Generalized Linear Targeting and Guidance

This memo describes the mathematical background and practical implementation of an explicit
generalized linear targeting and guidance (GLTG) algorithm. The GLTG algorithm uses a
combination of modern control theory and linear mathematical techniques to formulate and solve a
variety of orbital transfer problems. The targeting and guidance solutions are formulated using
generalized mission constraints which relate the classical orbital elements to the state vector predicted
by the GLTG scheme. In the closed-loop guidance mode, this iterative algorithm resolves the orbital
boundary value problem during the actual finite-burn maneuvers. A key assumption for this
algorithm is that each propulsive maneuver is treated as a single impulsive delta-velocity during
targeting and as a series of fixed-attitude, finite-burn maneuvers during closed-loop guidance.

The robustness and computational speed of the generalized linear scheme make it ideal for on-board
targeting and guidance flight software. The proper combination of numerical methods makes it
possible to carry a three-degree-of-freedom (3 DOF) simulation on-board and solve the trajectory
targeting and guidance problems as the vehicle performs the mission.

In the following discussion, bold letters and symbols indicate vectors and matrices.

Equations of motion

The first-order, nonlinear equations of orbital motion are given by

x  f  x, λ 

where x is the state vector and λ is the control vector.

The inertial, Cartesian equations of motion of a spacecraft subject to gravitational and thrust forces
can be written as
 r   v 
x      
 v  g  a  t  uT 

where r is the position vector, v the velocity vector, g is the acceleration due to gravity, a  t  is the
magnitude of the thrust acceleration and uT is the unit thrust vector given by

cos  cos  
uT   sin  cos  
 sin  

In this equation,  is the in-plane steering angle and  is the out-of-plane steering angle.

The control variables used to solve the trajectory problem consist of the impulse time and three
components of each delta-v maneuver. The steering angles provided to the control system are
determined from the components of the delta-v vector.

page 1
The inertial components of a J 2  perturbed Earth gravity model are as follows;

rx  3 J 2 req  5rz2  
2

g x    3 1  1  2  
r  2 r 2  r  

ry  3 J 2 req2  5rz2  
g y    3 1  1  2  
r  2 r 2  r  

rz  3 J 2 req  5rz2  
2

g z   1   3  2 
r 3  2 r 2  r  

where

rx , ry , rz  inertial position vector components of the spacecraft

r  rx2  ry2  rz2  geocentric distance of the spacecraft
  gravitational constant of the Earth
req  equatorial radius of the Earth
J 2  oblateness gravity coefficient for the Earth

However, the equations of motion used to predict the spacecraft trajectory can include a higher order
gravity model as well as other perturbations such as aerodynamic drag, solar radiation pressure, third-
body effects and so forth.

Linearization

The equations of motion can be linearized about a reference trajectory according to

 x  F t  x  G t  λ

where
x x
F G
x λ

Furthermore,
x  0 I 
F 
x  F 0 
and
g I rrT
F     3  3 5
x r r

where F is the variation of the gravity vector due to position and can be formulated by ignoring
second order terms in the gravity vector. I is the 3 by 3 identity matrix.

page 2
Note that the F and G partial derivative matrices are functions of time and are evaluated on the
reference trajectory.

The solution to the linearized system of equations is given by

 x  t   Φ  t , t0   x  t0   Γ x  t , t0   λ

where Φ is the two-body state transition matrix and Γ x is the control matrix which satisfies

Γ x  FΓ x  G

and Γ x  t0   0 . Note that the use of velocity impulses in the GLTG algorithm permits a closed-form
solution for the Γ x matrix.

The state transition matrix Φ consists of the 36 partial derivatives of the final state vector defined by
the six components x, y , z, x , y and z with respect to the six components of the initial state vector
defined by x0 , y0 , z0 , x0 , y 0 and z0 . This matrix can be written as

 12 
  t , t0    11 
 21  22 
where
 x / x0 x / y0 x / z0 
 r  
11     y / x0 y / y0 y / z0 
 r0   z / x z / y0

z / z0 
 0

 x / x0 x / y 0 x / z0 

 r  
12      y / x0 y / y 0 y / z0 

 v 0   z / x z / y 0 z / z0 
 0

 x / x0 x / y0 x / z0 

 v  
 21     y / x0 y / y0 y / z0 
 
 r0   z / x z / y0 z / z0 
 0

 x / x0 x / y 0 x / z0 

 v  
 22     y / x0 y / y 0 y / z0 
v 
 0   z / x z / y 0 z / z0 
 0

The state transition matrix satisfies


Φ  t , t0   F  t  Φ  t , t0 
t

and Φ  t , t0   I .

