Astrodynamics 102
Part 1: Lambert+ Algorithm
Gim Der
DerAstrodynamics
July 5, 2013 AMOS
September 11~14, 2011
Launch / Impact Point Prediction
Satellite and Missile Targeting
Mission Planning
Orbit Maneuvers
Station-keeping
Initial Orbit Determination
UCT Cataloging
. . . . . .
: Lambert: v (t ) , v (t ) , Lambert : v (t ) , v (t )
2
1
1 21
2
, inc_to
1 2+Find
Given: r (t ) , r (t ) , t , t
Lambert Problem
For each inc_to:
Single revolution has one solution
Multi-revolution (N) may have
( 2N + 1 ) solutions
r (t )
v (t )2
r (t )1
v (t )1
1
object
at t2
2
2-Body
Transfer
Orbit
Orbit
Sphere /
Spheroid
objectat t
v (t )2
Kepler+Transfer
v (t )1
Analytic Astrodynamics Overview
Astrodynamics 101: Kepler+ Algorithm
Part1: Analytic Prediction Algorithms
Part2: Verifications
Astrodynamics 102: Lambert+ Algorithm
Part1: Analytic Multi-revolution Targeting Algorithms
(Orbit Determination for Radar Data)
Part2: Verifications
Astrodynamics 103: Gauss/Laplace+ Algorithm
Part1: Analytic Angles-only Algorithms
(Orbit Determination for Optical Sensor Data)
Part2: Verifications
Astrodynamics 102: Lambert+ Algorithm
Part 1. Analytic Multi-rev Targeting Algorithms
1. What, Why, How
2. Physics and Mathematics
3. Lambert Algorithm Implementation
4. Analytic Lambert Solutions
5. Applications for SSA
Part 2. Verifications
6. Matter of Reference
7. Numerical Examples
Satellites Rendezvous and Docking
2007
Space Shuttle Discovery and ISS
Chinese Spacecraft and CSS
Radar/Laser Weather/Satellite Tracking
Radar
Beam
Laser
Beam
Satellite
tracking
by Radar
Satellite
tracking
by Laser
Weather
forecast
by Radar
Radar Missile Tracking
2007 2009, . . , 2013
Space Debris Collisions
2007 2009, . . , 2013
2007 2009, . . , 2013
Space Debris and Satellite Growth
What? Satellite rendezvous and docking, weather
forecast, tracking of satellites, missiles
and debris, mission planning, . . . . . , SSA
are many challenging problems
Why? Need accurate, fast and robust utility
algorithms for multiple applications
How? Understand the Physics and Mathematics
of Astrodynamics for analytic orbit
determination using radar and laser data
1. What, Why, How
Solving These Challenging Problems
Requires Analytic Lambert+ Algorithm
Lambert and Lambert+ Algorithms
Lambert (2-Body) solution Lambert+ solution for SSA
(accurate and fast )
Sphere
r (t )
v (t )2
r (t )1
v (t )1
1
objectat t
object
at t2
2
2-Body
Transfer
Orbit
Given: r (t ) , r (t ) , t , t
Find: Lambert: v (t ) , v (t ) ,
2
1
1 21
2
Lambert Problem
For each inc_to:
Single revolution has one solution
Multi-revolution (N) may have
( 2N + 1 ) solutions
, inc_to
+1 2Lambert : v (t ) , v (t )
r (t )
v (t )2v (t )
1
object
at t2
2
Kepler+ Transfer
Orbit
Spheroid
2-Body
r (t )1
1
objectat t
( J , J , J , J , J , . .
Sun , Moon , Drag , . . )32 4 3122
add perturbations analytically
2. Physics
Physics and Mathematics
Five SSA Features of Lambert+
Classical Lambert+ Lambert
Inclination indication of transfer orbit,
inc_to, specified NO/maybe YES
Bounded independent variable NO/maybe YES
Robust multi-revolution capability NO/maybe YES
Accurate iterative method YES YES
Perturbations compliant and analytic NO YES
Inclination Indication of Transfer Orbit
Choice of additional input
for Lambert problem:
Better for
multi-rev
transfer
method
or transfer
motion
direction
Inclination
Indication of
Transfer orbit,
Inc_to
vs.
