NRL Spectra - Winter 2014

8/9/2019 NRL Spectra - Winter 2014

1/40

NAVAL RESEARCH LABORATORY

THE MAGAZINE OF THE NAV Y’S CORPORATE LABORAT

W INTER 20

updates on

NRL’S

AUTONOMY

RESEARCH


2/40

4555 Overlook Ave, SW

Washington, DC 20375

(202) 767-2541

www.nrl.navy.mil/spectra

COMMANDING OFFICERCAPT Mark C. Bruington, USN

DIRECTOR OF RESEARCHDr. John A. Montgomery

T H E

NAVAL RESEARCH LABORATORY

PUBLISHER

NRL Oce of Public AairsRichard Thompson, PAO

MANAGING EDITORDonna McKinney, Deputy PAO

EDITING, DESIGN, AND PRODUCTION NRL Technical Information Services

Kathy Parrish, Head

EDITINGClaire Peachey

DESIGN AND PRODUCTIONJonna Atkinson

PHOTOGRAPHY AND ARCHIVES

Gayle FullertonJamie Hartman

James Marshall

AN OFFICIAL PUBLICATION OF THE NAVAL RESEARCH LABORATOR

Welcome to the fourth issue of SPECTRA, a magazine

designed to inform you of exciting science and technology

developments at the U.S. Naval Research Laboratory (NRL).

In the spring of 2012, NRL opened the Laboratory for

Autonomous Systems Research (LASR). Our vision for this

shining, state-of-the-art facility was that it would be a focal

point for basic and applied research related to unmanned and

autonomous systems – a place to develop, integrate, and

test systems and prototypes, to advance the Navy’s scientic

leadership in this domain. Now, a little more than two years

later, the LASR has become the collaborative and energetic

research space we envisioned, with scientists and engineers

from all across NRL working together on interdisciplinary

projects in the one-of-a-kind facilities and environments that

LASR oers. Autonomy research for the Navy and Department

of Defense benets greatly from this convergence of diverse

talent and expertise.

In its simulated desert, jungle, and littoral environments

and other well-equipped laboratories, the LASR is hosting

creative and innovative research in intelligent autonomy,

sensor systems, power and energy systems, human–system

interaction, networking and communications, and platforms.

This issue of SPECTRA provides a snapshot of some of the

LASR projects now moving forward:

• an autonomous underwater vehicle named WANDA that

borrows its design from a coral reef sh,

• a next-generation hydrogen fuel cell built using 3-D printing,• a battery that draws its power from the ocean bottom

sediment,

• a robot that is learning to understand how people think, and

• a vehicle that travels through air and through water – a ying

submarine.

We hope you enjoy this issue of SPECTRA and share it with

others. To request additional copies or more information, please

email [email protected].


3/40

Contents

30 Navy Launches UAV from Submerged Submarine

31 Captain Mark C. Bruington Relieves Captain Anthony J. Ferrari

and Captain Anthony J. Ferrari Receives Legion of Merit

2 Bio-Inspired Designs for Near-Shore AUVs

6 Flimmer: A Flying Submarine

10 Cognition for Human–Robot Interaction

14 NRL Autonomy Lab Hosts Shipboard Fire

Robotics Consortium

18 NRL Technology Assists FDNY

20 The Benthic Microbial Fuel Cell

24 3-D Printing in Hydrogen Fuel Cells for

Unmanned Systems

28 More LASR Projects

F E A T U R E S

N E W S B R I E F S

O U T R E A C H

32 NRL and Aerospace Industry Host 10th AnnualCanSat Student Challenge

MAK I NG CONN EC T I O N S

34 Sharing Science ... Starting Conversations

T ECHNOLOGY TRANSFER

36 Dr. Jeremy Robinson Receives Presidential Early Career Award

37 In Memory of Richard J. Foch

35 New York City Tracks Fireghters to Scene with NRL Radio Tags and Automated Display

FOCUS ON PEOPLE

ON THE COVER

Flight-testing the Flimmer UAV.See page 6.


4/40

NRL FEATURES

SPECTRA2

Autonomous underwater vehicles

(AUVs) have demonstrated manycapabilities in inspection, surveil-lance, exploration, and object detec-tion in deep seas, at high speeds,and over long distances. However,the low-speed, high-maneuverabilityoperations required for near-shore andlittoral zone missions present mobil-ity and sensing challenges that havenot been satisfactorily solved, despitenearly half a century of AUV develop-ment. Very shallow water and littoralenvironments are complex operating

zones, often turbid, cluttered withobstacles, and with dynamicallychanging currents and wave action.

AUVs need to be able to maneuveraround obstacles, operate at lowspeeds, hover, and counteract surgeand currents. To develop AUVs thatcan successfully navigate and oper-ate in these dynamic and challengingenvironments, Naval Research Labo-ratory (NRL) researchers have taken

inspiration from nature — from sh,

in particular — to design and developnovel underwater propulsion, control,and sensing solutions.

For AUV propulsion and control, NRLhas developed an actively controlledcurvature robotic n based on the

pectoral n of a coral reef sh, the

bird wrasse ( Gomphosus varius ). Inthe most recent n iteration, four rib

spars connected by a exible synthetic

membrane dene the n geometry,

and control of each rib deection angle

enables the shape deformation neededto generate 3-D vectored thrust. Usingcomputational uid dynamics (CFD)

tools in conjunction with experiments,sets of high-thrust-producing n stroke

kinematics, or n gaits, have been

identied. This kind of articial n tech-nology can adapt to varying ow condi-tions and provide the thrust controlnecessary for low-speed maneuveringand precise positioning.

The NRL articial pectoral n has

been integrated into a man-portableunmanned vehicle named WANDA,the Wrasse-inspired Agile Near-shorDeformable-n Automaton. Four

side-mounted ns, two forward and

two aft, provide all the propulsion ancontrol necessary for the vehicle. Aset of custom control algorithms useinformation about the vehicle motionand surrounding environment to in-form changes to the n stroke kine-matics. By weighting and combiningvarious n gaits, the magnitude and

direction of thrust generated by eacn can be controlled to produce the

desired eects on vehicle motion.

Computational and in-water experi-mental results have demonstratedWANDA’s capabilities. WANDA canperform low-speed maneuvers incluing forward and vertical translation,turn-in-place rotation, and stationkeeping in the presence of waves.

Bio-Inspired Designs forNear-Shore AUVs


5/40

NRL FEATURES

WINTER 2014

WANDA can also successfully coordi-nate maneuvers to achieve waypoint

navigation. WANDA is designed tooperate at speeds in excess of twoknots, or hold position in the presenceof two-knot currents, giving it the pro-pulsion and control authority neededin many harbor and other near-shoreoperational zones.

Solutions for vehicle propulsion andcontrol address only part of the chal-lenge of operating in shallow under-

water areas, so NRL is also develop-ing new sensor systems for AUVs,

and again looking at sh for ideas. Areliable system for navigating littoralenvironments requires real-time knowl-edge of current velocities and objectpositions. Conventional onboard sen-sors such as sonar and vision-basedsystems have shortfalls in a cluttered,shallow environment: sonar can suer

from multipath propagation issues, andvision-based systems are limited byturbidity.

Additionally, these are active sys-tems that emit sound or light, which

uses energy and exposes an AUV todetection.

Fish use a system of hair-like ow

and pressure sensors, called a lateral line, to detect changes and obstaclesin the environment around them. NRLhas studied these natural systemsand is developing an articial lateral

line, a system of pressure sensorsmounted on the AUV hull to provide

The bird wrasse ( Gomphosus varius ) isa coral reef dwelling sh operating in

depths of 1 to 30 meters. It is known touse its paired median pectoral ns to

produce thrust for maneuvering in thesecomplex environments.

The geometry of the bird wrasse was modeled and computational uid

dynamics simulation was validated against experimental results.

Following design studies for a robotic pectoral n, CFD simulation

was used to determine eectiveness of this device as a means of

propulsion and control for an AUV.


6/40

NRL FEATURES

SPECTRA4

passive sensing of surrounding watervelocities and near-eld objects. This

system uses quartz crystal–basedpressure sensors with high sensitivityto measure minute pressure dier-entials between various points onthe vehicle hull. The amplitude andphase dierences between sensor

signals provide information about thesurrounding environment that the ve-

hicle can use to perform maneuverssuch as orienting into a ow or avoid-ing obstacles. NRL’s articial lateral

line will help WANDA and other AUVsachieve better performance in envi-ronments where sonar or vision alonecannot be relied on.

Much of the development and testingof the WANDA platforms takes place

Demonstrating the WANDA AUV in the Littoral High Bay pool in NRL’s Laboratory for Autonomous Systems Research. The four-n design enables the propulsion and control

needed for station-keeping and maneuvering in the presence of external disturbancessuch as waves and currents. The cameras in the foreground of the top photo are used tocapture the three-dimensional motion of the vehicle in real time.

in NRL’s Laboratory for AutonomousSystems Research (LASR). This facil-ity houses vehicle prototyping tools,such as 3-D printers and metalwork-ing equipment, and large-scale testenvironments. The 45 foot by 25 footpool in the LASR Littoral High Bayserves as the primary underwatertest environment for WANDA, wherea 16-channel wave generator creates


7/40

NRL FEATURES

WINTER 2014

real-world simulated near-shore con-ditions, and a 12-camera underwatertracking system provides groundtruth position and orientation mea-surements for validation of vehicleperformance.

