From Fledgling

to Flight

MARK HARDESTY
THE BOEING COMPANY
SANDY KENNEDY, SHEENA DIXON
NOVATEL, INC.
TRAVIS BERKA, JASON GRAHAM, DON CALDWELL
THE BOEING COMPANY

For several years now the

Boeing Unmanned Little Bird
program has been examining
various methods for precisely
navigating a vertical takeoff and
landing unmanned aerial system
to a moving vessel for launch and
recovery operations. An engineering
team describes the development
and testing of a GNSS/inertial
system that uses relative navigation
techniques to do this successfully.

42 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 Boeing’s Little Bird UAV

T he Boeing
Company initiated
the Unmanned Little Bird
(ULB) program in the fall of 2003
to create a developmental platform for
an optionally manned, vertical takeoff and
landing (VTOL) unmanned aerial vehicle (UAV).
Initial flight test activity employed a modified
MD530FF helicopter, with the first flight taking
place on September 8, 2004. Six weeks later the
program achieved a fully autonomous multiple
waypoint demonstration flight from takeoff
through landing.
After several hundred hours of simulated
autonomous flight with a safety pilot on board,
an unmanned flight was performed at the U.S.
Army’s Yuma Proving Ground on June 30, 2006.
The ULB team’s success in creating a powerful
VTOL UAV aircraft for technology innovation
and demonstration assisted in the rapid develop-
ment and understanding of operational concepts
and requirements. The platform’s autonomous
capabilities continue to be expanded through
low-risk testing in support of UAV subsystems
development.

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 43

 LITTLE BIRD UAV

This article describes a recent company-sponsored flight

test effort to integrate and demonstrate a novel and highly
precise VTOL UAV navigation system for use in a maritime
environment. In it we will describe modifications to the test
helicopter, training and qualification of the flight crew and
engineering test crew, operational theory, and an evaluation
of the precision navigation solution.
The result chronicled here portrays a method employ-
ing integrated GNSS and inertial navigation capabilities to
autonomously guide a VTOL UAV — in this case, a Boeing
H-6U helicopter — to a predetermined precision landing
anywhere on a ship deck, regardless of deck dimensions.

A Development and Testing History

Deck lock engaged in grid
Following its initial accomplishments with the MD530FF
aircraft, the ULB program built a second highly upgraded
developmental and demonstration test helicopter (H-6U)
to support continuing VTOL UAV concepts of operation
(CONOPS). This platform more closely resembles the Mis-
sion Enhanced Little Bird operated by the U.S. Army’s 160th
Special Operations Aviation Regiment, based at Fort Camp-
bell, Kentucky.
The H-6U offers a large increase in performance and pay-
load over the original MD530FF technology demonstrator.
The design approach and integrated test capability provided
by the ULB program supports rapid development and cost
avoidance in the growing VTOL UAV market.
Beginning August 18, 2008, Federal Aviation Administra-
tion (FAA) policy regarding civil unmanned aerial system
Deck lock mechanism (UAS) operations with a safety pilot on board forced flight
test procedural changes. An FAA memorandum stated that
— even if a safety pilot were on board clearing airspace, mon-
itoring all systems, and prepared to immediately take control
of the aircraft — if the aircraft trajectory was controlled from
outside the cockpit (i.e., a ground control station with an RF
data link), the aircraft would be regarded as a UAV. In that
event, the FAA insisted on a thorough review of all systems
and procedures (a process that would take months) before
considering to allow the safety pilot UAV to fly in civil air-
space.
Flight test validation and verification of the trajectory
H-6U landing to the trailer helipad control portions of Boeing’s proprietary COMC2 ground
control station software is currently executed in cooperation
with New Mexico State University’s Physical Science Labora-
tory facility near Las Cruces, New Mexico.
When trajectory control of the H-6U by the ground
control station is not required to accomplish the test objec-
tives, flight testing can be conducted in civil airspace. In this
environment, the automated flight control system (AFCS) is
programmed to behave as a “full authority” autopilot.
Navigation routes are pre-programmed and briefed, and
the safety pilot uses a simple button push to allow the H-6U to
progress between programmed waypoints. This button push
Pilot’s view of the trailer helipad emulates the command that would otherwise be provided by
the ground control station operator, and this simple technique

44 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 LITTLE BIRD UAV

Easting Error
allows the ULB team to comply with cur-

Difference
0.2
rent FAA policy.

(m)
0
The ULB program has realized tre- -0.2
mendous value by employing the safety 1.56
56 1.57 1.58 1.59 1.6
Northing Error
pilot approach. Flight control software

Difference
0.2
can be evaluated in flight, updated, and

(m)
0
re-flown in a single day. Lessons learned -0.2
about aircraft behavior can be modified 1.56
56 1.57 1.58 1.59 1.6
Altitude Error
in flight and fine-tuned for optimal sys-

Difference
0.2
tem performance.