(Escobal) (Others)
r1
r2
h = r x r21 k = (0, 0, 1)
direct transfer orbit
(inclination <= 90 )
r1
r2
h = r x r21
k = (0, 0, 1)
retrograde transfer orbit
(inclination > 90 )
r1
r2
h = r x r21 k = (0, 0, 1)
oo
o = 2
= k h = k h
o = <
o = 2
o = <
r1
r2
h = r x r21
Inclination indication
of transfer orbit,
inc_to = 1, posigrade
= 1, retrograde
in-plane
in-plane
out-of-plane
out-of-plane
= cos< o
1r r
2
1r r
2[ ]
Bounded Independent Variable
Solution is
guaranteed
Uncertainties in: 1. Initial guess of x
2. Convergence
x
Poor choice of independent variable:
1. x is unbounded
2. Slope d/dx is unfit for most
iterative method
= is giveno
x
Good choice of independent variable:
1. x is bounded
2. Slope d/dx is fit for many
iterative methods
2 solutions
1 solution
0 solution
Slope
2 solutions?o o
Bounded x Unbounded x
(d/dx)
non-dimensional
transfer time
Robust Multi-revolution Capability
Choice of bounded x, gives a
“vertical U”, allowing A be found easily.
Then with given, the number of
solutions, 0, 1, or 2 can be determined.
Wrong choice of x leads
to difficulties of finding A,
on a “horizontal U”
Multi-revolution, N > 0, needs to determine the minimum time point, A
(If x is the semi-
major axis, then
it is unbounded)
x
2 solutions
1 solution
0 solutionx
# of solutions?
+1-1 0 0
x x
Bounded Unbounded
N > 0
N > 0
A A ?
ooo = is given
flat curve
(If x is the “path
parameter”, then
it is bounded)
Vertical U
Horizontal U
0
Accurate Laguerre Iterative Method
Solve
with
where , , are known, and
Vary as needed n = 2, 3, . . . ,
A micro-second slower, but convergence assured
Perturbations Compliant Analytic Lambert+
Astrodynamics 101
Vinti
(J2, J3, J4
included)
Astrodynamics 102
Lambert+ (J2, J3, J4
and other
perturbations))
Classical Lambert
(2-Body) +
Targeting
by Kepler+
Kepler+ (J2, J3, J4
and other
perturbations))
Classical Kepler
(2-Body)
3. Analytic Lambert Algorithms
Lambert Algorithmic
Implementations
Classical Lambert Algorithm
r (t )2
v (t )
r (t )1
v (t )1
object at t1
object at t2 2
2-Body
Transfer
Orbit
Spherical
Earth
Equations of Motion (2-Body)
1. Classical and Universal Lambert Equations
ii+1 i
22ii i i i
i
n F (x )for i 1, 2, ..
F (x )F (x ) (n 1) F (x ) n (n 1) F (x )F (x )
F (x )
x x
2. Implementation by Laguerre Iterative Equation
Lambert Algorithm
3aF (a ) = [ ( sin ) sin ) ] t 0
F ( x ) = ( x ) y ) 0
(single revolution)
(fixed n)
2
2 3
d=
d t r
rr
Given: r (t ) , r (t ) , t , t , inc_to
Find: v (t ) , v (t )
2
1
1 21
2
Lambert Problem
inc_to given: one solution
if not given: two solutions
Note:
Multi-revolution Lambert Algorithm
References: Sun, F.T., “On The Minimum Time Traj . . ”, AAS 79-164
Der, G. J., “The Superior Lambert Algorithm”, AMOS,2011
Lambert's Equation for multi-revolutions (Sun)
Iterative Equation (Laguerre)
F ( x ) = (x ) y ) N 0
ii+1 i
22ii i i i
i
n F (x )for i 1, 2, ..