As WANDA’s sh-inspired technolo-gies are perfected, the AUV is beingprepared for payload testing. The ve-hicle’s modular construction enableseasy integration of dierent mission-

specic payload packages, and one

such payload that will be developedand tested on WANDA starting thisyear is a biochemical sensing systemfor trace level detection of chemicalsignatures. This sensor system builtonto a capable low-speed platformsuch as WANDA will enable missionsin plume tracking and target localiza-tion in shallow water environments.

The WANDA program has spawnedother related programs to enableNavy critical missions using AUVs.NRL’s new Flimmer (Flying Swim-mer) program seeks to develop anunmanned platform that will be de-ployed from the air, glide to a watersurface landing in an area of interest,and transition into a swimming AUV.NRL is leveraging its expertise inunmanned air vehicle (UAV) technolo-gies to design and build the Flimmervehicle. Modications have been

made to the WANDA ns to enable

them to function as aerodynamiccontrol surfaces and to survive theimpact of landing (see page 6 forfurther details on Flimmer).

As the Navy’s focus on autonomyand unmanned systems intensi-es, NRL’s bio-inspired research

The WANDA AUV deployed for testing in the Littoral High Bay in NRL’s Laboratory for Autonomous Systems Research.

into capable propulsion and controltechnologies for low-speed operationin near-shore environments is helpingto close a clear gap in AUV technol-ogy. An unmanned vehicle that caneectively operate in these areas,

where traditional platforms experi-ence stability and control problems,will improve performance for criticalmissions including harbor monitor-ing and protection, hull inspection,covert very shallow water operations,and riverine operations.

By Jason GederNRL Laboratories for Computational Physics

and Fluid Dynamics


8/40

NRL FEATURES

SPECTRA6

The Flimmer (Flying Swimmer)

program at the NavalResearch Laboratory (NRL) ismerging two research areas

to provide a novel airbornedelivery method for unmannedunderwater vehicles (UUVs).Underwater vehicles are limited intop speed by the drag produced in

water. However, air is approximately1000 times less dense than water, sothe power required to overcome dragis substantially reduced in air. Byying over the surface of the water

rather than swimming long distancesthrough it, delivery of a UUV to anoperational area can be accom-plished much more quickly, and de-livery is possible to areas not directlyaccessible through continuous water

pathways. The Flimmer programseeks to investigate the potential ofrapidly ying a submarine over the

ocean’s surface into position, tran-sitioning from ight to underwater,

and then enabling a swimming modeonce underwater.

Major Design Considerations

In general terms, an aircraft is aspecic outer shape held rigidly in

place with a series of internal struc-tural elements. Weight is the enemyof an aircraft designer: an increase inweight requires additional lift to keepthe vehicle aloft. Drag is increaseddue to this extra lift, which requiresmore power to overcome that drag.To keep aircraft lightweight, they aretypically manufactured as a thin skin

supported by minimal internal struc-tures with low safety factor margins.

On the other hand, underwatervehicles are built completely the op-posite way. As an underwater vehicledescends, the pressure applied bythe water column increases rapidly,requiring the use of a strong pressure

vessel to protect the electronics fromwater. These pressure vessels aretypically thick metal pressure-resistant shapes with substantialseals. Inherent to enclosing air vol-ume is an increase in the vehicle’sbuoyancy, requiring sucient ballast

weight to bring the overall vehicle tonear neutral buoyancy. Instead of us-ing ballast weight, many UUV design-ers elect to make the pressure vessel

Flimmer:A Flying Submarine

Flimmer team (left to right): Trent Young, Jason GedeEdwards, Marius Pruessner (not shown: Ravi Ramam


9/40

NRL FEATURES

WINTER 2014

heavier than required for the desireddepth pressure.

Combining these two diametricallyopposed vehicles to design a ying

submarine comes down to a balanc-ing act between buoyancy, weight,

and structural elements. For a sub-marine to y, the enclosed air volume,

which is the main driver of weight fora submarine, needs to be reduced asmuch as possible. For an aircraft toland on the water, its structural ele-ments need to be more robust to sur-vive the high impact of splashdown.

The Flimmer program adds a furthercomplication into the design: apping

ns are used for underwater propul-sion. As learned from NRL’s WANDA

UUV (see article on page 2), a four-nned conguration provides high

maneuverability and good stabilityunderwater. In air, however, the ns

add weight and are relatively fragilemechanisms that need to be able tosurvive the forces of splashdown.Bringing all these design elementstogether is the central challenge of theFlimmer program.

Test Sub

The Flimmer program started with justthe combination aircraft and subma-rine, without the additional complexityof ns. The “Test Sub” conguration

was born from combining a traditionalsubmarine shape with a traditionalaircraft shape. Test Sub uses a xed

geometry in both air and water, carry-ing the drag penalty of large surfacearea wings in the water, instead ofspending weight and complexity ona wing folding mechanism. Test Subalso carries a weight penalty in the

aircraft mode so that it can enter thewater at full ight speed, spending

weight to make the structures survivethis high impact loading.

The Littoral High Bay test pool inNRL’s Laboratory for AutonomousSystems Research (LASR) was hometo Test Sub during its initial develop-ment. Underwater testing focused ondevelopment of the submarine sub-

systems, to include primarily buoy-ancy control and kinematic controls.

A custom-built buoyancy controlsystem uses two inatable bladders

placed at the front and rear inside thehull fairing. A control loop pumps airinto the two bladders for a combina-tion of both pitch and heave controlwhen the vehicle is stationary (in mo-tion, the tail surfaces are used). Air isscavenged from inside the pressurevessel to inate the bladders. While

using air in bladders is a lightweight

solution, it does have the drawbackof being unstable with depth, requir-ing constant control inputs.

The most interesting aspect of run-ning Test Sub in the LASR pool wasthat controlling the vehicle in forwardmotion underwater was identical to“ying” the aircraft. The conguration

of an aircraft shape and traditionalaircraft control surfaces of rudder, el-evator, and ailerons functions exactlythe same whether in air or water.

Several test runs showed excellentcontrollability and maneuverability ofTest Sub in the water while movingforward.

For ight testing, Test Sub was taken

to a local test range. Three free-ights started with an air-drop of the

vehicle from a mother ship at ap-proximately 1000 feet altitude. TestSub was guided manually in a series

of descending orbits to a splash-down landing in the water. Test Subew as any other aircraft, controllable

in three axes and exhibiting sucient

stability for man-in-the-loop ight.

Once nearing the water surface, TestSub was guided along a standardapproach at an airspeed of approxi-mately 40 knots before splashdownwith wings level. Upon touching thewater surface, the aircraft saw a dra-matic increase in drag and deceler-ated abruptly. After the splash, Test

Sub submerged and started movingunderwater.

Flying WANDA With the success of Test Sub, theFlimmer team applied the lessonsto designing a method for ying

NRL’s WANDA vehicle. The “Flying

WANDA” conguration has four ns

and the addition of a wing, with thetwo aft ns mounted on the tips of

the wing. This allows keeping thesame control techniques developed

for the four-nned UUV design, butprovides a lifting surface for carryingthe weight of all the apping mecha-nisms.

As the ns are a signicant surface

area relative to the wing area, theyhave been designed to pull double-duty. While swimming, the ns act

as apping propulsors. In ight, the

wingtip-mounted ns are turned up

Test Sub maneuvering as a submarine in the Littoral High Bay pool in NRL’s

Laboratory for Autonomous Systems Research.


10/40

NRL FEATURES

SPECTRA8

to act as xed vertical stabilizers

for lateral stability, and the forward-mounted ns act as close-coupled

canards ahead of the wing. With thisdual-use of the n mechanisms, the

incremental drag penalty for Flying

WANDA is only the addition of thewing.

Four test ights of this conguration

have conrmed acceptable stability

and control. Half-area forward ns

have been own and show strong

control power for acting as canards.Full-area wingtip ns have been

own in a xed conguration and

provide sucient vertical tail surface

for stable lateral modes.

Test ights have also begun explor-ing the landing mode that will bestprotect the n mechanisms. Flying

WANDA is designed for a traditionalapproach and splashdown, with aplaning hull surface to stretch thedeceleration time across a longerlanding maneuver. All four ns are

well protected using this landingtechnique, as they are on upperportions of the vehicle. However, in

moderate to heavy sea states, thisplaning landing is not possible with-out large increases in vehicle size orautopilot complexity. Therefore, twonose-down landing styles have beentested to investigate survivability ofthe canards in a load condition con-sisting primarily in the drag direction.Since the apping propulsion mode

already must be sti in the drag

direction, the ns’ survivability in the

plunge style landing has been betterthan initially expected.

Experimentation with the FlyingWANDA conguration continues.

Future ights will explore the perfor-mance envelope using the ns as ac-tive control surfaces in the air and will

continue the landing technique work. A hollow, oodable wing is underconstruction so that the swimmingphase of experimentation can beginin the LASR pool.