(m)
0
The safety pilot can allow the AFCS to -0.2
misbehave long enough to insure that suf- 1.56 1.57 1.58 1.59 1.6
ficient data is collected to define a system GPS Time × 10 5

problem, enabling the engineering staff to FIGURE 1 Plots comparing accuracy of H-6U-mounted GNSS/INS real-time kinematic performance
versus post-processed results from Spaceport America flight test
gain a quicker understanding of malfunc-
tions and thus correct issues faster. Ulti-
mately, the safety pilot is tasked with insuring that the H-6U The helipad was equipped with a NATO-standard har-
does not depart to an attitude or situation where the helicop- poon grid, which allows the deck lock device to engage
ter cannot be recovered without damage or injury. and lock to the deck. The forward deck of the trailer was
The Boeing Unmanned Little Bird H-6U program recent- equipped with the RF navigation system, a tactical common
ly partnered with French companies Thales and DCNS, data link (TCDL) for VTOL UAV command and control,
which builds warships including a type of frigate on which a differential GPS/inertial measurement unit (IMU) truth
the VTOL UAV would be designed to land. data system, and various video cameras. A specially modi-
The objectives of this partnership included development fied command and control vehicle towed the rig at a precisely
and demonstration of a radio frequency–based navigation maintained speed from 5 to 25 miles per hour.
system, a ship “green deck window” safe landing period pre- This test method allowed the accurate and rapid evalua-
dictor, and a deck-lock aircraft capture device, all intended tion of the RF navigation and harpoon deck-lock system to
for VTOL UAV shipboard terminal operations. successfully navigate to a landing and secure the H-6U to the
The terminal area navigation system, known by the heli-deck. The in-motion test activity took place at the vast
French acronym DAA, was designed to minimize ship emis- runway facility at Spaceport America in New Mexico, which
sions and be independent of satellite-based navigation solu- operates within a restricted airspace controlled by White
tions such as GPS or GLONASS. The “green deck window” Sands Missile Range.
predictor and the deck lock system were designed to mini- During flight testing at Spaceport America, the GPS/
mize human error and the risk of airframe or ship damage IMU system operated in a real-time kinematic (RTK) mode
during decking operations in a variety of weather and sea with a data link to a local reference station; the baseline never
state conditions. exceeded 10,000 feet. Figure 1 demonstrates the level of accu-
To support these activities, the Little Bird program need- racy in each dimension, comparing the RTK solution versus
ed to create a tool suitable for evaluating the performance of a post-processed solution. This test vetted the RTK solution
systems such as the Thales DAA radionavigation system. The for use as a “truth” source to evaluate the performance of the
GNSS/INS described here is designed to meet that need. DAA radionavigation landing system.
The ULB team broke the test program into several phases.
Initial trials of the navigation system included the use of a six H-6U Cockpit Instrument Panel Upgrades
degree-of-freedom motion platform to examine the ability of The H-6U cockpit instrument panel was originally equipped
the navigation system to compensate for ship motion. Con- for visual flight rules (VFR) operations in a non–visually
currently, we evaluated the “green deck window” predictor. degraded environment. This cockpit instrumentation was
The accompanying photos show the mechanical deck lock considered adequate for flight visibility conditions that almost
during static lab testing. The tests progressed to manual, then always exist in the desert southwest of the United States, where
automatic engagements while landing on a platform that was most flight test activity has occurred. However, operations
under way. under visibility conditions that can be expected in a maritime
Cost, safety, and logistical constraints demanded a novel environment such as the western Mediterranean demanded a
developmental facility to support the intermediate phase of complete redesign of the cockpit instrument panel.
the test program. A tractor-trailer rig was highly modified to Boeing Flight Operations in Mesa, Arizona flies a Eurocop-
emulate the landing deck of a frigate (see accompanying pho- ter AS-350B3 helicopter for chase and crash rescue duties. This
tos). The trailer deck was extended to 16 feet wide with an aft helicopter is equipped with a glass cockpit display system. The
load-bearing helipad measuring 16 x 16 feet. expense of training pilots on different cockpit designs and the

46 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

complexities of operating various avionics suites made com-
mon cockpit avionics architecture a logical decision.
The H-6U was also in need of a new radar altimeter for
terminal and near-Earth flight operations. We decided to
equip our test helicopter with the same avionics suite as the
chase helicopter.
Output from the radar altimeter could also provide data
to the H-6U AFCS where it could be integrated into the flight
control laws. Boeing and its avionics supplier agreed to work
together to evaluate the performance of the new device, with
H-6U RADAR altimeter antennas
antennas mounted on the tail boom of the H-6U.
Often, radar altimeter antennas are mounted on the belly
of a helicopter, an installation that can render the device use-
less when interference below the helicopter exists. The tail
boom antenna placement allows use of the radar altimeter
data during slung load operations, as well as while landing to
a NATO standard deck-lock grid.

