Transcript Thesis Defense Presentation

Progress on Target Tracking &
Engagement Demonstration
Presented by Lane Carlson1
M. Tillack1, T. Lorentz1, J. Spalding1, D. Turnbull1
N. Alexander2, G. Flint2, D. Goodin2, R. Petzoldt2
(1UCSD, 2General Atomics)
HAPL Project Review
PPPL, Princeton, NJ
December 12-13, 2006
Omnipresent Target Engagement Requirement
Final Key Requirement:
• 20 µm engagement accuracy in (x,y,z) at ~20 m (10-6)
• All individual tracking and engagement
components have been operated successfully.
• All components necessary for a glint-only
hit-on-the-fly demo have been integrated and
are operational.
Current tabletop target engagement demo is
complete with one simulated driver beam
crossing
sensors
Poisson
spot
camera
Glint laser
now in
operation
C1
C2
C3
drop
tower
verification
camera
alignment &
driver beam
(635 nm)
coincidence
sensor
microlens
array
collimating
lens
focusing mirror
pulsed glint
laser (1064 nm)
fast
steering
mirror
chamber
center
fringe
counter
Poisson (632 nm) &
fringe counting
(1540 nm) beams
wedged
dichroic mirror retroreflector
Poisson & fringe
counting systems not
used in glint-only demo
Hit-on-the-fly experiment has demonstrated
engagement on moving target
1) Update time of Poisson spot centroiding algorithm down from
10 ms to 3.5 ms in software.
2) Fringes off falling target counted over 3 mm.
3) Using crossing sensors, the timing & triggering system triggers all
necessary lasers/components on-the-fly.
4) Engaged moving targets with a simulated driver beam by using
the glint return signal to steer a fast steering mirror.
5) Verification system has been used to measure error in target
engagement.
• 150 µm range
• 6 µm resolution
A range of tracking/engagement scenarios call
for different requirements
Example #2: in-flight, pre-steering
corrections by Poisson, fringe counter
Example #1: no in-chamber gas,
glint provides final mirror steering
Example #1
Ultra-FSM
Example #2
FSM
Achieved (1)
± 1 mm
± 5 mm
4 mm
Poisson spot centroiding
(x,y position)
N/A
± 100 µm, 5 ms
update time
5 µm,
3.5 ms
Fringe counting (z position)
N/A
± 100 µm
5 µm over
5 mm
Crossing sensors (zero
crossing, velocity prediction)
± 0.7 µs,
± 70 µm
± 10 µs,
± 1 mm
45 µm
Glint/coincidence sensor
± 10 µm
± 10 µm
± 10 µm
± 5 µm, ~2 ms
± 5 µm, ~2 ms
± 6 µm, 2 µs
± 20 µm
± 20 µm
150 µm
~20% of time
Sub system
Target Injector (prob. TBD)
FSM pointing precision
Target engagement
#1 Transverse Target Tracking
Using Poisson Spot Centroid
Poisson
spot
camera
Poisson (632 nm) &
fringe counting
(1540 nm) beams
Reduced region of interest (ROI) technique further
improves update time
• Update time reduced from 10 ms to 3.5 ms by implementing a
“dynamic ROI” in software.
• Number of pixels to process is reduced from 307k to 10k.
6 mm
• The smaller ROI assumes the
target will not move more than
a few pixels between frames.
• ROI is recalculated each frame
to follow the Poisson spot
centroid.
30x less
pixels
4 mm
CMOS
sensor
reduced
Poisson spot
4mm sphere
region on
of
interest
translation stage
New centroiding algorithm implements dynamic ROI
1) Capture image
2) Brightness &
contrast adj.
3) Threshold pixels
above a certain
value
7) Pixels outside ROI
excluded
5) Particles filter,
centroid computed
6) Dynamic ROI drawn
4) Remove border
objects
8) Subsequent
frames contain
reduced number
of pixels
 Target position update time:
3.5 ms (5 µm 1)
 Closing in on goal of 1-2 ms
#2 Axial Target Tracking Using
Interferometic Fringe Counting
reference
leg
fringe
counter
Poisson (632 nm) &
fringe counting
(1540 nm) beams
Fringe counting has been demonstrated over 3 mm
• Lower-noise photo-detector
Similar intensities
• Higher-power laser (60 mW)
stationary
reference
leg
=> Fringes off falling target counted over 3 mm
Target releasing from
vacuum needle
Doppler
reflection
IR telecom
laser
photo diode
fringe counter
Free-falling target
Mini drop tower setup
5 ms/div
5 V/div
time
Signal processing
required to obtain
countable signal
20 µs/div
#3 Crossing Sensors & Axial
Position Prediction
crossing
sensors
C1
C2
C3
drop
tower
Crossing sensors & real-time operating system
compute predicted target location on-the-fly
Last time: Established crossing sensors to be sufficiently
precise (45 µm 1) to trigger glint laser.
