implications for key activities during the lunar
landing approach phase especially with crew within the
module. When human beings are in, additional
considerations are to be accounted for while landing on
the Moon.
Allan Y. Lee [9] proposed an optimal control
law for a lander to achieve a soft touchdown. For
simplicity, the vehicle is modeled as a point mass and
IAC-15-C1.7.6
only two-dimensional planner motion is considered. In
this study, a cost functional that penalizes both the
terminal vertical and horizontal velocities of the vehicle
as well as the fuel cost of the descent main propulsion
system is considered.
Bong et al. [10] address a minimum-fuel, two-
dimensional trajectory optimization for a soft lunar
landing from a parking orbit to a desired landing site.
Designing precise trajectories for landing at a desired
site demand the landing site to be considered as the
terminal constraint in the formulation.
Zhou et al. [11] considered an optimal control
problem arising from the guidance of a lunar module to
achieve soft landing, where the description of the
system dynamics is in a three-dimensional coordinate
system. Optimal guidance law for soft landing of the
lunar module with the terminal attitude of the module to
be within a small deviation from being vertical with
respect to lunar surface has been proposed.
The above literature review gives an idea of
various designs carried out for lunar landing missions.
Research has been carried out using point mass model
representing the lander craft. Some researchers were
interested only in the terminal descent phase assuming
that the initial starting point can be precisely arrived at.
These studies consider only two dimensional analysis
since out of plane maneuver is too costly in terms of
fuel. However, in hindsight, mission constraint of
landing at particular landing site demands three-
dimensional trajectory optimization as well. It is found
that the algorithm with point mass model provides
utmost accuracy and comparable solutions for the soft
lunar landing problem from a lunar parking orbit.
I.III Organisation of the paper
The terminal phase of the lander problem only
is considered and the de-orbit phase and transfer-orbit
phase have been tacitly assumed solved suitably. The
equation of motion used for optimal control formulation
is three dimensional. By optimal control formulation all
mission constraint has been formulated and included in
the design. An outline of paper is given below.
The equation of motion for guidance is derived
in Section 2. A point mass based derivation in spherical
coordinates has been presented.
In Section 3, a generic theory of Legendre
pseudo-spectral method is discussed and formulation is
presented to explain the implementation for the Moon
landing problem.
The results obtained from the trajectory
optimization methods of Legendre Pseudo spectral
algorithm of the Moon landing problem are presented in
Section 4. The parameters selected and the inferences
from the simulations have been presented in detail.
Finally, the results and conclusions of the work
is consolidated and provided in Section 5.
Page 3 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.

II. GOVERNING DYNAMICS control variables at the arbitrary subintervals. However,

II.I Governing Point Mass Dynamics for Guidance
In order to describe the motion of a dynamic system
it is necessary to define a suitable coordinate system and
formulate equations for the motion in accordance with
physical laws governing the system.
Considering the equation of motion of a vehicle
considered as a point mass flying inside a planetary
atmosphere. The motion of the vehicle is defined as
rw
u
r cos
v
r X t
the Legendre PS method uses globally interpolating
Lagrange polynomials as trial functions to approximate
the state variables, control variables, and state
differential equations at the nodes, which are called
Legendre-Gauss-Lobatto (LGL) points. These points
are given by tl , l 0,.., N
t0 1,
tN 1
The other LGL points 1 l N are the zeros
of the derivative of the Legendre polynomial. The
transformation from the real time domain to the LGL
time domain is required because the LGL points are
defined between [-1 1].
As stated we start by approximating the
continuous state and control variables by N degree
polynomial of the form
N
X N t X tl l t

w
T sin
2


u 2 v 2 2u cos l 0

2
N
m r r r cos
2 U t U N t U tl l t
l 0
T cos uw [2]

cos uv tan 2 w cos Equation 10 is Lagrange polynomials of
u order which interpolate the functions at the LGL points.
m r 2v sin It can be shown that