The Sky’s the Limit for this

SubmarineWhile the diametrically opposed re-quirements for a ying vehicle versus

a swimming submarine vehicle seem

incompatible, the Flimmer program isshowing a feasible path can be foundsomewhere in the middle. Thereare important trade-os between

enclosed air volume and structuralweight, with a particular emphasis onsurviving a splashdown in water atight speeds. However, testing has

already shown that Test Sub cruiseswell above 50 knots in the air, whiletop speed in the water is below 10knots, illustrating the ultimate benet

of a ying submarine: assuring quick-

reaction access to underwater areas.

Test Sub splashdown sequence.

By Dan EdwardsNRL Tactical Electronic Warfare Division


11/40

NRL FEATURES

WINTER 2014

Flying WANDA takes ight, moments

after departing the pneumatic launcher.

The Flimmer vehicle is an adaptation ofNRL’s WANDA for air-delivery.

Flying WANDA approaches splashdown

in the Potomac River.


12/40

NRL FEATURES

SPECTRA0

Our overarching goal is to giverobots a deep understandingof how people think. There arethree benets of this to the scien-

tic community. First, by improvingunderstanding of how people thinkand interact with the world, we arepushing the boundaries of the eld

of cognitive science. Second, we canleverage this knowledge of people tohelp robots be better at tasks theytypically are unable to perform well,such as multimodal communicationand computer vision. Third, this deepunderstanding of humans allowsrobots to better predict and under-stand a human partner’s behavior

and, ultimately, be a better and morehelpful teammate.

To accomplish this, we build pro-cess models of fundamental humancognitive skills – perception, memory,attention, spatial abilities, and theoryof mind – and then use those modelsas reasoning mechanisms on robotsand autonomous systems. Most ofour work is done using the computa-

tional cognitive architecture ACT-R/E. A cognitive architecture is a process-level theory about human cognition.It provides a rich environment for

modeling and validating dierenthypotheses about how people think. ACT-R/E provides a set of computa-tional modules, correlated with dif-ferent functional regions of the brain,that work together to explain boththe limitations of human cognition(i.e., we don’t have unlimited work-ing memory) and the strengths (i.e.,we are good at inferring connectionsbetween concepts and interactingwith the physical world):

• Declarative Module: Managesthe creation and storage of factualknowledge; selects what chunks (facts or memories) will be thoughtabout at any given time.

• Procedural Module: Manages thecreation and storage of proceduralknowledge, or chunks; selectswhat production (if-then rule) willre at any given time.

• Intentional and Imaginal

Modules: Provide support for goaoriented cognition and intermediatproblem state representations.

• Visual and Aural Modules: Enablethe architecture to see and hearperceptual elements in the model’sworld.

• Confgural and Manipulative

Modules: Enable the architectureto spatially represent perceptualelements in the model’s world.

• Temporal Module: Allows thearchitecture to keep track of time;

acts as a noisy metronome.

• Motor and Vocal Modules: Pro-vide functionality for the model tomove and speak with an appropri-ate time course.

Each module, save the proceduralmodule, is associated with a limited-capacity buer, representing what the

model is thinking about / planning to

Scientists from the Naval Research Laboratory’s Navy Center for Applied Research in Articial Intelligence demonstrate advances in

cognitive science, cognitive robotics, and human–robot interaction with the help of teammate Octavia (center). Left to right: Ed LawsoGreg Trafton, Laura Hiatt, Sunny Khemlani, Bill Adams, Priya Narayanan, Frank Tamborello, Tony Harrison, Magda Bugajska.


13/40

NRL FEATURES

WINTER 2014

say / looking at / etc. Together, thecontents of these buers make up

working memory in ACT-R/E. Whileeach of the modules and buers is

theoretically motivated and validatedon its own, ACT-R/E’s strength liesin the complex interaction of these

components, shown below.

To illustrate, consider an exampleof a conference attendee meetingsomeone at the conference and thenattempting to remember his or hername later in the evening. The at-tendee would need to not only attendto the new person’s face, voice, andclothing, but also bind those indi-vidual concepts to the name of theindividual. The attendee would thenneed to rehearse the name several

times so it would not be forgotten.When the attendee saw the new per-son a bit later, he or she would needto take available cues (perhaps onlyvisual cues like face and clothing)and attempt to retrieve from memorythe name associated with thosecues. Priming from contextual cues,like the face, clothing, and the partyitself, would boost the activation ofthe memory, and the earlier rehearsal

would allow the memory to becomefamiliar enough to be remembered.

ACT-R/E’s delity to the way the

human mind integrates, stores, andretrieves all this information is whatprovides its true predictive and ex-

planatory power. In this case, model-

ing the human’s experiences at theconference provides both descrip-tive information about how a humancould err in this situation (e.g., peoplethat look similar, or that were met atsimilar times or similar parties, maybe easily confused) and predictiveinformation (e.g., someone’s nameis unlikely to be remembered after along break without further rehearsalor strong contextual cues).

In this article, we present two dier-ent systems in which ACT-R/E’s del-ity to human cognition is leveragedto develop intelligent systems thatcan more functionally assist a humanteammate. In the rst example, ACT-

R/E’s familiarity and context mecha-nisms help an autonomous systembetter perceive ambiguous or ob-fuscated objects that a robot might

see in the real-world environmentsour warghters encounter (such as

jungles, deserts). In the second, ACT-R/E’s goal structure, together with itsfamiliarity and context mechanisms,allow an intelligent system to helphumans avoid certain classes of er-rors, such as those commonly madein Navy vehicle maintenance proce-dures.

Computer Vision

Recent work in computer vision,performed by Hoiem, Efros, and Her-bert in 2006, and Oliva and Torralbain 2007, among others, has shownthat including contextual informationcan greatly aect the eciency of

object recognition in terms of both

the speed and the accuracy of theprocessing process. These existingcomputer vision approaches, howev-er, typically rely on static, aggregatedstatistics of various features in thescene to provide their context. Whilesuch approaches are promising, thislimitation renders them unable toperform competitively with humanperception; a richer, more human-likerepresentation of contextual infer-ence, learned over time, like that in

ACT-R/E, may be the key to their

success.

Context in ACT-R/E takes the formof associations between relatedconcepts that are learned over time.Concepts become associated whenthey are thought about at roughly thesame time; the more they are thoughtabout in proximity to each other, thestronger their association becomes.This type of context is rich, in that itcan capture nuanced relationshipsbetween concepts and facts that

are not explicitly linked; this contextis also dynamic, in that it is learnedonline and adjusted over time basedon the model’s experience. Objectrecognition algorithms, like LVis, abiologically plausible object recogni-tion system developed by RandallO’Reilly at the University of ColoradoBoulder, can utilize this information toimprove recognition in cases whereusing visual features alone is dicult.

The ACT-R/E architecture. The complex interaction of ACT-R/E’s many

theoretically motivated and validated components provides its true predictive and

explanatory power.


14/40

NRL FEATURES

SPECTRA2

One of our robots, Octavia, correctly recognizes an orange ball when it is in context.

Octavia is a humanoid robot we use to investigate how to build intelligent systems

that can serve as better teammates to their human partners.

Sample objects from the RGB-D

dataset developed by Lai et al. at

the University of Washington. The

objects in the dataset are relatively

low resolution, similar to what a

robot would see as it moves aroundthe world. Here, a battery (left) and

dry eraser (right) can potentially

have very similar visual outlines and

contours, but are typically found in

dierent situations; context can help

dierentiate between these two object

classes.

For example, when looking in thekitchen, context may suggest relatedconcepts such as oranges or lemons.

Any ambiguities that might arise fromother similar objects (such as an or-ange ball) can be quickly resolved byincorporating contextual information,

resulting in the correct identication.

Our initial experiments using a largedatabase of objects have shown thata system that combines context andLVis’s object recognition algorithms isable to increase recognition accuracyas compared to using LVis withoutcontext (see below). While theseresults are simple, they shed light onthe powerful tool that context can be,and demonstrate how it can be usedto build to autonomous systems that

are better able to support and extendthe Navy’s capabilities.

Error Prediction

With the rapid rise of communica-tion technologies that keep peopleaccessible at all times, issues of

interruptions and multitasking havebecome mainstream concerns. Forexample, the New York Times in 2005and Time magazine in 2006 bothreported stories about interruptionsand multitasking, and how they aect

performance by increasing human er-ror. In 2005, the information technol-ogy research rm Basex estimated

the economic impact of interruptionsto be around $588 billion a yeardue to losses from increased task

completion time or errors. Given theprevalence of interruptions, build-ing systems that can help remind anindividual what they were doing orwhere they were in a task can have alarge impact on individual and groupproductivity.

We built an ACT-R/E model thatemulates the process people gothrough as they get interruptedduring a procedural task and then

have to resume it. As ACT-R/E pro-gresses through a task, it maintains arepresentation of where it is currentlyengaged in the task. For familiartasks, this representation always as-sociates, via context, an action to thenext action to be performed. When atask is interrupted, the model losestrack of its goal and, upon resump-tion of the task, must remember whatit was working on. Sometimes, con-textual cues associate to future steps


15/40

NRL FEATURES

WINTER 2014

beyond the next correct one, andthe system skips a step. Alternately,sometimes the familiarity of past ac-tions surpasses the familiarity of thecurrent action, and ACT-R/E remem-bers an action prior to the last one itcompleted, and repeats a step.