Relative Navigation Methodology

The UAV VTOL application requires continuously precise
and accurate relative positioning of the helicopter and the
ship. The solution implemented on the Little Bird uses GNSS
positioning and inertial navigation and is a modernization of
a system previously demonstrated in 2005.
The conventional way to achieve precise positioning with
GNSS is to transmit code and carrier phase corrections from
a stationary base station at a known coordinate to the rover
receiver. The position of the rover receiver is computed with
respect to the base station, with typical accuracies of one cen-
timeter (cm) plus one part per million (ppm) of the distance
between the base and the rover (baseline), when a fixed-inte- Prototype antenna installation
ger carrier phase solution is possible.
In order to achieve fixed-integer accuracies, a minimum
of five common satellites must be tracked by both the base
and rover. Furthermore, a continuous and robust radio link
must be maintained at all times. The failure of either of these
criteria, whether due to environmental masking of satellite
signals or an intermittent radio link, will result in the inabil-
ity to achieve the highly accurate differential solution.
When landing a helicopter onto a ship, a number of dif-
ficulties with the conventional approach to precise GNSS
positioning arise. Due to the mobility of the ship and its abil-
ity to operate in remote locations, establishing a stationary
base station becomes highly impractical if not impossible.
Even if it were possible, the varying dynamics of the ship and
helicopter can result in highly variable constellations with
respect to the base station. Moreover, substantial changes in H-6U instrument panel
satellite constellation can weaken the geometric quality of the
available observations to the point of losing the GNSS solu-
tion altogether. used with two GNSS receivers that do not move with respect
RTK algorithms solve for the position offset vector from to each other — a fixed-baseline RTK implementation — can
the base to the rover receiver. The base receiver does not have solve for the heading and pitch of the fixed baseline.
to be stationary, and it does not need a highly accurate known This algorithm can also be used with two receivers that
coordinate if the only quantity of interest is the relative dis- are moving with respect to each other — a moving base-
placement of the rover with respect to the base. An algorithm line implementation. In this case, the base receiver obtains

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 47

 LITTLE BIRD UAV

15 11 gation-grade IMU would reduce this drift to 5–8 centimeters

Relative Change in Height (meters)

over the same time interval.

Another usually beneficial aspect of GPS/INS is that the
10 6 integration filter used to combine the two systems results in

Number of SVs
a smoother solution than GNSS alone. A GNSS single-point
5 1
position will have a fair amount of variation due to mul-
tipath, atmospheric errors, and especially changes in satellite
constellation. Consequently, the typical single-point GNSS
0 -4 position standard deviation is approximately three meters,
while a typical single-point GNSS/INS position standard
deviation is less than a meter. (See Figure 2.)
-5 -9 The difficulty this poses in the inertial moving baseline
321000 322000 323000 324000 325000
case is that the ship and helicopter INS may not smooth out
GPS Time of Week (seconds)
the GNSS variations in the same way. Just as the drift of both
GNSS Height GNSS/INS Height Num. SVs. INS systems are not related to each other, the smoothing
done by both INSes is also not directly tied to the other.
FIGURE 2 Single-point GNSS height versus single-point GNSS/INS height
GNSS/INS Moving Baseline Solution
a single-point (autonomous) GNSS position solution and The rover GNSS/INS needs an absolute coordinate to update
transmits code and carrier phase corrections to the rover the INS. The coordinate used to update the rover INS is com-
based on that position. The rover then uses those corrections puted by adding the estimated relative RTK vector to the base
to compute the vector from the base to itself, resulting in an receiver’s single-point position. Both the ship and helicopter
RTK-quality solution between the two receivers, even though systems are using coordinate updates that have single-point
the absolute position solutions for the two receivers are only absolute accuracy, and the rover’s update coordinate error
of single-point quality. follows the base’s coordinate error.
The moving baseline RTK solution has the same benefits Because the rover and base single-point GNSS solution
and drawbacks as a fixed baseline RTK solution. The main errors could vary significantly due to different constellation
benefit is a very precise relative solution because the distance views or multipath, this approach minimizes the difference in
between the base and rover is quite short. The drawbacks errors in the coordinate updates used by the base and rover.
are the usual challenges of requiring constant communica- To further strengthen the relative accuracy of the INS
tion between the rover and the base, as well as maintain- solutions, delta phase updates are continually applied as well.
ing enough common satellites in view during the landing The delta phase update computes a precise position displace-
maneuvers as the helicopter approaches the ship deck. ment from carrier phase measurements differenced between
satellites and over time. This position displacement update is
GNSS/INS Integration accurate to the several millimeter level (with cycle slip detec-
An inertial navigation system (INS) is typically added to a tion in place) and is available whenever two or more satellites
GNSS solution to address issues such as these. With a GNSS/ are available. The delta phase updates will constrain the drift
INS system, the INS can “coast” through periods of GNSS of the INS solutions if a partial GNSS outage (fewer than four
signal blockage or degraded GNSS solution quality. An INS satellites) occurs and also help to further smooth out dis-
provides good relative accuracy over time, allowing it to continuities from GNSS position jumps, usually the result of
“hang onto” a high-accuracy solution. changes in the satellite constellation.
For very precise relative positioning between two systems, The differential corrections are sent from the base to the
a few limitations apply to the accuracy an INS can provide rover at 10 hertz. (The rate limit on this is imposed by the
during GNSS blockages or communication failures. The INS data link capacity not the GNSS receiver.) The INS is updated
relative accuracy is with respect to itself, and both the ship at 1 hertz. While the relative RTK solution is available — i.e.,
and helicopter GNSS/INS solution will start to drift without data link is working properly and an RTK solution is possible
GNSS aiding. Their drifts will depend on their respective — a position correction is applied to make the output GNSS/
residual inertial errors, which are not dependent on each INS position match the RTK position exactly.
other and could be drifting in opposite directions, maximiz- The update coordinate approach described previously
ing the relative ship-to-helicopter solution disparity. seeks to minimize the size of the position correction. In
The quality of the IMU incorporated into the INS will the event that the RTK solution is no longer available, this
dictate how quickly the free inertial solution will drift. For a post-update correction is only applied for 10 seconds. After
tactical-grade IMU used in a synchronized position/attitude approximately 10 seconds, the error from the inertial drift
navigation system, the position will drift 10–15 centimeters becomes larger than the GNSS to GNSS/INS offset, and
over 10 seconds in the absence of any external aiding. A navi- applying the position correction no longer has a benefit.