T sin vw
l t
1 t 2 1 LN t
2 N N 1 LN tl t tl
cos u tan 2u sin
v
r sin cos 1 if l k
2
m r
l t
T 0 if l k
m
I sp g [3]
[1] From this property of it follows that
X N tl X t l
In equation 1, kinematics of a point mass is [4]
included and the two control parameters considered are U N tl U tl
thrust and its orientation. The dynamics represent the
three dimensional motion of the point mass model. For To carry out the approximation of the LHS of the
further details one can refer to [12]. state equations, we impose the condition that the
approximations above satisfy the differential equations
III. DETAILS OF GUIDANCE SCHEME at the LGL collocation points. To express the derivative
in terms of at the collocation points we differentiate and
III.I. Legendres Pseudo-spectral Method evaluate the result to obtain matrix multiplication of the
The lunar craft is de-boosted from a circular following form
N N
X tk X tl l tk Dkl X tl Ck
orbit of 100km. De-boosted lander craft enters into
Hohmann transfer trajectory from 100Km to 18 Km.
l 0 l 0
At perilune height of 18 km, probe enters into powered
[5]
descent phase. Near this phase, the propulsive engines
are turned on and the probe velocity is decreased in Dkl l tk are entries of the
order to enable the craft to soft land on the lunar
surface. N 1 N 1 differentiation matrix
Collocation methods usually use piecewise-
continuous functions to approximate the state and

IAC-15-C1.7.6 Page 4 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.

LN tk equipment for lunar explorations and to return in the

k l event that the mission cannot be completed.
LN tl tk tl The dynamical equation 1-2 is formulated as
N N 1 NLP problem as explained above resulting in

D : Dkl : k l 0
4 N
i
D r t wk 0
f

N N 1
kl l
k l N l 0 2

0
4 N
i u
D t 0
f
otherwise
r cos k
kl l
[6] l 0 2
N
i v
D t 0
Next the RHS of the state equations is f
kl l
discretized by substituting equation 13 and using Gauss- l 0 2 r k
Lobatto integration rule. Thus the optimal control
problem is approximated by the following nonlinear N
i
Dkl w tl
f
optimization problem:
0
l 0 2

J N X N , f
f
T sin u 2 v2
2 2
N m r r
X t ,U t , t w
k k k k 2u cos 0
k 0
r 2 cos 2
[7]
k
2 1
wk
N N 1 LN tk 2
N
i
D u t
f
kl l
2
l 0
f 0
f X t k , U t k , t k Ck 0 T cos uw
2
cos uv tan
gl g X tk ,U tk , tk g u m r 0

0 l g X t0 , U t0 , t0 0 u 2 w cos
2v sin

fl g X t f ,U t f , t f fu k

[8] N
i
D v t
f
For more details on Legendre PS method refer [13]. kl l
l 0 2
T sin vw
III.II. Problem Specific Formulation 2
For the optimization algorithm, all state cos u tan
variables should be of same unit. So we do the m r 0
normalization using these basic variables
Vnorm sqrt / rini 2u sin
r 2 sin cos
d norm rm k
mnorm mini N
i T
D m t 0 [9]
f
[17] kl l
So the thrust, gravity, time, omega, gravitational l 0 2 I sp g e k
parameter can be normalized using above terms.
A space probe has to be minimizing fuel k 0,..., N
consumption if it is to carry a heavy payload of mission

IAC-15-C1.7.6 Page 5 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.