Using ACT-R/E in this way, we canexplain (and thus can better predict)how, when interrupted, people tendto skip or repeat steps, even in afamiliar task, based on the task’scontext and strengthening. This ac-complishment will help us developintelligent systems that can mitigatehuman error risks in dangerousprocedures, with both monetary andfunctional benets.

Conclusion

Our approach to intelligent systemsis multidimensional. First, in thecognitive science tradition, we attaina deep understanding of how peoplethink: our ACT-R/E models faithfullycapture people’s behavior as theyperceive, think about, and act on theworld around them. Second, we usethis understanding to make intelligentsystems better by taking advantageof people’s strengths. Finally, the

models help to reveal limitations andpotential failings of human cogni-tion, which our intelligent systemscan then take steps to correct.Overall, our eorts to attain a deep

understanding of human cognitivestrengths and limitations allow usto build more functional intelligentsystems that are better able to servetheir human teammates.

Navy Center for

Appli ed Research i n

Artificial Intelligence

By Laura M. Hiatt, Frank P. Tamborello, II,

Wallace E. Lawson, and J. Gregory TraftonNavy Center for Applied Research in

Articial Intelligence

The Navy Center for Applied Research

in Articial Intelligence (NCARAI) hasbeen involved in basic and applied research inarticial intelligence, human factors, and human-

centered computing since its inception in 1981. It

is part of the Information Technology Division ofthe Naval Research Laboratory and is directed by

Alan Schultz, who is also director of the Labora-tory for Autonomous Systems Research.

The Adaptive Systems group conducts research

focused on techniques that allow systems to change

their level of autonomy on the y, adapt to changes in

their environment and in their own capabilities, learn new

behaviors through interaction with the world, and interact

more naturally with humans.

The Intelligent Systems group performs research in

cognitive science, cognitive robotics and human-robot

interaction, predicting and preventing procedural errors,

the cognition of complex visualizations, interruptions and

resumptions, and spatial cognition, to enable intelligent

systems that can work more eectively with people.

The Interactive Systems group develops and enhances

computer interfaces for autonomous and intelligent

systems, spanning human-computer and human-robot

interactions. Research includes linguistic, auditory, and

phonological analyses to link natural language to modes

of human-machine interaction.

The Perceptual Systems group examines issues related

to sensors required to support autonomous platform navi-

gation, scene interpretation, and teamwork. Techniques

include passive monocular vision, and passive and active

stereo and triocular ranging methods, along with algo-

rithms to rapidly interpret the scene.

http://www.nrl.navy.mil/itd/aic/


16/40

NRL FEATURES

SPECTRA4

The U.S. Naval Research Labo-

ratory (NRL) Laboratory forAutonomous Systems Research(LASR), a partner in the Navy’sDamage Control for the 21stCentury project (DC-21), re-

cently hosted robotics researchteams from the Virginia Poly-technic Institute and State

University (Virginia Tech) andthe University of Pennsylvania(Penn) to demonstrate themost current developments

of advanced autonomoussystems to assist in discovery,control, and damage controlof incipient res.

Fighting res can at times prove chal-

lenging to even the most seasoned

reghting veteran — a reghter must

deal with extreme unpredictability, high

temperatures, and rapid decline of

environmental and structural integri-

ties. Add to this scenario a cloistered

platform, say many levels down inside

a seagoing ship, and the challenge is

exponentially increased, resulting in

extreme risks to human life. Despite

these risks, a shipboard re must be

contained and extinguished for the

safety of the crew and continued mis-

sion readiness of the ship.

To mitigate these risks, NRL research-

ers at LASR and NRL’s Navy Center

for Applied Research in Articial Intel-

ligence (NCARAI), under direction and

funding from the Oce of Naval Re-search (ONR), are working with univer-

sity researchers to develop advanced

reghting technologies for shipboard

res using humanoid robots, an eort

led by the NRL Chemistry Division.

“As part of the Navy’s ‘leap ahead’

initiative, this research focuses on the

integration of spatial orientation and

the shipboard mobility capabilities

of future shipboard robots,” said Dr.

Thomas McKenna, managing program

ocer of ONR’s Computational Neuro-

science and Biorobotics programs. “Th

goal of this research is to develop the

mutual interaction between a humanoid

robotic reghter and the rest of the

reghting team.”

This highly specialized research, to pro

mote advanced reghting techniques,

includes development of a novel roboti

platform and re-hardened materials

(Virginia Tech), algorithms for percep-

tion and navigation autonomy (Penn),

human–robot interaction technology,and computational cognitive models

that will allow the robotic reghter to

work shoulder-to-shoulder and interact

naturally with naval reghters (NCARA

“These advancements complement

highly specialized NRL research that

focuses specically on the human–robo

interaction technology and shipboard-

based spatial interrogation technology,

said Alan C. Schultz, director of LASR


17/40WINTER 2014

NRL FEATURES

and the NCARAI. “Developments

made from this research will allow a

Navy reghter to interact peer-to-peer,

shoulder-to-shoulder with a humanoid

robotic reghter.”

The NRL LASR, where the articialintelligence portion of the research is

performed, hosted the consortium of

university researchers to demonstrate

their most current developments.

The LASR facility allows the research-

ers from Virginia Tech and Penn to

demonstrate, in a controlled environ-

ment, progress in the critical steps

necessary for shipboard re suppres-

sion using variants of their Shipboard

Autonomous Fireghting Robot,

or SAFFiR. In 2013, human–robotinteraction technology and cognitive

models developed by NRL were also

demonstrated at the laboratory.

“The LASR facility, with its unique

simulated multi-environments and

state-of-the-art labs allows us to ‘test

out’ our ideas before we go to the

eld,” Schultz said. “In essence, our

facility gives us a cost-saving method

for testing concepts and ideas before

we go to the expense of eld trials.”

While at LASR, the researchers dem-

onstrated the complex motion, agility,

and walking algorithms of the robots

over natural and manmade terrain and

simulated shipboard sea state (pitch

and roll) conditions. Also demonstrated

were “seek-and-nd” algorithms for

locating a re emergency, in this case

an open ame, and the use of “articial

muscle” for the lifting and activation

of re suppression equipment, such

as opening a water valve, lifting andwalking with a re hose, and activating

a nozzle.

“SAFFiR is being designed to move

autonomously throughout a ship to

learn ship layout, interact with people,

patrol for structural anomalies, and

handle many of the dangerous reght-

ing tasks that are normally performed

by humans,” McKenna said. The robot

is designed with enhanced multimodalSAFFiR — the Shipboard Autonomous Fireghting Robot.


18/40

NRL FEATURES

SPECTRA6

Researchers demonstrated the complex

motion, agility, and walking algorithms of the

robots over natural and manmade terrain

and simulated shipboard sea state (pitch

and roll) conditions. Also demonstrated were

“seek-and-nd” algorithms for locating a reemergency, in this case an open ame, and

the use of “articial muscle” for the lifting and

activation of re suppression equipment, such

as opening a water valve, lifting and walking

with a re hose, and activating a nozzle.


19/40WINTER 2014

NRL FEATURES

sensor technology for advanced navigation and

a sensor suite that includes a camera, gas sen-

sor, and stereo infrared (IR) and ultraviolet (UV)

cameras to enable it to see through smoke and

detect sources of excess heat. SAFFiR is also

capable of walking in all directions, balancing in

sea state conditions, and traversing obstacles

such as “knee-knocker” bulkhead openings.

“Today’s display demonstrates the integration of

perception through multiple sensors, and of lo-

comotion through biped walking,” said Dr. Daniel

Lee, director of the General Robotics Automa-

tion, Sensing, Perception Lab and professor at

the University of Pennsylvania. Tasks as humans

we take for granted, such as standing and

remaining upright, become increasingly complex

with the addition of full body mobility required for

walking and lifting. Dr. Brian Lattimer, associate

professor at Virginia Tech’s Department of Me-

chanical Engineering, additionally commentedthat what we are now seeing is the result of a

multidisciplinary project combined to perform all

the critical tasks necessary for re suppression

by a humanoid robot.

“In dark or smoke-occluded and noisy environ-

ments found in shipboard reghting conditions,

tactile feedback — touch — is an important form

of communication between human reghters,”

said John Farley, project ocer of the re test

ship ex-USS Shadwell , NRL Chemistry Division.

“Moving forward, the team will integrate NRL’s

human–robot interaction technology with the

SAFFiR platform so that there is a greater focus

on natural interaction with naval reghters.”

In the short term, however, to protect

robotic mechanisms and electronics from

intense heat, researchers in the Advanced

Materials Section of the NRL Chemistry Division

have developed a class of lightweight, high-

temperature polyetheretherketone (PEEK)-like

phthalonitrile resin that can be molded to any

shape and remain strong at temperatures up to

500 degrees Celsius. The robotic teams are ex-pecting to soon conduct shipboard trials aboard

Shadwell , the Navy’s only full-scale re test ship,

moored in Mobile, Alabama.

SAFFiR is being designed to move autonomously

throughout a ship to learn ship layout, interact with

people, patrol for structural anomalies, and handle

many of the dangerous reghting tasks that are

normally performed by humans.

SAFFiR taking a break after a hard day at work.