page 3
Since a velocity impulse v is a discontinuity in velocity only, the state vector immediately
following an impulse time t I is simply
 r 
x  tI    
 v  v 

and the partial derivatives of the state vector with respect to this impulse are

 x 0

v  I 

The partial derivatives of the state vector with respect to the impulse time are found from the first
order variation according to

t I  t I v t I  t I  v  v   v t I 
x    g  dt   tI  g  dt   0 
tI
     

Therefore, for the i th delta-v and impulse time

x  v i 

t Ii  0 

Generalized mission constraint equations

Define the mission constraint vector y at the final time t F to be

h   rv 
y  f  xF       
 e  F  v  h   rˆ  F

Linearizing the constraint vector produces

 y  H x

where H is the constraint partial derivative matrix given by

y
H
x

The equations for the elements of this matrix can be found in Appendix A.

The solution to the linearized constraint vector can be written as

 y  H x F  HΦ x 0  HΓ x λ

page 4
 We can define a fundamental constraint-control matrix as

Γ  H Γx

The relationship between the mission constraints and control variables is

 y  Γ λ
with
 x0  0

The condition that  x0 is zero implies that a reference trajectory can be created at each targeting or
guidance update. The Γ x matrix is propagated during coast arcs using the state transition matrix.

Updated control variables

The control variable vector at successive targeting or guidance updates can be determined from

λ i 1  λ i   λ
where
 λ  Γ 1 y

The linear system defined by this last equation is solved using the Householder Orthogonal
Decomposition method. A key characteristic of this numerical method is that it can be used to solve
overdetermined, exact or underdetermined linear systems which may be encountered during different
flight phases of a complex space mission.

Mission constraints and classical orbital elements

The following table illustrates the relationship between classical orbital elements and the
corresponding mission constraint subsets. In this table, a is semimajor axis, e is orbital eccentricity, i
is orbital inclination,  is argument of periapsis, and  is the right ascension of the ascending node
(RAAN).

Table 1. Orbital Elements and Mission Constraint Subsets

orbital constraint subset

element h, r ,  h, r ,  h, hz , r,  h, hz , C3 h, hz , C3 , ez h,C3 h, e
a constrained constrained constrained constrained constrained constrained constrained
e 0 0 0 0 0 0 0
i 0 0 0 0 0 constrained constrained
 undefined undefined undefined free constrained free constrained
 undefined constrained free free free constrained* constrained*
*
undefined if i  0

page 5
 where
h angular momentum vector
e orbital eccentricity vector
r geocentric radius
 flight path angle
C3  specific orbital energy
h  angular momentum magnitude
hz  z-component of angular momentum vector
ez  z-component of eccentricity vector

This next table illustrates example constraint subsets for each classical orbital element. Constraints
for special cases such as circular or equatorial orbits are also noted.

Table 2. Example Mission Constraints

constraint elements equivalent constraint

C3 or r and v (circular orbit) a

h and C3 or r, v and  (circular orbit) e

h and hz or h  i  0  i

h, hz , ez and C3 

hx and hy 

Targeting to an Outgoing Hyperbola

For lunar and interplanetary missions, the “target specs” are usually specified by the specific orbital
energy C3 , and the right ascension (RLA) and declination (DLA) of the outgoing asymptote of the
launch or departure hyperbola. This section describes the computation of an “energy-scaled” mission
constraint useful for these types of missions.

A unit vector in the direction of the departure asymptote is given by

cos   cos   
 
sˆ   cos   sin   
 sin  
  

where
   right ascension of departure asymptote
   declination of departure asymptote

page 6
The specific orbital energy is computed from the following expression

2
C3  v 2 
r

The outgoing unit asymptote vector is given by

1  C3   1  C3  
sˆ    h  e  e   
  h  e  e 
h 2     1  C3 p   

1  C3 2  
 

Finally, the targeting or constraint vector consists of the “energy-scaled” unit asymptote vector given
by
s  C3 sˆ

which is usually evaluated at the end of the final finite-burn orbit transfer maneuver or perhaps at a
trajectory interface point (TIP) some time after final stage burnout.

Calculating the mission constraints

This section summarizes the equations used to compute the complete set of mission constraint
components from the inertial state vector. In the equations that follow, a “hat” symbol indicates a
unit vector.

The angular momentum vector is normal to the orbit plane and is given by the following right-handed
cross product
h rv

The unit angular momentum vector is given by

rv
hˆ 
rv

The angular momentum magnitude can be computed from

h  r  v  hx2  h y2  hz2   a 1  e 2 

The inertial flight path angle can be determined from

 r v 
  sin 1    sin  rˆ  vˆ 
1

 r v 

Flight path angle is positive above the local horizon and negative below. The sine of the flight path
angle is often used as a mission constraint.

page 7
Twice the specific (per unit mass) orbital energy can be determined from

2 
C3  v 2  
r a

The eccentricity vector is in the direction of perigee and is given by

vh 1  2   
e r    v   r   r v  v 
   r 

Calculating the classical orbital elements

This section describes the equations used to compute the classical angular or orientation orbital
elements from the state vector. In the equations to follow, the correct quadrant for each angular
orbital element can be determined from the corresponding sine and cosine equations using a four
quadrant inverse tangent evaluation.