C1
 Timing and triggering system
fully operational for our demo
C2
Overview of Timing Sequence:
• Timing sequence initiated by target crossing C1.
Glint
• C2 crossing yields target velocity.
• Velocity info used to trigger alignment beam, glint laser,
verification camera, and driver beam.
• Disparity between predicted and actual target location
is detected by PSD and corrected by steering mirror…
Driver
#4 Glint System
alignment &
driver beam
(635 nm)
coincidence
sensor
focusing mirror
pulsed glint
laser (1064 nm)
microlens
array
collimating
lens
fast
steering
mirror
wedged
dichroic mirror retroreflector
All necessary components have been integrated
for glint-only target engagement demo
Glint laser - final
component of hiton-the-fly demo
has been installed
Pulsed diode laser
(simulated driver)
alignment &
driver beam
(635 nm)
New Wave 35 mJ, 1064 nm
glint laser
coincidence
sensor
focusing mirror
pulsed glint
laser (1064 nm)
microlens
array
collimating
lens
fast
steering
mirror
wedged
dichroic mirror retroreflector
Simulated wedged
dichroic mirror
Optics In Motion
fast steering mirror
Wedged dichroic mirror compensates for
glint/chamber center offset
1 cm
target at
glint
location
Co-axial glint return
and driver
target at
“chamber
center”
Verification
camera
Simulated wedged
dichroic mirror
We have engaged moving targets with a
simulated driver beam
• Last time we used a simulated glint return from a
stationary target to steer a mirror.
• Now, we have used the “real” glint return signal
from a moving target (5 m/s) to steer a simulated
driver beam to engage the target.
Targets fully
engaged ~20% of
the time (in 150 µm
verification range)
150 µm diam. verification beamlets
20%
40%
outside range
40%
outside range
150 µm
verification
range
Snapshots of engaged targets
But does not meet the 20 µm
spec yet…
Effort to improve engagement accuracy to 20 µm must
address & minimize all uncertainties
• Errors & uncertainties from every
subsystem contribute to engagement
accuracy.
PSD signal with ground-looping
• We are working to understand errors
and to address each one.
– Air fluctuations, sensor noise,
bandwidth limitations, response times…
Error contributions to engagement
accuracy:
- Reading glint return
(PSD, LabView)
~100’s µm
- Air fluctuations
~20 µm
- Verification camera ~12 µm
- Mirror pointing
~6 µm
Erratic 50 mV signal translates
to significant mirror steering!
(100s microns)
 Resolved by plugging all
electrical components into
same circuit
• Most dominant uncertainty so far is
deciphering the glint return on PSD…
Glint return is used to make one final steering
mirror correction depending on PSD reading
• Glint return on PSD gives target’s final position.
• LabView reads PSD signal, then calculates steering gain to give FSM.
• LabView loop time is 50 ± 20 µs due to non-deterministic operating system.
• Inconsistent voltage
readings grossly and
falsely steer the mirror.
Steering signal
to FSM X-axis
100 µs/div
Glint return on PSD X
2V/div
Glint return on PSD Y
Glint laser q-switch
Peak-hold circuit picks off PSD voltage more
consistently than software
• Peak-hold circuit holds the peak voltage until LabView can read it.
• Also researching other ways to read glint (photo-diode, quad-cell)
Rise time may also influence
reading consistency
=> More work must be
done on glint return
characterization, PSD
response, error/noise
reduction.
Glint returns on PSD
Signal held by peak det.
(not mirror command)
50 µs/div
2V/div
#5 Engagement Verification for
Hit-on-the-fly Demo
alignment &
driver beam
(635 nm)
microlens
array
collimating
lens
verification
camera
chamber
center
Applet post-processes snapshot to verify
target engagement accuracy
target
beamlets
PSD
• Applet computes light centroid of
obscured and un-obscured beamlets.
imaginary target shadow
No target
• Resolution of 6 µm, 150 µm range
Target equally eclipsing beamlets
Target offset to the right ~ 100 µm
Summary of progress & plans
#1 Moving target engaged by simulated driver beam
#2 Transverse Tracking System Using Poisson Spot:
Progress: Improved update time to 3.5 ms using dynamic ROI
Plans: Implement into system to help pre-steer mirror
#3 Axial Tracking & Triggering Prediction:
Progress: Faithfully triggering glint, simulated driver beam & verification camera.
Also, fringes off moving target counted over 3 mm
#4 Glint System:
Progress: Glinted target & steered FSM to engage in-flight target
Plans: Better characterize glint return & PSD response to meet 20 µm goal
#5 Target Engagement Verification:
Progress: Engagement verification resolution of 6 µm, 150 µm range
Future Goals to Consider:
- increase speed capability to couple with a 50 m/s injector
- add more driver beamlines at different angles.