By Daniel ParryNRL Public Affairs Ofce


20/40

NRL Technology Assists FDNY David DeRieux (right) of the Naval Research L

RFID tracking system grew out of the realization during the events of 9/11 that rst responders did no

FDNY vehicles and helps to answer, Who exactly is on scene? Where are they? Are they safe? NRL re

saving work. See details on page 35. (Photo, left to right: George Arthur, Dan Orbach, and David DeR


21/40

nvented a system for Fire Department New York (FDNY) to automatically track reghters. NRL’s active-

liable method to account for all members responding to an incident. The system is now in use on 15

014 Federal Laboratory Consortium Award for Excellence in Technology Transfer for this potentially life-


22/40

NRL FEATURES

SPECTRA0

A benthic microbial fuel cell

(BMFC) is an oceanographic

power supply that can gen-

erate power indenitely. TheBMFC sits at the sediment/water(benthic) interface of marine environ-ments, where it generates electricalpower using organic matter naturallyresiding in marine sediments as itsfuel, and oxygen in overlying wateras its oxidant. At the Naval Research

Laboratory (NRL), we are developingBMFCs to power persistent, in-waterintelligence, surveillance, and recon-naissance capabilities presently lim-ited in operational lifetime by batterydepletion.

Thousands of battery-powered sen-sors are deployed each year thatprovide valuable scientic informa-tion about marine environments. The

prospect of using BMFCs to powerthese sensors indenitely, or at least

far longer than possible with batteries,is enticing – we can acquire long-termuninterrupted data and signicantly

reduce the cost and logistics burdenof keeping sensors running. A BMFCis maintenance free and nondeplet-ing: the organic matter and oxygenare constantly replenished by natu-rally occurring diusion and advec-tion; the electrode catalysts consist

of self-forming biolms comprised ofmicroorganisms naturally inhabitingthe benthic interface; and there areno moving, degradable, or depletablecomponents.

NRL has been a leader in developingthe BMFC and optimizing it for Navyoperational use. We have conductedlaboratory and eld research to study

the microbial activity underlying

power generation: to determine theeects of dierent oceanographic

and biogeochemical parameters, topair prototype BMFCs with sensors,and to develop easily deployablecongurations. We have successfully

powered many useful devices withsmall-scale BMFCs (less than 0.1watt continuous output), and have

just begun development of full-scaleBMFCs (greater than 1 watt continu-ous output).

Early Laboratory Experiments

The bottom sediment of manymarine environments is reductantenriched, due to metabolic activityof sediment-dwelling microorgan-isms, whereas the overlying wateris oxidant enriched due to supply ofoxygen from the atmosphere. Thistransition is referred to as the benthicredox gradient. As a result of this


23/40

NRL FEATURES

WINTER 2014

gradient, when an inert electrode isstepwise inserted into such sedi-ments from overlying water, its opencircuit potential shifts by as much as−0.8 volts.

Taking advantage of this natural volt-age gradient at the sediment/waterinterface, my colleagues and I setout to create a battery for poweringoceanographic sensors by embed-

ding one electrode into reductant-enriched marine sediment, where itwould act as an anode, and plac-ing the other in oxidant-enrichedoverlying water, where it would actas a cathode. Our rst experiments

involved a benchtop shoebox-sized aquarium containing marinesediment, seawater, and a platinumanode and platinum cathode, whichgenerated miniscule current across aresistive load. The notion at the timewas that the anode was oxidizing

microbial-generated reductants in thesediment (such as sulde), and the

cathode was reducing oxygen in theoverlying water. Since the net reac-tion is thermodynamically favorable(electron transfer from a reductantto an oxidant, just like in a battery orfuel cell), it was reasonable to ex-pect that power could be expendedacross a resistive load connectingthe electrodes as long as oxygen was

replenished at the cathode, and masstransport (assumed to be diusion)

supplied the anode reactants andremoved the anode products. A mostinteresting result was that currentincreased over time to a steady-statelevel. In most electrochemical experi-ments, current decreases over timedue to depletion of the reactant and/ or diminished catalytic activity of theelectrode. I remember being dumb-

founded watching the current rise.

We immediately followed with anotherbenchtop experiment using graphiteelectrodes for the anode and cathode,resulting in a signicantly higher cur-rent density. After the system gener-ated power for some time across aresistor, we removed the anode fromthe sediment and examined it to nd

a biolm adhering to its surface — a

coating comprised of a community ofmicroorganisms from the sediment. A

genetics-based analysis of the biolmindicated that it was enriched in aspecic class of microorganisms, Del-

taproteobacteria. The microorganismavailable in pure culture most similarto the microorganism enriched on theBMFC anode was Desulfuromonas

acetoxidans, a dissimilatory metal-reducing bacteria (DMRB) found inmarine sediment that couples oxi-dation of sedimentary acetate with

reduction of insoluble iron oxidemineral deposits. Such microorgan-isms are fascinating (especially toan electrochemist like me) for theirability to transport respired elec-trons from inside the cell to insolubleoxidants residing outside the cell(referred to as extracellular electrontransport), acquiring a small amountof energy for themselves in the pro-cess. It is thought that D. acetoxidans

in a BMFC anode biolm oxidizesorganic matter in marine sedimentand transports the acquired electronsto the underlying anode as if it were amineral deposit, eectively catalyzing

the anode reaction.

Early Field Deployments

In one of our rst eld experiments,

at the Rutgers University Marine FieldStation in the Great Bay, New Jer-sey in 2002, a BMFC with graphiteelectrodes generated power continu-

ously for nine months without indica-tion of depletion in power before theexperiment was terminated. This andsubsequent longer-term eld and

laboratory experiments revealed animportant attribute of microbial an-ode catalysts: as living entities meta-bolically beneting from the reaction

they catalyze, they self-maintainthemselves and do not degrade overtime as long as their environment

The combined power from six small-scale BMFCs was used to persistently operate a

meteorological buoy (center) in the Potomac River. The buoy measured local weather

conditions and transmitted data every ve minutes to a land-based receiver.


24/40

NRL FEATURES

SPECTRA2

is hospitable. In this way, they canmaintain catalytic activity indenitely.

Analysis of the anode biolm of the

New Jersey BMFC indicated enrich-ment not only of D. acetoxidans, butalso of microorganisms in the Desul-fobulbus or Desulfocapsa genera that

oxidize and disproportionate sulfur.In sulde-enriched sediment such as

the New Jersey site, a graphite an-ode can develop a passivating sulfurprecipitate on its surface that canshut down current. It is thought thatthese microorganisms act to clear thesulfur from the anode surface, form-ing sulfate and sulte, both soluble.

The next BMFC eld experiments

were performed in 2003, at 1000meters depth in the Monterey Can-yon o the coast of California on a

cold seep, a ssure on the seaoor

eusing organic-rich water. We had

hypothesized that the high masstransport rate of organic matterwould benet BMFC power gen-eration and we were correct. ThisBMFC, equipped with a spear-likegraphite anode inserted verticallyinto the seep, generated signicantly

more power per unit footprint areaand geometric surface area of the

anode compared to our earlier ex-periments. This was exciting becauseit demonstrated that placement of aBMFC in an area with a high rate oforganic matter mass transport, suchas a subliming methane hydrate out-crop, could result in very high poweroutput, which could be of great useto the Navy.

We next deployed a BMFC in 2004 inthe Potomac River o the NRL pier.

Here, the BMFC consisted of graph-

ite electrode slabs attached with zipties to the top and bottom of milkcrate spacers that were positionedon the river bottom with one elec-trode in the mud and the other in theoverlying water. An attached buoyoating on the water surface mea-sured air temperature, pressure, rela-tive humidity, and water temperature,and transmitted this data to my oce

every ve minutes. The entire buoy

including the radio transmitter was

powered by the BMFC. This requireda novel solution to convert the lowDC voltage output of the BMFC from0.35 volts (dictated by the benthicredox gradient) to 6 volts to charge acapacitor, which in turn was used topower the buoy. The capacitor was

needed to buer the low but steadypower output of the BMFC with theduty cycle of the buoy, wherebynearly 99% of energy consumption bythe buoy occurred during the fractionof a second during which the datawas transmitted. Between trans-missions, the BMFC recharged thecapacitor. This was the rst time that

a BMFC was used as a free-standingpower supply. The buoy ran awlessly

for seven months. The river froze thatwinter, trapping the buoy in ice, but

it kept running until the ice thawedenough that large pieces began toow down river, dragging the buoy

and severing the mooring line con-necting it with the BMFC. I got a callfrom the Coast Guard who found thebuoy resting against a pier of the Wil-son Bridge. (Lesson: always put yourphone number on any oceanographicinstrument you want back.)

A Practical Power Supply

Since 2004, a number of researchershave been working to develop BM-FCs into practical power supplies thatare straightforward to deploy. I amfortunate to pursue this concept withmy NRL colleagues Dr. Je Book, Mr.

Andy Quaid, and Mr. Justin Broders-en, oceanographers; Dr. Yoko Furuka-wa, marine geochemist; and Dr. Je

Erickson and Mr. Marius Pruessner,engineers; with funding from the Of-ce of Naval Research and NRL. We

have deployed small-scale BMFCs in

coastal locations including California,Gulf of Mexico, Maine, New Jersey,North Carolina, and the Adriatic Sea.These BMFCs have operated overdurations ranging from less than sixmonths to more than two years with-out indication of depletion in poweroutput. They have powered a hydro-phone with a radio transceiver link, anacoustic modem, and a surveillancecamera with a cellular link.