48 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 The initial alignment poses another
challenge as well. A stationary coarse
alignment can be performed with tac-
tical grade IMUs, but only when the
system is truly stationary. A transfer
alignment can be performed with the
GNSS course-over-ground azimuth and
pitch, but only when the vehicle’s for-
ward direction of travel is aligned to the
IMU’s forward axis (or there is a fixed,
known offset between them).
With a ship or helicopter, these
alignment conditions cannot be assured
due to crab angles, the angular dif-
ference between heading and actual
The Allure Shadow
ground path. A ship will often be mov-
The relative attitude measurement between the ship and ing enough to prevent a stationary alignment and its move-
helicopter does not benefit from the moving baseline RTK ment without any crab angle cannot be guaranteed. Even if
implementation, computed by differencing the ship and heli- an alignment is achieved, the dynamics will likely be too low
copter GNSS/INS attitude solutions. The variance of the rela- for good GNSS/INS convergence. This will degrade the qual-
tive attitude solution is effectively the combined variance of ity of the projected coordinate at the landing pad, which is
the ship and helicopter attitude solutions. where the helicopter is aiming.
The helicopter system suffers a similar challenge in initial
The VTOL UAV Application alignment. Helicopters are not an ideal platform to use a
In our application in which a helicopter lands aboard a mov- transfer alignment from GNSS course-over-ground measure-
ing ship, the quality of the attitude solution on the ship’s sys- ments, due to their maneuverability.
tem plays the most significant role in determining the overall To solve the initial alignment problem (on ship and heli-
relative accuracy. The ship’s GNSS/INS system is mounted copter) and to address the attitude error convergence/observ-
in a convenient location away from the landing pad, but the ability problem (on the ship), the GNSS/INS was augmented
landing pad is the true point of interest. Similarly, the land- with a second GNSS receiver and antenna, using the fixed
ing gear is the point of interest on the helicopter, not the loca- baseline implementation of the relative RTK algorithm. The
tion of the inertial measurement unit (IMU). ship’s GNSS/INS has two GNSS antennas associated with it,
Both GNSS/INS systems must project their solutions from as does the helicopter’s GNSS/INS. The offset vector from the
the IMU to the point of interest. To implement this coordi- IMU to both antennas must be measured and input.
nate projection, the offset vector from the IMU must be mea- The pitch and heading of the baseline between the two
sured in the IMU frame, and the rotation matrix between antennas is used for the initial INS alignment. Because the
the IMU reference frame and the GNSS’s Earth-centered roll angle cannot be observed with just two antennas, it is
Earth-fixed (ECEF) frame must be known. The accuracy of assumed to be zero in the initial alignment. After alignment,
the solution at the point of interest therefore depends on the the GNSS azimuth is used as a heading update to the INS.
quality of the measured offset as well as the quality of the This solution is critical for the ship system, because the
rotation matrix from the IMU frame to the ECEF frame. ship will be experiencing low dynamics, making the attitude
This rotation matrix is maintained as part of the INS errors less observable. For the helicopter system, the GNSS
solution. The quality of the rotation matrix is very dependent azimuth updates are not as vital because the helicopter
on the quality of the initial INS alignment (i.e., finding the maneuvers much more and its attitude errors are generally
IMU’s orientation with respect to gravity and north), and the observable via the vehicle dynamics.
overall convergence of the GNSS/INS solution. The longer the
offset vector is to the landing pad, the larger the effect of the Test Crew Training For Maritime Operations
rotation matrix errors (i.e., a classic pointing error in survey Flight test operations involving the trailer-borne helipad
terminology). resulted in more than 100 landings to the moving rig.
Attitude errors in GNSS/INS are best observed with vehi- Although landing on this test helipad was still a challeng-
cle dynamics. In particular, horizontal accelerations allow ing feat given the landing deck’s 16x16-foot dimensions and
the azimuth error to be observed and controlled. Depending the H-6U landing gear width of slightly more than 8 feet,
on the size and speed of the vessel, the dynamics observed the only motion environment the heli-deck presented to the
aboard a ship can be very low, leading to degradation in the safety pilot was linear translation down the Spaceport Amer-
azimuth solution. ica runway.