The semimajor axis can be determined from the following equation:

r  h2 r 
a   
2  rv 
2

 e   
2 2

 2  r v  C3
2

The sine and cosine of orbital inclination can be determined from the following two expressions:

sin i  kˆ  hˆ

h
cos i  hˆ kˆ  z  hˆz
h

where k̂ is a geocentric unit vector in the direction of the planet’s spin axis and is given by  0 0 1 .
T

An orbital inclination constraint can be enforced using the z-component and magnitude of the angular
momentum vector according to the second equation.

A unit vector in the direction of the ascending node is given by the following cross product

nˆ  kˆ  hˆ

The sine and cosine of the right ascension of the ascending node (RAAN) can be determined from


sin   ˆi  nˆ kˆ 
cos   nˆ  ˆi

page 8
where î is a geocentric unit vector in the direction of the x-axis of the planet-centered-inertial (PCI)
coordinate system and is given by 1 0 0 . A RAAN constraint can be enforced using the x and y
T

components of the angular momentum vector according to   tan 1  hx , h y  .

The sine and cosine of the argument of periapsis for elliptical and hyperbolic orbits can be
determined from the next two expressions
sin    nˆ  eˆ hˆ
cos   hˆ eˆ

Finally, an expression for the scalar orbital eccentricity in terms of specific orbital energy and angular
momentum is given by
h C3   2 h 2
e


and the eccentricity vector which points in the direction of perigee is

vh
e  rˆ


Applications

The GLTG algorithm has been used to solve the following orbital transfer problems;

(1) Fixed one and two impulse orbital transfer

For this type of problem, the delta-v components and the impulse times are the control variables.
However, the scalar magnitude of each delta-v is fixed.

(2) Minimum one and two impulse orbital transfer

The minimum impulse orbital transfer problems are solved by wrapping an unconstrained
minimization routine around the GLTG algorithm. As in the previous example, the delta-v
components and the impulse times are the control variables. For the “outside” or minimization loop,
the optimization control variables are the impulse times and the objective function is the scalar
magnitude of the impulsive maneuver(s).

(3) One and two impulse Lambert trajectories

For the Lambert applications, the impulse times are fixed and the mission constraints consist of the
three components of the final position (one impulse) and velocity vector (two impulse).

(4) Single impulse deorbit maneuver

The mission constraints for this problem are the inertial flight path angle and geodetic altitude at the
entry interface (EI). The delta-v components and the time of the deorbit impulse are the control
variables.

page 9
References and Bibliography

“Two-body Linear Guidance Matrices”, L. P. Abrahamson and R. G. Stern, NASA CR 70465, June
1965.

“An Analysis and Evaluation of Guidance Modes”, C. G. Pfeiffer, NASA CR 113902, May 1970.

“Gamma Guidance for the Inertial Upper Stage”, J. W. Hardtla, AIAA 78-1292, August 1978.

“Linear Guidance Laws for Space Missions”, W. Tempelman, AIAA Journal of Guidance, Control
and Dynamics, Vol. 9, No. 4, July-August 1986.

“Linear and Nonlinear Perturbation Analysis Applied to the Two-body Problem”, W. Tempelman,
AIAA-1988-4216, AIAA/AAS Astrodynamics Conference, August 15-17, 1988.

page 10
 Appendix A
Elements of the H Matrix

This appendix summarizes the elements of the H matrix which represent the partial derivatives of the
mission orbit constraint vector y with respect to the state vector x.

position vector
r
 rT r 0 
x 

velocity vector
v
 0 vT v 
x

angular momentum vector

h
 I  v r  I
x

angular momentum magnitude

h
 hT  I  v  h hT  r  I  h 
x

eccentricity vector

e
  v 2   1 r  I  rrT r 3  vvT  2rvT    vr  I   vrT  
x

specific orbital energy

C3
  2 rT r 3 2 vT 
x

sine of flight path angle

 sin 
  vˆ T  I r  rrT r 3  rˆ T  I v  vvT v3  
x

scaled unit asymptote vector

s sˆ C
 C3  sˆ 3
x x x

C33  h   e   C3 e  3 C3 2  e   1 C3

   e    h     2   2  e   s  2 e
 e2  x   x   e x  2C3 x e  x   2e x

page 11