We are working to optimize the BMFCdesign to maximize power outputwhile making anode embedment asimple and reliable procedure. Maxi-mizing power involves anode designsthat expose as much mass-transport-accessible surface area to sediment

as possible within mass and volumeconstraints of the intended deploy-ment platform, all while being easyto properly embed into sediment. Wehave been making steady progress,hampered at times by the vagaries ofeld research: in 2012, all our mea-surement equipment and 26 BMFCsdeployed at a New Jersey site werelost when directly hit by HurricaneSandy.

In 2013, we began development of

full-scale BMFCs in NRL’s newlyopened Laboratory for AutonomousSystems Research (LASR), where wehave assembled a six-meter-diameter,8000-gallon benthic mesocosm,essentially a large indoor aquariumlled with sediment and seawater to

replicate the benthic interface. This fa-cility enables full-scale BMFC testingin a controlled environment, greatlyreducing the cost and risk of BMFCdesign development compared to eld

testing. We have already made sevendeployments of a two-meter-diameter900-kilogram oceanographic mooringequipped with various BMFCs to testnew designs. The LASR mesocosm isgreatly compressing the BMFC devel-opment horizon and I feel we are ontrack to transition the BMFC in 2015.

By Leonard TenderNRL Center for Bio/Molecular Science and

Engineering


25/40

NRL FEATURES

WINTER 2014

Dr. Leonard Tender (above, and with MariusPruessner at right) working in the benthic

mesocosm in the Laboratory for Autonomous

Systems Research. The mesocosm contains 18

inches of seawater and 24 inches of synthetic

marine sediment that has been inoculated with

real marine sediment to provide necessary

microbial activity. On the water surface are BMFC

graphite bottlebrush cathodes; when deployed in

the eld, these cathodes are beneath the water

surface but above the sediment surface. The

vertical tubes indicate the locations of BMFC

anodes embedded in the sediment.

BMFC graphite bottlebrush cathodes.

The benthic mesocosm in the Laboratory for

Autonomous Systems Research.


26/40

NRL FEATURES

SPECTRA4

The Naval Research Labora-

tory (NRL) has been leading

the development of lightweight

hydrogen fuel cell systems for

long-endurance unmanned air

vehicles (UAVs), starting with

the modest 3-hour ight of the

Spider Lion in 2004, followed

by the 24-hour ight of the Ion

Tiger in 2009, the 48-hour ightof the Ion Tiger in 2013, and the

deployment of the XFC tactical

UAV from the torpedo tube of a

submarine in 2013. NRL also as-

sisted in a demonstration ight

of Insitu’s ScanEagle vehicle on

a hydrogen fuel cell in 2010.

In a recent project, an NRL teamleveraged 3-D printing and otherprototyping tools in NRL’s Laboratoryfor Autonomous Systems Research(LASR) for developing and integrat-ing their next generation of hydrogenfuel cells into unmanned systems.

Additive manufacturing processessuch as 3-D printing promise rapidprototyping of new designs so theycan be tested and veried without the

expense of traditional manufacturing.

The team used new tools in the LASR,plus commercial tools, to investigatehow 3-D printing can be used towardimproving fuel cells for unmannedsystems.

The motivation for NRL’s research intohydrogen fuel cells is clear. Fuel cellsdirectly convert the energy in hydro-gen to electricity by electrochemicallycombining with the oxygen in air.

The only byproducts are water andlow-grade heat, and no combustionoccurs. The electrochemical reactionsare ecient, resulting in systems that

are over 50% ecient and operate at

low temperatures (near 80 degreesCelcius). Combined with the highenergy of hydrogen, the result is anelectric power source that is light-weight and has high energy per unitweight. For the Ion Tiger system, thismeant a sevenfold increase in endur-

ance over battery propulsion whengaseous hydrogen was used for the24-hour ight. Hydrogen fuel cell sys-tems also respond rapidly to changesin throttle, and are easily turned o

and on. The fast response time andhigh energy density are two reasonsmajor automobile manufacturers suchas General Motors, Toyota, Honda,and Daimler have invested cumula-tively $10 billion to $12 billion in hy-

3-D Printing in

Hydrogen Fuel Cells for Unmanned Systems

The NRL crew after the successful ight of a 3-D printed fuel cell aboard the Ion Tiger UAV. Left to right: Dan Edwards,

Drew Rodgers, Ben Gould, Steve Carruthers, Mike Schuette, and Karen Swider-Lyons. Not shown: Chris Bovais.


27/40

NRL FEATURES

WINTER 2014

drogen fuel cells for next-generationautomobiles.

Despite integration into many suc-cessful vehicles, hydrogen fuel cellsare still too costly for commercialuse, coming in about 5 to 10 timesmore expensive than a battery sys-tem. Platinum electrocatalysts wereonce a signicant cost factor, but

eorts by the automotive industry

have cut the precious metal content

of fuel cell systems drastically. Thebigger cost now is in the balance of

plant used to operate the fuel cell:this includes air blowers to push airthrough the system, a humidier to

keep the fuel cell at the right humid-ity, hydrogen recirculation systemsto make sure 99% of the hydrogenfuel is used, and power electronics toconvert the voltage of the fuel cell toone usable by the air blower, autopi-lot, and payloads. Another major costdriver is the bipolar plates that are

used as the backbone of the fuel cellto physically hold together catalystlayers and to provide channels for air,fuel, and coolant ow, and electric

pathways for the power produced bythe system. These plates are whereNRL focused much of its most recentresearch.

The bipolar plates hold a lot of thescience and engineering of the fuel

cells. The air, fuel, and coolant ow

elds are typically created through

iterations of design, computationaluid design (CFD) modeling, fabrica-tion, and testing. Traditionally, bipolarplates have been made of carbon, butthese are being replaced in favor ofmetal bipolar plates made of stampedfoils that are coated for corrosion-proong against the acidic environ-ment in the fuel cell and then weldedtogether. Looking at the complexity

of developing new metal plate de-signs for NRL’s UAV systems, seniormechanical engineer Mike Schuette

said, “Stamping of metal foils is a

proven industrial practice, but I wasconcerned about how long it wouldtake to design the fuel cell ow elds,

develop custom tools to stamp the

plates, and develop the welding pro-cesses. So I thought this was a greatopportunity to try 3-D printing of metaplates so we could quickly try out sev-eral designs without having to committo any tooling or welding.”

First, the NRL team came up withseveral ow eld designs. To check

their viability, the team used the 3-Dprinting tools in the LASR to build upprototype plastic ow elds for ow

visualization. In parallel, physicist Dr.Ravi Ramamurti carried out CFD pre-dictions of the gas and coolant ows

and heat rejection. A successful platedesign was achieved through severalrounds of design, building, testing,and model verication.

The challenge now was to make ahollow metal bipolar plate with exter-nal channels for air and fuel and aninternal channel for coolant in a singlethree-dimensional structure that was

also thin enough to meet the stringentlow-weight requirements for ight. The

team investigated dierent 3-D metal

printing processes and found that di-rect metal laser sintering, DMLS, hadthe most potential for fabrication ofthis kind of geometry. NRL success-fully worked with the DMLS company3T RPD Ltd. (UK) to make thin tita-nium bipolar plates from NRL designs

Schematic of a hydrogen fuel cell. Pressurized hydrogen (H2) is

distributed through ow elds in the bipolar plate to the anode

catalyst where it is converted to protons (H+) and electrons. The

electrons pass from the anode bipolar plate to the cathode bipolar

plate, while the protons travel through the polymer electrolyte

(peruorosulfonic acid) to the cathode. At the cathode, the protons

and electrons combine with oxygen from air to form water and heat.

Concept drawing for a bipolar plate,

showing air, coolant, and fuel channels.


28/40

NRL FEATURES

SPECTRA6

The next challenge was to compressseveral bipolar plates together in afuel cell stack with no leaks. “The

no-leak part is really tough,” says

mechanical engineer (Joseph) DrewRodgers. “You have to put all the

catalyst layers together with the gasdiusion media, and properly align

the gaskets so that none of the gaschannels are blocked, and then com-press them together in a gas-tightblock.” A consultant to the team,

Doug Wheeler (who had worked atInternational Fuel Cells on fuel cellsfor the space shuttle), pointed us to

high-performance softgood materi-als, including gasket materials andcommercial catalyst membranes,which we precision-cut with NRL’s

laser cutting tools.

Once the full stack was together,chemical engineer Dr. BenjaminGould found that it was much moreelectrically resistive than he hadexpected. “We looked through the

literature and found that the nish

on the bipolar plates strongly aects

the plate-to-plate conductivity.” Ben

then worked with Morris Technolo-

gies (Cincinnati, OH) to develop theappropriate surface roughness forthe plates, and aeronautical engineerChris Netwall joined the team to work

on processes to compress the platesand softgoods together eectively

to eliminate as much resistance aspossible. Additionally, NRL workedwith TreadStone Technologies Inc.(Princeton, NJ) to apply a coatingof titanium oxide impregnated withnanoscale dots of gold to providea highly conductive, robust coat-ing to the titanium bipolar plates.With these improvements, the team

Details of the hollow titanium bipolar plates made by the 3-D printing process of direct metal laser sintering (DMLS).