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 49

 LITTLE BIRD UAV

Ships at sea, however, exhibit the following heli-deck An H-6U helicopter has the following dimensions: main
motion: pitch; roll; yaw; heave; sway, and surge. Ships also rotor diameter, 27.5 feet; tail rotor diameter, 4.75 feet; total
don’t move across the Earth in the same direction as their helicopter length, main rotor tip to rotor tip, 32.3 feet. The
heading due to local water currents, a factor that must be stinger, the lowest part of the H-6U’s vertical stabilizer, is
accommodated in the flight control laws. Moreover, conduct- approximately 2.5 feet above the landing surface.
ing terminal flight operations in the intended operational The Squadron advised a minimum of 3 feet lateral clear-
environment must also deal with the wind turbulence gen- ance from the stinger to the edge of the helipad where the
erated by a ship’s superstructure. These factors created a safety net frames protruded upwards, and a minimum of 10
requirement for safety pilot training in a maritime environ- feet lateral clearance between the main rotor blades and the
ment. closest ship structure.
Because the flight tests on the French frigate would A careful survey of the helipad yielded a zone of approxi-
require four Boeing engineers and technicians to reside on mately 5 feet fore and aft in which the safety pilot could
board for a period of two weeks, the Little Bird program allow the H-6U to land and insure safe structural clearance.
needed to find both a suitable training vessel and quali- Simple but highly effective markers were installed to create a
fied trainers to work with the engineers. After an extensive visual cue environment that could enhance the flight crew’s
search, the ULB team located a helipad-equipped yacht — judgment regarding a safe landing zone. The proximity of
Shadow Marine’s Allure Shadow — based in Fort Lauderdale, the helicopter rotors to the yacht structure, while fairly tight
Florida, that was available for lease by the week. compared to dimensions generally found on DoD vessels, is
Boeing’s ULB program engaged The Squadron, a com- common in the super yacht world.
pany that specializes in training both flight crews and deck Risk mitigation dictated that Boeing provide H-6U trained
hands in the super yacht industry, to provide maritime envi- fire/crash rescue personnel and firefighting equipment, inde-
ronment training to the test team. Staffed by helicopter pilots pendent of the Allure Shadow crew. The Squadron conducted
formerly with the U.S. Department of Defense (DoD) and a review of all yacht safety equipment and emergency proce-
the United Kingdom Ministry of Defense, The Squadron was dures, provided maritime environment training to Boeing
able to provide deck qualification pilot training equal to or fire/crash rescue personnel, and trained the flight test engi-
exceeding U.S. and U.K. military standards. (An interesting neering staff in shipboard flight operations procedures.
aside: the Federal Aviation Administration has no certifica-
tion similar to the deck qualification training common to a Ship State Monitoring
military training program.) Knowledge of the ship motion while under way is crucial to
An additional safety requirement for the flight crew was insuring that the limitations of the test helicopter are respect-
helicopter dunk tank training, which they completed at Loui- ed during landing and takeoff operations. A system devel-
siana State University’s facility in Lafayette, Louisiana, before oped for the U.S. Navy by Hoffman Engineering Associates
the flight test program began. — the “Landing Period Designator” or LPD — was installed
and operated by the developer during the flight test program.
Squeezing a Helicopter into a
Moving Landing Zone
Landing the H-6U helicopter safely on a
moving yacht had everything to do with
the relative dimensions of both and the
adjacent physical structures.
The Allure Shadow is equipped with
a helipad that measures 34 feet wide by
50 feet long and is surrounded on three
sides by horizontal safety nets, which
are raised about 5 inches above the heli-
pad surface. At the forward edge of the
helipad is an overhang from a pool deck
located next to and above the landing
zone. Both features are visible in the
accompanying photos.
The pool deck overhang presents a
contact hazard for the helicopter main
rotor system while the helipad’s safety net
system presents a contact hazard for the
helicopter’s tail structure. Structural clearances

50 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 This system provided trend information, absolute deck
motion data, and an indication of deck conditions (color-
coded green, yellow, or red; see Figure 3). Use of the LPD
ensured that terminal operations were conducted within
the limitations of the test helicopter. Of particular interest
was the ship motion in the sway axis (lateral back and forth
motion), which could contribute to a dynamic rollover event
if limits were exceeded.