F (x )F (x ) (n 1) F (x ) n (n 1) F (x )F (x )
F (x )
x x
2-Body Lambert Algorithm
v (t )
object at t2
2
Transfer
Orbit 2
v (t )1
r (t )1
r (t )2
Transfer
Orbit 1
v (t )1
v (t )2
Sphere
object at t1
2
1
1 21
2
Lambert Problem
Given: r (t ) , r (t ) , t , t , inc_to
Find: v (t ) , v (t )
Transfer orbit 1: inc_to = 1, i < 90o
Transfer orbit 2: inc_to = 1, i > 90o
Multi- revolutions (N)
may have ( 2N + 1 ) solutions
for each inc_to and a given
Targeting to Lambert+
Celestial Pole
Final Keplertrajectory
k
ji
Final position at t
Central
body
v2
Given: r , r , t = t t , Computed v , v (Lambert) 1 2 t1 t22 1
Find: v , v (Lambert )v1 v2
1
2r
v2
v1 vt1
vt2
v1
2
1
C
A
B
C
A
InitialKepler
Lambert
Step 1
Step 2Initialposition at t
A
vv1
vv2
v1
FinalKepler
r2
r2
Lamberttrajectory
+
+
+
+
v2
r
+ 2
Targeting by
Kepler at t
4. Analytic Lambert Solutions
Lambert Solutions
and
Orbit Determination
Single Revolution Lambert Solutions
Sun, F.T., AAS Paper 79-164,
“On the minimum time trajectory and
multiple solutions of Lambert problem”
0 < < , 0 < < 1
Transfer orbit 1: inc_to = 1, i < 90o
Transfer orbit 2: inc_to = 1, i > 90o
Single revolution has one solution
for each inc_to and a given
v (t )
object at t2
2
Transfer
Orbit 2
v (t )1
r (t )1
r (t )2
Transfer
Orbit 1
v (t )1
v (t )2
Sphere
object at t1N
orm
aliz
ed T
ime
Elliptic Orbits
Path parameter,1 < x < 1
x
High Path Low Path
Parabolic Orbits
Path parameter, x = 1
Hyperbolic Orbits
Path parameter, x > 1
3
3
1/
2
N = 0
ME: Minimum Energy
( G
ive
n tim
e d
iffe
ren
ce )
( Unknown to be solved for )
( Single revolution, N = 0, 0 < < 360 )o
ME
Pat
h l
ine
t
m
]
Multi- Revolution Lambert Solutions
Sun, F.T., AAS Paper 79-164,
“On the minimum time trajectory and
multiple solutions of Lambert problem”
Transfer orbit 1: inc_to = 1, i < 90o
Transfer orbit 2: inc_to = 1, i > 90o
Multi- revolutions (N)
may have ( 2N + 1 ) solutions
for each inc_to and a given
0 < < , 0 < < 1
Nor
mal
ized
Tim
e
Elliptic Orbits
Path parameter,1 < x < 1
x
High Path Low Path
Parabolic Orbits
Path parameter, x = 1
Hyperbolic Orbits
Path parameter, x > 1
3
3
( Multi revolution, N > 0, > 360 )
o
1/
2
N = 1
N = 0
N = 2
ME: Minimum Energy
ME
Pat
h l
ine
t
m
]
( G
ive
n tim
e d
iffe
ren
ce )
( x = Unknown to be solved for )
N = Revolution number
v (t )
object at t2
2
Transfer
Orbit 2
v (t )1
r (t )1
r (t )2
Transfer
Orbit 1
v (t )1
v (t )2
Sphere
object at t1
2-Body vs. Lambert+ Solutions
Given: r (t ) , r (t ) , t , t , inc_to
Find: v (t ) , v (t )
2
1
1 21
2
Lambert Problem
Spherical
Earth
2-Body/Keplerian
Trajectory
Good: 2-Body Lambert solution
Spheroidal
Earth
Better: Targeting by Kepler+
Lambert+ solution
Kepler+ Trajectory
(J , J , J + N Body ,
+ J , J , Drag , . .) 2 43
31 32
_
r (t )2
v (t )
r (t )1
v (t )1
1
objectat t
object
at t2
2
Orbit
Sphere /
Spheroid
2-BodyKepler+
TransferOrbit
2-Body Lambert+
Lambert+Transfer
2-Body
Lambert Algorithm Developers
2. Sun/Der Lambert algorithm applies to multi-rev and rarely has singluarity
(simple theory + straightforward implementation = speed, accuracy, robustness)+
Theories/
Formulations
Implementations/
Iterative Methods
Gauss
Battin
Shepperd
Gooding
Klumpp
Two Equations
and
Two Unknowns
One Equation
and
One Unknown
Newton,
Halley,
and Others
Laguerre
and Modified
Laguerre
Lambert
Gauss
Battin
Lancaster/Blanchard
Godal
Vinh
Sun
Everyone
(almost)
Conway
Der
1. Most (over 90%) Lambert algorithms apply to zero rev and have singularities
Lambert Algorithm Characteristics:
Orbit Determination and Prediction/Propagation
SP is accurate, but
slow for SSA
Analytic algorithms in OD and P/P
need to process 100,000+ Objects
in less than 12 hours with accuracy
of 10 km to centimeters for SSA
SGP4 needs TLE
conversion
(not efficient for SSA)
Kepler+ is accurate
and fast for SSA
1
3
2
Radars /Opticalsensors
RawObservationdata
Orbitalelement set
r and v
(Estimated initial)positin andvelocity vectors
Prediction /
Propagation
SGP4 Kepler+Numerical
Integration
Osc2Mean
Ephemerides
Close approach(miss distance)
Rise/SetSite visibilityOthers . . . . . .