The bipolar plates assembled into a gas-tight stack.


29/40

NRL FEATURES

WINTER 2014

put together 41 bipolar plates for aworking stack. With help from theUniversity of Hawaii, quality assuranceprocesses were developed to conrm

the fuel cell performance.

Putting the plates together in a stack

exposed a critical problem with the3-D metal printing process: warp-ing. In DMLS, titanium particles aresintered into place with a laser, andthen the whole body is sintered. Clas-sic metallurgy teaches that residualstrains in the metal-to-metal contactpoints lead to deformations as thematerial undergoes strain relief. Therst set of metal bipolar plates had

dimensions of 4 centimeters (cm) by8 cm and could be pressed togethereectively with other plates to make

good contact; however, an attempt tomake larger 8 cm by 12 cm plates wasunsuccessful, as they all came in fromthe manufacturer warped by severalthousandths of an inch. NRL is stillinvestigating how to eliminate warpsso that the plates have an adequateatness tolerance. Developing ac-curate tolerancing is likely to be a keyaspect of future research in metal 3-Dprinting.

With the stack ready, we built somenew balance of plant and electroniccontrol components, and soon acomplete fuel cell system was runningon the bench in the LASR Power andEnergy Lab, ready to be packagedinto NRL’s Ion Tiger test vehicle.

The fuel cell system appeared readyto go from bench testing, but ight

tests are the true graduation exercisefor UAV technology. After an unusu-ally snowy winter, the team jumpedat the chance to y the new system

on a warm day in mid-March. Theteam rolled into the ight range at

Blossom Point, Maryland, we startedup the fuel cell, Chris Bovais loadedthe Ion Tiger onto the winch — ando it went. Pilot Steve Carruthers let

the vehicle climb above the tree lineand then ew some laps. A thumbs-

up from Dan Edwards on the groundstation controller. Everything worked.NRL had successfully built a fuel cellusing 3-D metal printing.

The team showed that 3-D printingwas denitely the right way to go

toward prototyping a fuel cell concept.“The 3-D printing helped us sort out

the ow eld designs and verify Ravi’s

CFD results, so now we are ready tocommit to the more time-intensiveprocess of developing tooling andwelding for stamped titanium-foil bi-polar plates,” says Ben. NRL has used

this experience to publish and sharenew information on designing bipolar

plates and stacks, decreasing con-tact resistance in fuel cell stacks, andusing CFD to design eective coolant

ow elds. The resulting computa-tional tools will allow NRL to rapidlydevelop future fuel cell systems in the100 to 5000 watt range for air, under-sea, and ground vehicles.

Michele Anderson, the NRL team’sprogram manager from the Oce of

Naval Research, summarized the val-ue of the eort: “3-D printing met its

promise as an R&D tool to make andtest complex assemblies for researchprototypes without the investment ofcostly tooling. Clearly it is an impor-tant research tool for the Navy for rap-idly demonstrating new energy systemprototype technology, and identifyingoptimal designs prior to investment inmore expensive manufacturing meth-ods.” With additional research and

development, 3-D printing combinedwith advanced computer design tools

will help to accelerate the demonstra-tion and elding of new Department ofDefense capabilities. As NRL’s LASRfacility expands its 3-D printing tools,it will continue to be the home forrapid prototype development at NRL.

Mike Schuette, Ben Gould, Drew Rodgers, and

Karen Swider-Lyons display a hydrogen fuel

cell stack made with 3-D printed bipolar plates

(right), a microcontroller (left), and the nose of

the Ion Tiger UAV (center) which houses the

fuel cell system during ight.

By Karen Swider-LyonsNRL Chemistry Division


30/40

SPECTRA

8

NRL FEATURES

The DC-21 project is developing advanced technologiesfor shipboard damage control. The project includes thedevelopment of advanced autonomous systems to assist indiscovery, control, and damage control of incipient res. Our

focus is on the human–robot interaction technology that willallow a Navy reghter to interact peer-to-peer, shoulder-to-

shoulder with a humanoid robotic reghter. Computational

cognitive models are used to create reasoning mechanisms for

the robot based on required human cognitive skills. In addition,the project investigates perception, including aural, gesture,and touch, and how context helps to disambiguate perceptionsHuman–robot interaction is achieved through robust parsing ofhuman speech, integrated with human gesture and touch, all ofwhich are important in reghting teams.

Damage Control for the 21st Century (DC-21)

L A B O R A T O R Y

F O R

A U T O N

OMO U S S Y

S T E M

S R

R C H

• W

A S H I N G T O N , D

. C .

•

The MANTISS program is adapting techniques inspired

by nature to enable advanced information gathering bymulti-agent autonomous teams. Under the Oce of Naval

Research Science of Autonomy (SoA) program, numerousbiologically inspired techniques have been developedfor controlling teams or swarms of autonomous vehicles.These include foraging algorithms based on how animalherds gather food, ocking behaviors observed in birds,

schooling behaviors observed in sh, and articial physics

(physicomimetics) based on how particles interact invarious states of matter. We adapt these techniques togather information (rather than food, for example), anduse them to perform area coverage tasks such as search,reconnaissance, and surveillance.

Mobile Autonomous Navy Teams for Information Surveillance and Search

LABORATORY FOR AUTONOMOUS SYSTEMS RESEARCH


31/40

WINTER 2014

NRL FEATURES

This project is developing technologies for advancedtest and evaluation of the control software forintelligent autonomous systems. Machine learningtechniques in the form of evolutionary algorithms andreinforcement learning are applied to learn the minimalnumber of faults that cause minimal or failed behaviorof an autonomous system that is under the controlof autonomy software — the autonomy software isthe system under test. The method uses high-delity

simulations of the vehicle and the environment, andtests are performed in that simulation to test therobustness of the autonomous controller under variousunanticipated conditions.

Adaptive Testing of Autonomous SystemsNRL FEATURES

In this simulation, a UUV is

performing its mission correctly.

In this simulation, the vehicle

controller has failed to deal with a

limited number of faults introduced

by the adaptive testing.

One suspect in the failure of lithium-ion batteries is the formation ofmetallic lithium dendrites that can puncture the cell’s separator andform a short circuit between the cathode and anode. NRL researchusing in situ optical microscopy has identied temperature-

dependent morphologies of metallic lithium dendrites: at 23 °C,spherical Li electrodeposits agglomerate to form a loosely packedstructure, while at −5 °C, the Li extrudes into a continuous wire

ball structure. We are examining the link between temperature,morphology, instability, and susceptibility to failure, withimplications for improving lithium-ion battery safety.

Characterization of Lithium-ion Batteries

The Robotic Touch Sensing project is developing an articial sensate

skin for robots, to extend the perceptual capabilities of roboticmanipulators to include touch. The sensate skin uses an embeddedarray of piezoresistive sensors and an algorithm that analyzes theresponses of the sensors aected by each environmental contact/

impact, to detect and classify contacts (location and magnitude).The detection algorithm uses K-means clustering, rules for contactdisambiguation, and 3-D trigonometric calculations.

Robotic Touch Sensing, Manipulation, and Fault Detection

For more research highlights, see www.nrl.navy.mil/lasr/highlights.


32/40

SPECTRA

0

NRL NEWS BRIEFS

NAVY LAUNCHES UAV FROM

SUBMERGED SUBMARINE

The U.S. Naval Research Labo-ratory (NRL) with funding fromSwampWorks at the Oce of

Naval Research (ONR) and the De-partment of Defense Rapid ReactionTechnology Oce demonstrated the

launch of an all-electric, fuel-cell-powered, unmanned aerial system(UAS) from a submerged submarine.

From concept to eet demonstration,

this idea took less than six years toproduce results at signicant cost

savings when compared to traditionalprograms often taking decades toproduce results.

“Developing disruptive technolo-gies and quickly getting them intothe hands of our sailors is what ourSwampWorks program is all about,”

said Craig A. Hughes, Acting Directorof Innovation at ONR. “This demon-stration really underpins ONR’s dedi-cation and ability to address emergingeet priorities.”

The successful submerged launchof a remotely deployed UAS oers a

pathway to providing mission-criticalintelligence, surveillance, and recon-naissance (ISR) capabilities to theU.S. Navy’s submarine force.

Operating under support of theLos Angeles class submarine USSProvidence (SSN 719) and the NavalUndersea Warfare Center–Newport

Division (NUWC-NPT), the NRL-developed XFC (eXperimental FuelCell) UAS was red from the subma-rine’s torpedo tube using a Sea Robinlaunch vehicle system. The Sea Robinlaunch system was designed to t

within an empty Tomahawk launchcanister (TLC) used for launchingTomahawk cruise missiles alreadyfamiliar to submarine sailors.

Once deployed from the TLC, the SeaRobin launch vehicle with integratedXFC rose to the ocean surface whereit appeared as a spar buoy. Uponcommand of Providence Command-ing Ocer, the XFC then vertically

launched from Sea Robin and ew

a successful several hour missiondemonstrating live video capabilitiesstreamed back to Providence, surfacesupport vessels, and Norfolk beforelanding at the Naval Sea SystemsCommand Atlantic Undersea Test and

Evaluation Center (AUTEC), Andros,Bahamas.