Test Setup and Description

The ULB team used survey instruments to measure the lever
arms (offset vector from IMU to GNSS antenna) and point-
of-interest offset vectors while the ship was docked. During Boeing fire/crash rescue crew
the survey, it was exceptional windy, leading to ship motion
and lower accuracy lever-arm determinations than desired.
The H-6U was equipped with the primary antenna on
the “T” tail and a secondary antenna on the nose. A laser
micrometer mounted on the belly center would measure
absolute displacement of the belly above the heli-deck on
initial touchdown and the final height after the landing gear
had settled.
Tables 1 and 2 show the measured offset vectors. Because
the lever arm quality was suspect during the test, the helicop-
ter executed a specific set of figure-eight maneuvers to allow
for lever arm estimation in post-processing.. In post-process-
ing, a significant error in the height component appeared in
estimation of the primary lever arm for the ship system. Table Portable fire suppression system
3 gives the post-processed ship offset vector sum, which are
accurate to approximately 10 centimeters.
Data links transmitted differential correction data
between the ship and the helicopter and also transmitted the
real-time relative ship-to-helicopter solution, output in the
log RELINSPVA, back to the “command center” via radio
link. The GNSS/INS receivers on board the Allure Shadow
logged raw inertial and GNSS data in order to be able to
post-process the ship and helicopter conventional RTK tra-
jectories. In post-processing we also used data collected at the

Command center view

GPS antenna installations

FIGURE 3 Landing Period Designator indicating a “yellow deck” condition

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 51

 LITTLE BIRD UAV

System Lever Arm X (m) Y (m) Z (m)

Starboard Antenna 2.80 -0.36 5.44
GNSS/IMU on ship (Primary)
Port Antenna -2.61 -0.36 5.44
Starboard Antenna 0.42 -4.74 1.20
GNSS/IMU on (Primary)
helicopter
Port Antenna 0.28 2.13 -0.39
TABLE 1. Surveyed lever arms

From To X (m) Y (m) Z (m)

GNSS/IMU on ship Helipad Target 0.09 -12.52 -0.13
GNSS/IMU on helicopter Aircraft Belly 0.28 0.31 -1.32
GNSS/IMU installation
TABLE 2. Surveyed IMU to point of interest offsets

Lever Arm X (m) Y (m) Z (m)

Starboard Antenna (Primary) 2.80 -0.36 5.08
Offset to Helipad Target 0.09 -12.52 -0.09
TABLE 3. Post-processed ship system offset vectors

National Geodetic Survey’s continually operating reference

station (CORS) “LAUD” located near Fort Lauderdale about
25 kilometers from the test area.
The accuracy of each post-processed trajectory was about
three centimeters. For performance analysis, the real-time
ship-to-helicopter relative position vector was compared to
Total station survey the post-processed ship-to-helicopter relative position vector.
The true test of the system’s performance, however, came
in the real-time testing as demonstrated in several successful
autonomous landings. Tests were undertaken on July 4 and 5,
2012, at sea off the coast from Fort Lauderdale, Florida.

Test Results from July 4

For most of the morning, the aircraft performed maneuvers
behind the boat, following its movement. Although we would
have preferred a sea state or 3 or 4 (wave height between 0.5
and 2.5 meters) during the trials, the water surface was essen-
tially flat, sea state 0. The H-6U was also allowed to approach
the landing pad and hover over the landing point to provide
a sufficient confidence level that the system was functioning
as expected. The aircraft then performed a single automated
Primary and nose GPS antennas
landing before returning to the airport for fuel. Figure 4
shows the trajectories of the boat (green) and the aircraft
(red) during these operations.
The aircraft autonomously landed on the helipad at GPS
time 316350–316772 seconds. The GNSS/INS system on the
helicopter reported a real-time relative position to the heli-
pad center of 0.024m North, -0.028m East, and 1.09m Up.
The helicopter belly height measured was approximately 64
centimeters; so, the real-time results seem to have about 40
centimeters of vertical error, which matches the vertical error
of the lever arm.
Laser micrometer installation In post-processing, the new lever arm was used and
the average relative position values of the helicopter on the

52 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 the real-time solution shows about 35 centimeters of height
error due to the lever arm used in real-time.
The nature of the test program did not allow for extensive
tuning of the automated flight control system to respond in
an optimal fashion to the navigation data input. Neverthe-
less, the results from the initial test program were impressive.
Table 4 presents the difference between the H-6U position
at 10 feet above the helipad and after landings to the helipad
during one sortie.
The radar altimeter output compared very favorably with
the RTK solution, as shown in Figure 7. Installation of the
antenna on the tail boom should offer excellent functionality
for slung load operations in which the load interferes with the
radio wave returns, or landing to a deck-lock-grid–equipped
deck where multipath below the mechanical grid surface will
render the output completely unreliable.
Incorporation of reliable radar altimeter data into the ter-
FIGURE 4 July 4 morning test trajectory minal operations solution enhances the reliability and redun-
dancy of the navigation method.
10’ over the pad On the pad
Landing Longitudinal
Lateral (ft)
Longitudinal
Lateral (ft)
Conclusion
(ft) (ft) Flight tests performed in 2005 and 2006 provided encourag-
1 0.5 Aft 0.1 Right 1.5 Fwd 0.1 Right ing results for the initial development of a moving-baseline
2 1.2 Aft 0.7 Right 0.5 Aft 0.8 Right relative navigation system. Extensive flight test activity at
1.0 Aft 0.2 Right 0.3 Fwd 0.6 Left
Spaceport America vetted the integrity of the GNSS/INS sys-
3
tem as a performance evaluation tool for navigation systems
4 0.7 Fwd 0.1 Left 2.6 Fwd 0.1 Left
such as the Thales DAA radionavigation system.
5 0.2 Fwd 0.5 Left 0.5 Fwd 0.3 Right Flight test activity in 2011 at Spaceport America demon-
6 1.0 Fwd 0.4 Right 1.5 Fwd 0.7 Left strated the integrity and accuracy of the further GNSS/INS
TABLE 4. Ship landing guidance and control errors solution, certifying that system for TSPI (time space posi-
tion information) applications. More than 100 landings were
landing pad were -0.383m North, -0.298 East, and 0.771 made to the moving trailer helipad.
Up, which agrees much better to the known helicopter belly Our most recent maritime flight tests during the sum-
height. Figure 5 shows the real-time relative solution of the mer of 2012 demonstrated the accuracy of the navigation
helicopter landing gear to the landing pad. Figure 6 shows the solution, as well as the integration of the navigation solution
difference between the real-time and post-processed relative with the automated flight control system on the Boeing H-6U
position solutions as the helicopter was landed. Recall that Unmanned Little Bird.