The radial constraint is given such that the problem has no solution for the parameter specified.
However, as part of the verification, we can use the
radius is always greater than radius of the Moon. The control profile to propagate the problem dynamics with
fuel mass constraint is an inequality constraint and small time step and the results can be validated.
should be greater than zero.
IV.II Optimal Control Constraint
rm r 0
m fuel 0 Initial condition for these dynamics is tabulated
in Table 1. The dynamics derived above is propagated
[10]
through RK-4 with parameters given in Table 2. In
The other constraints for above nonlinear
formulation, these condition become constraint for
programming problem are given below,
optimal control algorithm. The path constraints are
r t f rm 0 minimum time, minimum fuel consumption which is
also accounted in the formulation.
m t f mf 0
m t f mf 0 VARIABLES VALUES

w t f wf 0
Radial Vel. 1.69
Tangent Vel. 0
v t f v f 0 Cross Vel. 0
u t f u f 0
[11] Altitude 18.2
Table 1 Initial Constraint for Optimal Control
Problem

IV. IMPLICIT GUIDANCE NUMERICAL

RESULTS VARIABLES VALUES
Simulation analysis has been carried out using Thrust 3200
optimal control formulation explained earlier. The Gravity 1.62
platform used for simulation is MATLAB. The problem Gravitational 4902.779E9
is stated as mission requirement of landing is at equator. parameter
But due to inaccuracies of the de-boosting, at perigee Moon radius 1737.1
the lander is offset from the equator. The problem is to Rotational velocity 2.672E-6
ensure the lander craft to land on equator with zero Final time 420
touchdown velocity. Lat. Landing site 0
Long. Landing Site 30
IV.I. Legendre Pseudo-spectral Method Legendre degree 7
In this method, the problem is formulated as Time step 0.5E-5
explained in Section III.II. The optical control problem
is converted into nonlinear programming problem. The Table 2 Design Parameter for Legendre PS method
problem consists of set of equality and non-equality
bounds. In MATLAB, there is fmincon function to
handle such equality and non-equality nonlinear Figure 1 depicts the lander craft descent altitude due
constraint. Since the function is inbuilt, we have to to the firing of the engine. Figure 2-4 shows the velocity
define bounds in understandable format. Normalization components of the lander craft reaching the mission
is needed so that the algorithm searches for feasible constraints of ensuring soft landing. Figure 5 is the
solution in same non-dimensional domain which has demanded thrust magnitude for satisfying the mission
been discussed in the section III.II. constraints. Figure 6-7 depicts the thrust vector direction
i.e. pitch and yaw angle with terminal constraints
Set of initial conditions is provided for the satisfied making sure of vertical landing at desired site.
algorithm as guess variables though they didnt need to Figure 8-9 shows the latitude and longitude traced by
satisfy any constraints. The algorithm is set to use SQP the lander craft with crafter reaching desired landing site
for its optimization search. Once the final error is less at final time.
than the tolerance value specified, the computation
stops. If the tolerance value is never met then the

IAC-15-C1.7.6 Page 6 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.

Figure 1 Altitude Profile using Legendre PS method Figure 4 Cross Velocity Profile using Legendre PS
method