“This six-year eort represents the

best in collaboration of a Navy labora-

tory and industry to produce a tech-nology that meets the needs of thespecial operations community,” said

Dr. Warren Schultz, program devel-oper and manager, NRL. “The creativ-ity and resourcefulness brought to thisproject by a unique team of scientistsand engineers represents an unprec-edented paradigm shift in UAV propul-sion and launch systems.”

The NRL Chemistry and TacticalElectronic Warfare Divisions teamincludes the design-builder of the SeaRobin, Oceaneering International Inc.(Hanover, MD); the fuel cell developerProtonex Technology Corp. (South-borough, MA); and NUWC-NPT’s

Autonomous and Defensive SystemsDepartment for Temporary Alteration(TEMPALT) and test demonstrationsupport.

By Daniel Parry, NRL Public Affairs Ofce

XFC UAS launch in time-lapsed

photography. Deployed from

the submerged submarine USS

Providence, the XFC vertically

launched from a Sea Robin

launch vehicle (bottom right).

The folding wing UAS autono-

mously deploys its X-wing airfo

and after achieving a marginal

altitude, assumes horizontalight conguration.

(Photo: NAVSEA-AUTEC)

The XFC is a fully autonomous, all-

electric, fuel-cell-powered, folding wing

UAS with an endurance of greater than

six hours. The nonhybridized power plant supports the propulsion system

and payload for a ight endurance that

enables relatively low cost, low altitude,

ISR missions. The XFC UAS uses an

electrically assisted take-o system

which lifts the plane vertically out of its

container and therefore enables a very

small footprint launch such as from a

pickup truck or small surface vessel.


33/40

WINTER 2014

NRL NEWS BRIEFS

Captain Mark C. Bruington Relieves

Captain Anthony J. Ferrari

Captain Mark C. Bruington, USN, assumed theduties as the Naval Research Laboratory’s 38thCommanding Ocer during a formal Change

of Command ceremony in Washington, D.C.on Friday, August 1, 2014. Captain Bruingtonsucceeds Captain Anthony J. Ferrari, USN, whoretired following two years as Commanding Oce

of NRL, and 28 years of Naval service.

As NRL’s Commanding Ocer, Capt. Bruington

directs the activities of approximately 2500scientists, engineers, and support personnel intheir mission to conduct leading-edge researchand provide new technological capabilities to the

Navy and Marine Corps. Prior to his assumption ocommand of NRL, he was the Principal Director,Programs at the Defense Security Cooperation Agency where he led a team charged withDepartment of Defense humanitarian assistance,building partnership capacity and Foreign MilitaryTraining and Equipping U.S. partner nations.

“For exceptionally meritorious conduct in the performance of out-standing service as Commanding Ocer, Naval Research Labora-tory from August 2012 to July 2014. Captain Ferrari signicantly

increased Untied States national security by fostering the NavalResearch Laboratory’s world-class research capabilities and deliv-ery of critical advanced technologies to the Department of Defenseand other national level organizations. His eorts enabled the Naval

Research Laboratory to develop and transition critical science andtechnology programs needed to maintain overwhelming techno-logical superiority in the dynamic global operational environment.Captain Ferrari forged the way ahead, improving business practicesand methodically investing in a broad array of scientic and techno-

logical infrastructure projects such as the Marine Meteorology Cen-ter, which provides state-of-the-art research facilities for weatherforecasting for the eet. His innovative spirit, involved leadership,

and ability to surge technological solutions to short-fused, mission-critical national requirements have saved lives and contributed to aplethora of operational successes. Captain Ferrari’s superior perfor-mance of duties highlights the culmination of 28 years of honorableand dedicated service. By his dynamic direction, keen judgment,and loyal devotion to duty, Captain Ferrari reected great credit upon

himself and upheld the highest traditions of the United States NavalServices.”

Captain Anthony J. Ferrari Receives

Legion of Merit

Capt. Anthony Ferrari Capt. Mark Bruington

Chief of Naval Research Admiral Matthew L. Klunder presents the Legion

of Merit to Captain Anthony J. Ferrari.


34/40

NRL and Aerospace Industry Host 10th Annual

NRL OUTREACH

The 2014 CanSat competition, celebrating its 10-year anniversary, required teams

to develop a soda-can-size satellite-type system (CanSat) capable of harvesting

energy from the environment and transmitting telemetry in real time to a station on

the ground.

Created in 2004 by the American Astronautical Society(AAS) and American Institute of Aeronautics and As-tronautics (AIAA), the Texas CanSat Competition is an

undergraduate and graduate level design-build-launch eventsimulating the end-to-end life cycle of a complex engineeringproject.

Spanning a decade long commitment by the U.S Naval Re-search Laboratory (NRL) and other federal and commercialsponsors, the goal of CanSat is to foster student growth in

multiple disciplines of science, technology, engineering, andmathematics (STEM). The CanSat competition has becomean annual event providing a unique opportunity for univer-sity and college student teams to design, build, and launcha soda-can-size satellite-type system (CanSat) designed tomeet specic mission goals selected each year by competi-tion sponsors and co-sponsors.

This year’s 10-year anniversary CanSat competition, co-sponsored by NRL in cooperation with NASA Goddard SpaceFlight Center, NASA Jet Propulsion Laboratory, AIAA, AAS,

Ball Aerospace Technologies, Praxis Inc., and Kra-tos Integral Systems International, required teams todevelop a CanSat capable of harvesting energy fromthe environment and transmitting telemetry in real timto the team ground stations. The CanSat had to useaero-braking to slow its descent and protect a raw egduring the launch, deployment, descent, and landing.Parachutes or similar devices were not allowed.

Beginning last October and culminating in a nal com

petition in June, teams from around the nation, as weas from South America, Europe, and Asia, entered todesign and build a space-type system and then com-pete against each other at the end of two semesters tdetermine winners.

Preliminary design reviews were held in February andcritical design reviews in April. Those teams able tocompete attended the three-day launch event in Junein Burkett, Texas. The rst day was committed to

safety checks and preight briengs; on the second

2

SPECTRA


35/40

NRL OUTREACH

Provisions of the 2014 competition required deployed

CanSats to use aero-braking to slow descent and to

protect a raw egg payload during the launch, deploy-

ment, descent, and landing stages of the mission.

Parachutes or similar devices were not allowed.

Since its start in 2004, the CanSat competition

has become an annual event providing a

unique opportunity for university and college

student teams to design, build, and launch a

soda-can-size satellite-type system designed

to meet specic mission objectives. The goal

of CanSat is to foster student growth in mul-

tiple disciplines of science, technology, engi-

neering, and mathematics (STEM).

day, launch competitions were held; and on daythree, ight reviews were posted and nal awards

were announced.

Of the 54 original teams, 39 attended the launchcompetition, with rst place won by Team Arisat

from Istanbul Technical University, Turkey. Secondplace went to Team Wilkensat from Sri RamaswamyMemorial (SRM) University, India; third place wasTeam Wind Chargers from University of Alabama

Huntsville; fourth place, Team Tomahawk from Ryer-son University, Canada; and fth place was Tarleton

Aerospace Club from Tarleton State University,Stephenville, Texas.

WINTER 2014

U.S. Naval Research Laboratory engineer and

CanSat organizer Ivan Galysh was honored with the

2013 National Association of Rocketry (NAR) Howard

Galloway Spacemodeling Service Award for his contri-

butions over the last decade to educate and motivate

students in the technical disciplines of aerospace engi-

neering. As a mentor and organizer for student teams

in NAR’s Team America Rocketry Challenge, NASA’s

Student Launch Initiative, the Battle of the Rockets (of

which he is a founder), and other programs, Galysh

has spent hundreds of hours each year providing direct

hands-on instruction and motivation to middle school,

high school, and university students. In acknowledge-

ment of these outstanding accomplishments, the

NAR said, “there is no other individual in the National

Association of Rocketry who is more involved across

more programs in advancing our organizational goalof ‘paying forward’ to motivate and educate the next

generation of U.S. aerospace professionals and rocke-

teers.” Galysh has been in the NRL Space Applications

Branch since 1987, where he rst developed systems

to test the performance of atomic frequency standards

for GPS satellites, and has since worked on numerous

programs developing space-based payloads. Most

recently, these include a trio of experiments to demon-

strate miniaturized satellite technologies for improved

maritime security and measurement of space weather

and the radiation environment.


36/40

SPECTRA4

NRL MAKING CONNECTIONS

STEM FESTIVALS

C O N F E R E N C E S

SOCIAL MEDIA


37/40

WINTER 2014

NRL TECHNOLOGY TRANSFER

On 15 of its vehicles, Fire Depart-ment New York (FDNY) now can

automatically see which re-

ghters are nearby from the onboard

computer, and relay that information

to the city’s Operations Center. The

system was invented by David DeRieux

of the U.S. Naval Research Labora-

tory (NRL) Space Systems Develop-

ment Department, along with Michael

Manning of Mannin

NRL Spectra - Winter 2014

Documents

Transcript of NRL Spectra - Winter 2014