2500
0.35
2000
0.3
Position Displacement (m)

Position Displacement (m)

1500
0.25
1000 0.2
500 0.15
0 0.1
0.05
-500
0
-1000
-0.05
-1500
3.13 3.14 3.15 3.16 3.17 3.1635 3.1636 3.1637 3.1638 3.1639 3.164
GPS Time (s) × 105 GPS Time (s) × 105

FIGURE 5 Real-time relative position solution FIGURE 6 Real-time to post-processed relative vector differences

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 53

 LITTLE BIRD UAV

the French frigate-landing test experienced 84 landings to

and takeoffs from the yacht.
In all, seven sequential days were required to accomplish
the deck qualifications of two Boeing test pilots, integrate and
debug all systems and software, and carry out maritime ter-
minal operations until the operation became routine. The on-
board safety pilot was able to allow the autonomous systems
to misbehave enough so as to rapidly gain a good understand-
ing of system malfunctions and how to correct them.
Once again, these efforts demonstrated the value of the
Unmanned Little Bird program’s optionally manned system
architecture.

Manufacturers
Both the ship and the helicopter were outfitted with SPAN-
SE- dual antenna GNSS/INS receivers from NovAtel, Inc.,
Calgary, Alberta, Canada. The ship system used two NovAtel
702GL antennas. The ship system used an LN200 IMU from
Northrop Grumman, Woodland Hills, California USA,
while the helicopter system used a HG1700 AG58 IMU from
Honeywell Aerospace, Phoenix, Arizona USA. The data
links used between the SPAN-SE-D receivers were Micro-
hard 2.4 GHz IP2421 frequency-hopping spread-spectrum
RF modem radios from Microhard Systems Inc., Calgary,
Alberta, Canada. The relative RTK solutions was provided by
NovAtel’s ALIGN algorithm. Post-processing of the CORS
data with the shipboard and helicopter GNSS/INS data used
the Inertial Explorer post-processing software from NovAtel’s
Waypoint Products Group. The glass cockpit display system
GNSS/INS processor (lower left) and inertial measurement unit used in Boeing Flight Operation’s Eurocopter AS-350B3 and
installations the H-6U was the Garmin G500H dual-screen, flat panel
electronic flight display, from Garmin International, Olathe,
A total of 16 fully autonomous landings and 13 fully Kansas USA. The Garmin cockpit avionics suite (Figure
autonomous takeoff/departures comprised this latest effort, 8) consists of a GMA350H communications control panel,
with the flight crew closely monitoring the controls and the GTN635 VHF/GPS nav/com panel, GTS800 traffic collision
aircraft position when the aircraft was in close proximity to avoidance system (TCAS), GTX33H Mode S transponder,
the deck. In the process, the project test pilot responsible for G500H integrated primary flight display (PFD) and a multi-
function display (MFD), and a GRA5500 RADAR altimeter.

Acknowledgements
300 The authors wish to acknowledge the invaluable contribu-
tions of the following individuals without whom the safe and
efficient execution of the H-6U maritime operations flight
DGPS Altitude (ft)

200 test could not have been accomplished.

Radar Alt (ft)

Joshua Pfleeger, procurement agent, Boeing Test & Evalu-

ation for his tireless and determined efforts to get both The
Squadron and the Allure Shadow under contract, on time
100
and under budget.
Dan Deutermann and Jim Frean of The Squadron, whose
knowledge and professionalism made the difference between
0 success and failure of the test program.
2900 3000 3100 3200 3300 Brook O’Neill, super yacht captain, and Garrin Ham-
Time (s)
mond, rescue tender operator, for their invaluable contribu-
FIGURE 7 Radar altimeter vertical data compared to GNSS/INS vertical tions towards the safe and effective execution of the flight test
data
program.