Figure 2 Radial Velocity Profile using Legendre PS

method
Figure 5 Thrust Profile using Legendre PS method

Figure 3 Tangential Velocity Profile using Legendre

PS method Figure 6 Pitch Profile using Legendre PS method

IAC-15-C1.7.6 Page 7 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.
Figure 7 Yaw Profile using Legendre PS method
Figure 8 Longitude Profile using Legendre PS
method
Figure 9 Latitude Profile using Legendre PS method
IV.III Inferences
In this LPS method since all constraints
dynamic and terminal are converted into equality and
inequality constraints, the algorithm can take into
IAC-15-C1.7.6
account of additional constraints of desired landing site.
The offset in latitude trajectory at perilune start from the
equator in Figure 9 has been corrected on its way along
with other mission constraint. Legendre PS method is
mathematically complex. The number of Legendre-
Gauss-Lobatto (LGL) points determines the
dimensional complexity of the problem. If number of
points chosen is less, then the solution is not smooth. So
optimal number of points is chosen based on
experience; no theoretical proof is available. Mission
constraint of reaching latitude and longitude of landing
site is assured by manipulation of control thrust vector
magnitude as well as direction.
V.CONCLUSIONS
Simulation studies of a three dimensional trajectory
design for a soft and precise Lunar lander on specified
site with equality and inequality constraints are
presented. The mission constraints considered in
studying this problem are: reaching the desired landing
site with desired latitude and longitude along with
desired altitude, attitude and, velocity from a specified
perilune height. At touchdown, the lander orientation
should be vertically up favoring landing leg of the
Lander to touch the Moons surface as desired. The
commanded acceleration should be within the limit of
the main engine thruster capability and saturation limit
which enforces maximum and minimum thrust
constraints. The simulation results that are presented
clearly establish that the designed trajectory meets all
the constraints enforced with desired latitude and
longitude landing site.
ACKNOWLEDGEMENTS
The authors thank Director, ISRO Satellite Centre,
Bangalore and Deputy Director, Mission Development
Area, ISRO Satellite Centre, Bangalore for their
continuous support and encouragement while carrying
out this work.
VI.REFERENCES
1. Vijith Mukundan, Golak Prasad Sahoo and R.V.
Ramanan, Optimal moon landing Trajectory
Design with Solid and Liquid propulsion Using
SQP, VSSC, Thiruvananthapuram.
2. Alisa Michelle Hawkins, Constrained trajectory
Optimization of a Soft Lunar landing from a
Parking Orbit, Master of Science in Aeronautics
and Astronautics, MIT, 2005.
3. Alexander I. Kozynchenko, Enhancing the
Manoeuvring Capabilities of a Lunar Landing
Module using Predictive Guidance Algorithms,
Acta Astronautica, Vol 67, pp 406-416, 2010.
4. Chao Bei and Wei Zhang, A Guidance and
Control Solution for Small lunar probe precise-
Page 8 of 9
66th International Astronautical Congress, Jerusalem, Israel. Copyright 2015 by the International Astronautical Federation. All rights reserved.
landing mission, Acta Astronautica, Vol 62, pp
44-47, 2008.
5. Dong-Hun Cho, Boyoung Jeong, Donghun Lee and
Hyochoong Bang, Optimal Perilune Altitude of
Lunar Landing Trajectory, International Journal
of Aeronautical and Space Sciences, Vol 10 (1),
May 2009.
6. A. R. Rekha, A. Abdul Shukkoor and P. P.
Mohanlal, Challenges in Navigation System
design for Lunar Soft Landing, Trivandrum
Conference, 2012.
7. J. E. Vaeth and M. D. Sarles, A Lunar landing
Guidance System for Soft-Precision landings,
International Convention on Space Electronics and
telemetry, 1964.
8. Badak E. Cohanim, Thomas J. Fill, Approach
phase considerations for Lunar Landing,
IEEEAC paper#1195, Version 3, 2009.
IAC-15-C1.7.6
9. AllanY.Lee, Optimal Terminal Descent Guidance
Logic To Achieve a Soft Lunar Touchdown,
California Institute of Technology.
10. Bong-Gyun Park, Jong-Sun Ahn and Min-Jea Tahk,
Two-Dimensional Trajectory Optimization for
Soft Lunar Landing Considering a Landing site,
International Journal of Aeronautical & Space
Science, Vol 12 (3), pp 288-295, 2011.
11. J. Y. Zhou, K. L. Teo K, D. Zhou and G. H. Zhao,
Optimal Guidance for Lunar Module Soft
landing, Non-linear dynamics and Systems theory,
Vol 10 (2), pp 189-201, 2010.
12. N. X. Vinh, A. Busemann and R. D. Culp,
Hypersonic and Planetary Entry Flight
Mechanics, Univ. of Michigan Press, Ann Arbor,
MI, 1980.
13. Fariba Fahroo and Michael Ross, Pseudospectral
method knotting methods for solving optimal
control problems, Journal of Guidance, Control,
and Dynamics, Vol 27, pp 397-405, 2004.
Page 9 of 9