Multi-Agent Airborne Laboratory for Cryospheric Remote Sensing Final Report Shawn Keshmiri, Mark Ewing, Richard Hale, Carlton Leuschen, John Paden, Fernando Rodriguez-Morales and Jie Yan University of Kansas 2335 Irving Hill Road Lawrence, KS 66045-7612 http://cresis.ku.edu Technical Report CReSIS TR 164 May 30, 2016 This work was supported by grants from the Paul G. Allen Foundation and the National Science Foundation (#ANT-0424589)
15
Embed
Multi-Agent Airborne Laboratory for Cryospheric Remote ... · aircraft weight was reduced by approximately 5 lb. The strength of the new wing was validated through both analysis and
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Multi-Agent Airborne Laboratory for Cryospheric Remote Sensing
Final Report
Shawn Keshmiri, Mark Ewing, Richard Hale, Carlton Leuschen,
John Paden, Fernando Rodriguez-Morales and Jie Yan
University of Kansas
2335 Irving Hill Road
Lawrence, KS 66045-7612
http://cresis.ku.edu
Technical Report
CReSIS TR 164
May 30, 2016
This work was supported by grants from
the Paul G. Allen Foundation and the National Science Foundation
for a swarm of UASs which uses topological distances instead of metric values. Biologically
inspired, this method evades exploiting distance measurements in the absence of such
capabilities, as it happens in swarming birds [23-25];(2) Develop built-in collision avoidance
mechanisms to keep the formation as uniform as possible and prevent UASs from crossing
over or tangling [23-33]; (3)Validate flexibility and coherency of agents in following an
arbitrary sinewave-like, or curvature shaped flight formation; (4) Identify the aerodynamics
of UASs in the presence of environmental disturbances (e.g. cross wind or windsheer) and
validate the robustness of agent controllers [23-30]. A dozen papers were published on
guidance, navigation, and control of multi-agent UASs and unsteady and nonlinear
aerodynamics of UASs in prestigious national and international journals and conferences [20-
37].
For the first time in the history of UAS control, the KU team was able to develop a new
autopilot system that is capable of running computationally intense control (e.g. Nonlinear
Model Predictive Control) and artificial neural network-based system identification algorithms
in real time [37]. The new autopilot allowed the KU team to successfully validate the
developed guidance and control swarm algorithms between actual and virtual UASs in real-
time [Ref: Year 1 Progress Report].
F. UAS Flight Tests
The complexity of flight test on polar missions is an order of magnitude higher than flight
tests in the continental U.S. Successful UAS missions in polar regions are highly dependent
on the UAS reliability and demands for extensive crew training. The University of Kansas has
obtained nine certifications of authorization (COAs) from the Federal Aviation Administration
(FAA) to test UASs locally in Lawrence, Kansas. Local flight tests made frequent UAS
reliability tests and crew training more achievable and affordable. More than 200 UAS
validation and verification flight tests were successfully conducted in Lawrence Kansas prior
to Greenland deployment. Multiple UAS platforms were used in these flight tests including
the 40% scale Yak-54, Bird of Time, DG808, and G1X UASs.
With the support from the National Science Foundation, the successful UAS flight test
campaign in Lawrence, Kansas was followed by a polar mission in March- April 2016. The
Russell Glacier, a fast flowing glacier located northeast of Kangerlussuaq, Greenland, was
targeted for this science mission. Three different UAS platforms were flight tested 32
autonomous flights. To complete the science mission, the G1XB UAS was instrumented with
35 MHz radar. The UAS flight tests were conducted from two frozen lakes in proximity to the
Russell Glacier. Although the extreme weather (very cold and very windy), specific
geographic location of lakes (surrounded by high mountains), and manned aircraft flight
operations in the area increased the complexity of this polar mission by an order of
magnitude, the CReSIS UAS was able to successfully perform eight (8) over the horizon
missions covering about 400 km of ice sheets. The G1XB UAS was able to follow the glacier’s
steep slope and collect data over closely spaced lines (3.65~24 m) (see Figure 4). It is
extremely difficult for manned aircraft to hold altitude/heading @ 400-600 ft above the
ground level with relatively slow speed (30 m/sec) and to collect data over such closely
spaced grid lines, yet we were able to achieve this success using UAS for the first time during
this mission [38].
Figure 4:3D Trajectory of G1XB UAS over the Russell Glacier, Kangerlussuaq, Greenland. The top-right inset shows the
crevassed surface along the flight path, which makes radar sounding a challenge.
G. Results
We collected science data over three previously flown flight lines. During these previous
missions, we collected VHF (180-210 MHz) radar sounder data on these lines. The three
lines were chosen to represent a range of image quality from poor to good. In all cases, the
new HF sounder images are a substantial improvement over the VHF sounder images. The
HF sounder is also compared with data collected by the JPL WISE HF sounder (Rignot 2013
[39-40]). We also began work on 3D imaging and show some preliminary work. Additional
time is needed to tune the processing parameters, estimate and remove noise sources,
validate system parameters and performance, and finally to test different types of 3D imaging routines.