54 InsideGNSS M AY/ JUNE 2013 www.insidegnss.com

 Dr. Bernard Ferrier, Engineering Head, Aircraft/Ship lead for the SPAN software group in 2007, then finally the engineering man-
Dynamic Interface Program, Hoffman Engineering Corpo- ager for all SPAN products in 2009. Her area of expertise is primarily GPS/
ration for his support with the Landing Period Designator INS integration, exploiting navigation grade, tactical grade, and commercial
microelectromechanical system (MEMS) inertial measurement units for air-
system.
borne, land, and surface sea applications of varying accuracy and cost tar-
Dan Ciernia of CAI TV, who captured all the critical ele-
gets.
ments of the test program on still and video imagery and cre-
Sheena Dixon began her career at General Dynamics
ated a great story.
Canada after earning a bachelor of science degree
Mike Bobye and Tom Ford (retired) of NovAtel, Inc., for
in geomatics engineering from the University of
their willingness to support the development of this unique Calgary. She joined NovAtel in 2006 as a systems
navigation system capability, and for the ingenuity of their engineer in the aviation department, working on
methods. reference GNSS receivers and participating in the
Michael Brown, president of Steve Leiber and Associates, early development of Galileo reference receivers.
whose knowledge and technical support of NovAtel’s product Dixon joined the company’s applications engineering group in 2008, where
line is unparalleled. she provided technical assistance to customers using and integrating NovA-
Joe Killian, Killian Yacht and Ship Brokers, who “found a tel’s OEM products. Dixon eventually specialized in SPAN GNSS/INS product
way” to get the Allure Shadow under contract. applications, including unmanned vehicle navigation, hydrographic survey,
Dino Cerchie, Unmanned Little Bird Program manager, aerial survey, and mobile mapping. She is now the SPAN technical lead in
the applications engineering group.
Rick Lemaster, director of Unmanned Airborne Systems,
and Debbie Rub, vice-president and general manager, Mis- Travis Berka joined Boeing Directed Energy Systems
siles and Unmanned Airborne Systems for Boeing Military in 2007, after graduating from the University of
Arizona with a bachelor of science degree in aero-
Aircraft, for their trust in the H-6U flight test team, and their
space engineering. At Boeing, he supports a variety
determination to secure IRAD (internal research and devel-
of programs with both field testing and data analy-
opment) funding for this extraordinary test program. sis. Berka specializes in modeling, simulation, and
This article is based on a paper presented at the AUVSI algorithm development for both directed energy
Unmanned Systems 2012 conference. and space systems. His other areas of expertise include LIDAR data process-
ing and visualization techniques, integrated differential GPS and inertial
Additional Resources measurement system performance evaluation, and counter rocket, artillery,
Ford, T., and M. Bobye, and M. Hardesty, “Helicopter Shipboard Landing and mortar (C-RAM) mission support.
System,” Institute of Navigation GNSS 2005 conference, Long Beach, Cali- Jason Graham is the ground systems lead for the
fornia, USA, September, 2005 Unmanned Little Bird program, serving as the hard-
ware and software architect for the ULB ground
Authors control station since program inception in 2003. In
that role, he interfaces with customers and ven-
Mark Hardesty is the flight test director for Boeing’s
dors, coordinates efforts with remote team mem-
Unmanned Little Bird (ULB) program, having led
bers, establishes customer requirements, develops
numerous flight test campaigns including various
concepts of operation, designs and authors software in C++, conducts sys-
weapons, sensor, and communication relay inte-
tem level testing from desk side simulation through flight test, and sup-
gration efforts. Most recently Hardesty designed
ports planning and conduct of UAV operations. Graham joined Boeing after
and led test campaigns to demonstrate precision
receiving a B.S. degree in aerospace engineering from Embry–Riddle Aero-
autonomous terminal operations of the H-6U
nautical University and later attaining a M.S. in technical management from
Unmanned Little Bird for maritime operations, including a helipad equipped
the same institution.
trailer at Spaceport America in New Mexico, a super yacht off the Florida
coast, and finally on a French frigate in the Mediterranean Sea. He attended Donald G. Caldwell received a BSE degree from Ari-
North Carolina State University where he earned both B.S. and Master’s zona State University and completed course work
degrees in mechanical engineering. Since joining Boeing in 1986, he has for an M.S. in electrical engineering while working
authored numerous technical publications and earned several patents relat- at Sperry Flight Systems. He has worked in the area
ing to precision navigation and helicopter robotics technology. of fixed- and rotary-wing aircraft guidance, navi-
gation, and control (GNC) for most of his career.
Sandy Kennedy is the principal engineer for all
With more than 20 years at Boeing, Caldwell has
NovAtel SPAN products, managing the advanced
been the lead control-law designer for the Apache fly-by-wire demonstra-
development group and overseeing the transfor-
tor as well as the Unmanned Little Bird. From its inception in 2003, ULB has
mation of new and enhanced technologies with a
demonstrated various integrated sensor technologies and autonomous con-
one to three–year horizon into robust products for
trol for weapon delivery, obstacle avoidance, cargo pickup and delivery,
both commercial and military customers. She
and, most recently, shipboard takeoff and landing operations.
joined NovAtel in 2004 after completing a B.Sc.
with Honors and an M.Sc. in geomatics engineering at the University of Cal-
gary. Kennedy helped launch the Synchronized Position Attitude Navigation
(SPAN) product line in 2005, became the functional manager and technical

www.insidegnss.com M AY/ JUNE 2013 InsideGNSS 55