Un
suit
ab
le f
or r
eal-
tim
eau
tom
atic
pro
cess
ing
1 32
(Estimated future/past)position and velocity vectors
o
Osculating OrbitalElements at t
o o[ r (t ), v (t )]
r
v
[ r , v ]
Initial OrbitDetermination
DifferentialCorrection
ProcessedObservationdata
Batch / KF
Astrodynamics102: Lambert /103: Gauss/Laplace
SPAstro 101:Kepler
Object at t
Futurelook angles( pointing prediction )
DeterminationOrbit
TLEconversiondifficulties
Singularitydifficulties
5. Applications
SSA and Other
Applications
Applications of Analytic Algorithms(1)
Missile Launch- and impact-Point Predictions
BM – Impact Point
Predictions
• Numerical solutions
possible but too late
for countermeasures
• Analytic Lambert+ (speed & accuracy)
A Few Minutes Too Late for any Intercept
Applications of Analytic Algorithms(2) UNCLASSIFIED
SF
Example: I = J = K = 100
Correlation combination
= I J K = 1,000,000
Solve by New
angles-only algorithm:
90+ % of
objects to
catalog
Takes a few seconds for
a million combinations
Correlating
3 detected
ranges
3 computed
ranges =
UCT processing
I
J K
SF
?
Multi-sensor Multi-object UCT Cataloging using Radar Data
Fence or Radar
Applications of Analytic Algorithms(3) UNCLASSIFIED
Multi-sensor Multi-object UCT Cataloging using Optical Sensor Data
Applications of Analytic Algorithms(4)
Solution: New Analytic Astrodynamics algorithms
Problem: SSA
Next
Astrodynamics 102
Part 2: Verifications (Please download iOrbit:
http://derastrodynamics.com/index.php?main_page=index&cPath=1_7
and run lam for Astrodynamics 102 Verifications)
Numerical Example 1
Rev # Path, x v_eci (t1) (km/s) v_eci (t2) (km/s)
N = 0
Output: lambert2 converged to the correct 2-Body solution
inc_to = 1, (inclination of transfer orbit < 90 deg.), t1 = 0., t2 = 1618.50 (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
ICBM transfer orbit , Single revolution 2-Body solution
computed inclination of transfer orbit = 67.895 deg., transfer angle = 296.368 deg.
*
*
Numerical Example 2
Rev # Path, x v_eci (t1) (km/s) v_eci (t2) (km/s)
N = 0
Output: lambert2 converged to the correct 2-Body solution
inc_to = 1, (inclination of transfer orbit > 90 deg.), t1 = 0., t2 = 1618.50 (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
ICBM transfer orbit , Single revolution 2-Body solution
computed inclination of transfer orbit = 112.105 deg., transfer angle = 63.632 deg.
*
*
Numerical Example 3
Rev # Path, x v_eci (t1) (km/s) v_eci (t2) (km/s)
N = 0
N = 1
N = 1
Output: lambert2 converged to correct 2-Body solutions
inc_to = 1, (inclination of transfer orbit < 90 deg.), t1 = 0., t2 = 36000. (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
High Earth Orbit (Molniya transfer orbit) , Multi- revolution 2-Body solutions
*
*
* inc_to as input dictates the inclination of the transfer orbit.