Figure 5 shows a map of the 3 images used for comparison. The good quality line is green;
the medium quality line is yellow; and the poor quality line is red. Figure 6 compares each
of the lines between new HF sounder data and VHF sounder data. Where the VHF sounder
is able to see the ice bottom, the HF sounder matches it. In many places the VHF sounder
cannot see through the temperate ice layer which sometimes starts only a few tens of
meters below the ice surface – completely obscuring and attenuating the ice bottom in
many regions. The HF sounder is able to sounder nearly 100% of the ice bottom including
many regions where no ice bottom is detectable in the VHF imagery. The HF sounder, with
its long wavelength, is much less sensitive to the temperate ice water inclusions and
crevassing. Note that the nearly horizontal line at ~300 m depth in each VHF image is the
ice surface multiple: this is a multiple reflection or resonance between the large P-3 aircraft
and the ice surface. Figure 7 shows the WISE data from a line which crosses all three
CReSIS HF sounder lines. The intersections with the good, medium, and bad lines are
shown by the vertical bars. The corresponding intersections are also shown in Figure 6 with
a blue line. The ice bottom is difficult to impossible to detect in the WISE data and demonstrates the improved performance of the CReSIS HF sounder over the WISE system.
The first step in 3D imaging is to co-register the different passes. We have completed this
first step for the six passes over the good quality line. Figure 8a shows the phase of the
interferogram formed from two overlapping passes. The bright ice surface and ice bottom
are phase-coherent and standout from the noisy background. Because the scattering from
the ice surface and bottom mostly comes from the nadir direction, the phase is consistent
and near zero angle (cyan colored). The ice surface is flatter than the bed and we use it to
verify that the coregistration angle is small in Figure 8b. Some variation about zero angle is
expected since the ice surface does have some roughness due to crevasses and the overlap between the two flights is not perfectly zero.
Figure 5: Map showing the HF sounder data lines used for comparison over Landsat-7 imagery. The good quality line is green, the
medium quality line is yellow, and the poor quality line is red.
Ice Bottom
TemperateIce Layer
Ice Bottom
MissingIce Bottom
Ice Bottom
Ice Bottom
Ice Bottom
Ice Bottom
SurfaceMultiple
SurfaceMultiple
Figure 6: a) through f) HF and VHF sounder comparisons for good, medium, and poor quality lines. The point of intersection with the
WISE line is shown by the blue line.
a) b)
Figure 7: a) Comparison with WISE HF sounder data. The intersection with the good (green), medium (yellow), and poor (red) quality
lines are shown. The location of the ice surface and ice bottom detected by the CReSIS HF sounder is shown with an ‘x’. b) Magnified
region around the ice bottom shows that the WISE HF sounder data quality is significantly lower than the CReSIS HF sounder.
Ice Surface
Ice Bottom
a) b)
Figure 8: Phase coherence from co-registration of multiple passes. a) Phase-angle map of the interferogram formed from two overlapping
passes taken on two different flights. The ice surface and ice bottom show good coherence whereas the background noise has poor
coherence and the angle varies rapidly from pixel to pixel. b) The angle extracted from the strong ice surface in a).
3) What outcomes were not achieved?
As discussed in section 2.D, well-grounded research and experiments were conducted on
the swarm of multi-agent UASs. Additionally, all required hardware and software for such
missions were designed, developed and verified. However, flight test of multi-agent swarm
of UASs is pending due to Federal Aviation Administration (FAA) UAS regulations. The FAA
regulations prohibit any multi-agent flight test in U.S. airspace. KU has been in negotiation
with the FAA since 2014 and we anticipated receiving special authorization for operation of
multi-agent UASs in near future.
4) What barriers, if any, prevented you from achieving all of the expected outcomes? How
did you overcome the barriers during the grant period?
Lack of transparent UAS flight test regulations from 2014-15 was a major hurdle for our
flight test activities. Although KU filed for the UAS certification of operation in 2014, we had
to wait till May 2015 to receive approval from the FAA for UAS flight test in South West of
Lawrence Kansas. During this period, we closely collaborated with the U.S. Army to flight
test our UASs in their restricted area in Fort Riley, Kansas. More than 60 UAS flight tests
were conducted at Fort Riley, Kansas which is home of the 1st Infantry Division. KU UAS flight
test operation in Fort Riley was continued till February 2015 when the FAA forced the U.S. Army to shut down our UAS operation. UAS flight test operation was put till June 2015.
The European UAS flight test regulations were no less complex compared to the U.S. FAA
regulations. It took KU 18 months to negotiate UAS flight test operations in the Greenland
airspace. The complexity of this process was compounded by the need to convince both the
Danish Transpiration Authority and Greenland Air Traffic Control that our UAS operation is
safe. Our approval process was further exacerbated because of the proximity of Russell
Glacier to the Kangerlussuaq International Airport which is the main hub of manned aircraft
operations in Greenland with many daily flights. Fortunately, KU’s prior history in UAS flight
operations in the U.S. and in polar regions helped us to receive permission to flight test all
our UASs in western Greenland; however, the flight test multi-agent UASs was not
approved until similar flights are conducted in U.S.A. and proven to be safe. In the absence
of swarming UASs, the concept was proven using one UAS flight multi-pass on a closely
spaced grid (see Figure 4).
