Space Systems Laboratory Massachusetts Institute of Technology
Optimization of Separated Spacecraft
Interferometer Trajectories in the
Absence of A Gravity-WellAbsence of A Gravity-Well
Edmund M. Kong
Prof David W. Miller
MIT Space Systems Laboratory
20
th
March 1998
Space Systems Laboratory Massachusetts Institute of Technology
Objective & Approach
Objective : Determine the optimal synthetic imaging trajectory for a
Separated Spacecraft Interferometer
Image Quality Optimization
Comparison with Other Alternatives
Other Considerations
Trajectory Optimization
Mass Metric Time Metric
Uniformly Spaced
U-V Points
DS3
Performance Metric
Optimal System
Mass
Trade-offs
Space Systems Laboratory Massachusetts Institute of Technology
Image Quality
Model : 2 Collector and 1 Combiner Interferometer (DS 3)
Physics : Average Image Intensity
Combiner
CollectorCollector
∑
∑
=
=
?
?
?
?
?
?
+
?
?
?
?
?
?
≈
?
?
?
?
?
?
+
?
?
?
?
?
?
?
?
?
?
?
?
?
?
+
=
N
k
kj
k
i
N
k
kj
k
iD
D
ji
y
x
D
N
y
x
JD
N
I
1
22
1
2
sin
sin
1
2
2
)(cos2
21
)(cos2
)(
)cos1(1
),(
??
λ
π
λ
π
??
λ
π
λ
θπ
??
λ
θπ
λ
θπ
Space Systems Laboratory Massachusetts Institute of Technology
Image Quality - Mean Square Error
Minimize :
()
mm
II
MSE
m
i
m
j
o
jiji
×
?
=
∑∑
==11
),(),(
2
????
Approach : Choose N subset of all points
: Compare with Nominal PSF
: Simulated Annealing
Optimization Technique
Results : Image quality increases with
no. of imaging points (N)
: Diminishing rate of return
0 100 200 300 400
0
0.5
1
1.5
No. of Imaging Points
M
SE, (W
m
-2
)
2
(x
1
0
-3
)
Best MSE Performance
Space Systems Laboratory Massachusetts Institute of Technology
Point Spread Function Images
41 Imaging Points
201 Imaging Points
281 Imaging Points
121 Imaging Points
Space Systems Laboratory Massachusetts Institute of Technology
Trajectory Optimization - Mass Metric
Minimize :
?
?
?
?
?
?
∑
?±=
=
N
i
s
i
imageimage
fuel
fuel
a
N
TT
m
m
1
4
2
2
&
Assumptions : “Stop and Stare”
imaging mode
: Trapezoidal velocity
profile
: Constant acceleration
Parameters : Spacecraft masses
Collector = 150 kg
Combiner = 250 kg
: Cold Gas Propulsion
I
sp
= 62.5 s, F = 9 mN
Constraint : T
image
= 264 Hours
Approach : Traveling Salesman
Algorithm
0 100 200 300 400
0
0.2
0.4
0.6
0.8
1
No. of Imaging Points
Fu
e
l
(k
g
)
Fuel Expended Per Image
Result : Fuel mass increases with
no. of imaging points (N)
Space Systems Laboratory Massachusetts Institute of Technology
Trajectory Optimization - Time Metric
Minimize :
∑
=
=
N
n
i
s
a
T
1
2
0 100 200 300 400
0
50
100
150
200
250
300
No. of Imaging Points
T
i
m
e
(H
rs
)
Least Maneuvering Time Per Image
Assumptions : “Stop and Stare”
imaging mode
: Triangular velocity
profile
: Small Integration Time
Parameters : Spacecraft masses
Collector = 150 kg
Combiner = 250 kg
: Pulse Plasma Thrusters
I
sp
= 1000 s, F = 1.4 mN
Constraint : S/C Power 80 W
Approach : Traveling Salesman
Algorithm
Result : Imaging time increases
with no. of imaging points
Space Systems Laboratory Massachusetts Institute of Technology