Also allows multi-rev solutions better grouping, as all solutions (N = 0, 1) have the same inclination
and transfer angle.
computed inclination of the three transfer orbits = 63.388 deg., transfer angle = 44.705 deg.
Numerical Example 4
Rev # Path, x v_eci (t1) (km/s) v_eci (t2) (km/s)
N = 0
N = 1
N = 1
Output: lambert2 converged to correct 2-Body solutions
inc_to = 1, (inclination of transfer orbit > 90 deg.), t1 = 0., t2 = 36000. (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
High Earth Orbit (Molniya transfer orbit) , Multi- revolution 2-Body solutions
*
* computed inclination of the three transfer orbits = 116.612 deg., transfer angle = 315.295 deg.
* inc_to as input dictates the inclination of the transfer orbit.
Also allows multi-rev solutions better grouping, as all solutions (N = 0, 1) have the same inclination
and transfer angle.
Numerical Example 5
Algs. Rev# v_eci (t1) (km/s) v_eci (t2) (km/s)
2-Body N =0
Vinti
_targeting
N = 0
Lambert+ N = 0
Output:
inc_to = 1, (inclination of transfer orbit < 90 deg.), t1 = 0., t2 = 1618.50 (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
ICBM transfer orbit , Single revolution 2-Body, Vinti targeting, Kepler+ targeting solutions
*
* computed inclination of transfer orbit = 67.895 deg., transfer angle = 296.368 deg.
clock1 = at t1 (needed for Lambert+)
Numerical Example 6
Algs. Rev# v_eci (t1) (km/s) v_eci (t2) (km/s)
2-Body N = 0
Vinti
_targeting
N = 0
Lambert+ N = 0
Output:
inc_to = 1, (inclination of transfer orbit > 90 deg.), t1 = 0., t2 = 1618.50 (seconds)
r_eci (t1) (km) r_eci (t2) (km)
computed inclination of transfer orbit = 112.105 deg., transfer angle = 63.632 deg.
*
*
Input:
clock1 = at t1 (needed for Lambert+)
ICBM transfer orbit , Single revolution 2-Body, Vinti targeting, Kepler+ targeting solutions
Numerical Example 7
Algs. Rev# v_eci (t1) (km/s) v_eci (t2) (km/s)
2-Body, N = 0
Vinti_targ N = 0
Lambert+ N = 0
2-Body, N = 1
Vinti_targ N = 1
Lambert+ N = 1
2-Body, N = 1
Vinti_targ N = 1
Lambert+ N = 1
Output:
inc_to = 1, (inclination of transfer orbit < 90 deg.), t1 = 0., t2 = 36000. (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
computed inclination of the three transfer orbits = 63.388 deg., transfer angle = 44.705 deg.
clock1 = at t1 (needed for Lambert+)
Molniya transfer orbit , Multi- revolution 2-Body, Vinti targeting, Kepler+ targeting solutions
Numerical Example 8
Algs. Rev# v_eci (t1) (km/s) v_eci (t2) (km/s)
2-Body, N = 0
Vinti_targ N = 0
Lambert+ N = 0
2-Body, N = 1
Vinti_targ N = 1
Lambert+ N = 1
2-Body, N = 1
Vinti_targ N = 1
Lambert+ N = 1
Output:
inc_to = -1, (inclination of transfer orbit > 90 deg.), t1 = 0., t2 = 36000. (seconds)
Input:
r_eci (t1) (km) r_eci (t2) (km)
clock1 = at t1 (needed for Lambert+)
Molniya transfer orbit , Multi- revolution 2-Body, Vinti targeting, Kepler+ targeting solutions
computed inclination of the three transfer orbits = 116.612 deg., transfer angle = 315.295 deg.
Lambert+ Verifications
Astrodynamics 102
Part 2: Verifications (Please download iOrbit:
http://derastrodynamics.com/index.php?main_page=index&cPath=1_7
and run lam for Astrodynamics 102
for more Lambert+ Verifications)