5) What strengths and unexpected opportunities helped you achieve your outcomes?
In addition to the kind support from the PGAFF, it is extremely important to mention the
importance of matching funds from the National Science Foundation and the University of
Kansas. These resources were vital to achieving this initial validation of the proposed concept
in polar regions. We would not achieve any of scientific outcomes shown in Section 2. F, if it
was not for logistical support of National Science Foundation. By partnering with the PGAFF,
we were able to incorporate important improvements to the design and fabricate state-of-
the art radars and UASs for our studies. By leveraging federal and local support from the NSF, Fort Riley, KU, and the PGAFF we had the resources to deploy our team to Greenland.
Very recently, KU was granted a blanket certification of operation (COA) by the FAA for
UAS flight test operation in May 2016. This permission allows KU to perform UAS flight
tests anywhere in the continental U.S. as long as the UAS weighs less than 55 lbs and
flight test is conducted under 400 ft. AGL in a class-G airspace.
6) What other benefits, if any, has the grant project provided your organization or the
larger community?
The support of PGAFF has made a significant contribution to operational research on the
cryosphere with a domino effect on a multitude of Earth and Science disciplines. While our
proposed concept was transformative, it had yet to be proven viable. Thus, this seminal
investment from the PGAFF award made the creation of this high-risk, high-reward project
possible, and it was necessary to bridge the gaps in what can be supported at an early stage
by federal dollars. Without the support of PGAFF, we would not have had the resources to
launch a successful demonstration of this unproven, yet innovative technology and its
application as a pilot program. When innovative technology ideas are still too early stage for
federal support, it is critical to have the partnership of foundations to bridge this gap and
enable the proof of concept demonstration. Furthermore, as federal investments remain
scarce, the ability to leverage partnerships like ours with PGAFF can help us to unlock federal
support down the line by demonstrating the value and potential of our research to better
understand climate change in these difficult to characterize, yet very instrumental, regions
of glaciers and ice sheets.
Emerging ice sheet models for glacial flow require fine resolution data near the terminus of
key glaciers. With the support of the PGAFF, we demonstrated that achieving the resolution
required to reduce uncertainties in projecting rates of sea level change can only be
accomplished with a UAS flight based radar system.
7) For questions A & B below: If your project outcomes included serving
individuals, please indicate the number of people and characteristic of those
served in the questions below.
A.1. Indicate the number of individuals served below.
N/A
A.2. Indicate the primary characteristic of the individuals served (e.g., general program
participants, audience, youth).
Please select N/A if not applicable to your project.
N/A
B.1. If your grant outcomes include serving a second cohort of people, list below how many
were served.
N/A
B.2. Indicate the primary characteristic of the individuals served (e.g., general program
participants, artists, youth).
Please select N/A if not applicable to your project.
N/A
Attachments
Final Budget Report [Required]: Attach a final project budget below (reflecting a budget-to-
actual financial report). Include a budget narrative that explains how grant funds were
spent and notes important deviations from the original grant budget.
Please see attached financial report and justification.
Products Associated With the Grant Project: To help us better understand work completed
during the grant period, please submit the following attachments. (If you have multiple
documents for each of the categories below, upload each document along with the
respective Title in the dropdown list below).
A. Any key press or promotional materials associated with the grant project.
A bibliography of related publications is attached.
B. Significant work products (including evaluation reports, strategic plans, fundraising or
communication plans). Also include programs or major materials developed through the
grant project. Note: If any work product, such as videos or programs, cannot be attached
you may send it in by mail to our office.
Flight Test videos were shared with our project manager, Spencer Reeder, and
thousands of photos were taken during domestic and Greenland flight test
activities that can be shared with the PGAFF upon request.
C. The Foundation regularly updates its Web site with photos and descriptions of grant
projects. If you have photos from the project that you would like us to consider posting to
the Web site, please submit them (up to six photos total). By submitting photographs you
grant the Foundation a non-exclusive, transferable, royalty-free license to use and display
these photographs for non-commercial purposes. If a photo credit is required, please
include appropriate credit information.
We have previously submitted a blogpost for the PGAFF website, and we are
delighted to provide an updated blogpost upon request.
Bibliography
1. R. K. Pachauri and A. Reisinger (Eds.), Climate Change 2007: Synthesis Report,
Contribution of Working Groups I, II and III to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, IPCC, Geneva, Switzerland.
2. I. Joughin et al., Continued evolution of Jakobshavn Isbrae following its rapid speedup,J.
of Geophysical Research, vol. 113, F04006, doi:10.1029/2008JF001023, 2008.
3. R. Thomas, E. Frederick, J. Li, W. Krabill, S. Manizade, J. Paden, J. Sonntag, R. Swift,
and J. Yungel, Accelerating Ice Loss from the fastest Greenland and Antarctic Glaciers,