Other Alternatives
-500 0 500
-500
0
500
X (m)
Y (
m
)
67 Uniformly Spaced Imaging Points
-500 0 500
-500
0
500
X (m)
Y (
m
)
Proposed DS3 Trajectory
Proposed DS3
Reference : Linfield, JPL
Uniformly Spaced
Space Systems Laboratory Massachusetts Institute of Technology
PSF Comparison
281 Optimal MSE
Imaging Points
Proposed DS3
(261 Points)
231 Uniformly Spaced
Imaging Points
Space Systems Laboratory Massachusetts Institute of Technology
Fuel and Time Metrics vs MSE
10
-5
10
-4
10
-3
10
-2
0
0.2
0.4
0.6
0.8
1
1.2
MSE (W m
-2
)
2
Fu
e
l
(
k
g
)
Fuel Expended vs Image Quality
Proposed DS3
Uniform Spacing
Optimized MSE
10
-5
10
-4
10
-3
10
-2
0
50
100
150
200
250
300
350
MSE (W m
-2
)
2
T
i
m
e
(H
o
u
rs
)
Maneuvering Time vs Image Quality
Proposed DS3
Uniform Spacing
Optimized MSE
Result : Better MSE with lower fuel consumption or shorter imaging
time
Space Systems Laboratory Massachusetts Institute of Technology
Other Considerations
Psi
x
Ps
i
y
Main Lobe Comparison
DS3 (261 Points)
Uniformly Spaced (231 Points)
Optimized MSE (281 Points)
0 50 100 150 200 250 300 350
0
1
2
Image Quality versus Fuel Trade-off
No. of Imaging Points
M
SE,
(
W
m
-2
)
2
(x
1
0
-3
)
0
0.5
1
Fu
e
l
(
k
g
)
Main Lobe
Comparison
Trade-off
-500 0 500
-500
0
500
X (m)
Y (
m
)
Imaging Time = 74 Hours
“Image on
the Fly”
2
3
4
5
6
7
5
7
9
11
500
1000
1500
No. of S/C
Total System Mass Required (500 Images)
No. of Cornwell Points
M
a
ss (
k
g
)
Extension to
N spacecraft
Reference : Linfield, JPL
Space Systems Laboratory Massachusetts Institute of Technology
Conclusion
? Determined the optimal imaging locations
? Determined the optimal trajectories
– Mass metric
– Time metric
0 100 200 300 400
0
0.2
0.4
0.6
0.8
1
No. of Imaging Points
Fue
l (
k
g)
Fuel Expended Per Image
0 100 200 300 400
0
50
100
150
200
250
300
No. of Imaging Points
T
i m
e
(H
rs
)
Least Maneuvering Time Per Image
? Compared with other alternatives
? Future considerations
– MSE versus Mass trade-off
– MSE versus Time trade-off
– Extension to N spacecraft
– “Image on the Fly” mode
– Other Metrics
10
-5
10
-4
10
-3
10
-2
0
50
100
150
200
250
300
350
MSE (W m
-2
)
2
T
i
m
e
(Ho
u
rs
)
Maneuvering Time vs Image Quality
Proposed DS3
Uniform Spacing
Optimized MSE
2
3
4
5
6
7
5
7
9
11
500
1000
1500
No. of S/C
Total System Mass Required (500 Images)
No. of Cornwell Points
Ma
s
s
(
k
g
)
Space Systems Laboratory Massachusetts Institute of Technology
Simulated Annealing
? Statistical Approach
? Randomly select a configuration and calculate cost, C
r
– If C
r
< C
r-1
accept r
th
configuration
– If C
r
> C
r-1
accept r only if exp(-C
r
/T) > Random(0,1)
– when C
r
is accepted, decrease T (system temperature)
– Continue until system is frozen (no new solution accepted in N trials)
? Does not guarantee global minimum
? Quick and easy implementation
? Reasonable solution achieved in short computation time
? Reference:
– S. Kirkpatrick, C. D., Gelatt, Jr., M. P. Vecchi, “Optimization by Simulated Annealing”,
Science, Volume 220, Number 4598, 13
th
May 1983.