Preprint, 1998 Fall Simulation Interoperability Workshop, Sept. 13-18, 1998, Orlando, FL. A Distributed Simulation Facility to Support Human Factors Research in Advanced Air Transportation Technology Keith Amonlirdviman Todd C. Farley R. John Hansman, Jl: MIT International Center for Air Transportation Massachusetts Institute of Technology Room 33-113 77 Massachusetts Avenue Cambridge, MA 02139 617-253-2271 [email protected], [email protected], [email protected] http://web, mit.edu/aeroast ro/www/l abs/ICAT/ John F. Ladik Dana Z Sherer TASC, Inc. 55 Walkers Brook Drive Reading, MA 01867-3297 781-942-2000 [email protected], [email protected] Keywords: Human Factors, Distributed Simulation, Flight simulation, Air Traffic Control simulation ABSTRACT: A distributed real-time simulation of the civil air traffic environment developed to support human factors research i_1advanced air transportation technology is presented. The distributed environment is based on a OltStOm sinltdation architecture designed for simplici O' attd flexibility in human experiments. Standard Internet protocols are used to create the distributed environment, linking an advanced cockpit simulator, an Air Traffic Control simulator, attd a pseudo-aircraft control and simulation management station. The pseudo-aircraft control station also functions as a scenario design tool for coordinating human factors experiments. This station incorporates apseudo-pilot ilttetface designed to reduce workload for human operators piloting multiple aircraft simultaneously in real time. The application of this distributed simulation facility to support a study of the effect of shared infotvnation (via air-ground datalink) on pilot/controller shared situation awareness and re-route negotiation is also presented. 1. Introduction operator(s) in the loop. Because the systems being researched are typically evaluated early in the conceptual Human-automation interaction is a critical consideration phase, the ability to rapidly prototype and exercise many in the design and operation of advanced avionics and Air alternate designs is of particular importance. Given this Traffic Control (ATC) systems. The MIT International dynamic environment, flexibility and freedom in the Center for Air Transportation (ICAT) has developed an design of the simulation architecture are important integrated human-centered systems approach to the considerations. design and evaluation of new air transportation technologies such as terrain avoidance systems, heads-up One area of current research at ICAT is advanced display (HUD) systems, and air-ground datalink information and communication systems, including air- systems [1,2]. This approach, which considers the human ground datalink systems. Of particular interest is the as an element of the closed-loop control system, relies effect of such systems on air traffic controller/pilot heavily on the use of real-time, moderate-fidelity interaction and shared situation awareness. A distributed simulation to evaluate prototype systems with the human simulation of a portion of the national airspace environmewnat sdesigneadnddevelopetodsupporthtis (DistributeIdnteractivSeimulationo)rHLA(HighLevel researchf,acilitatingthe evaluationof alternativeArchitectureT).his architecturiencorporateesxisting datalinkconceptsT.hedistributedsimulationfacility applicationinstoasimulatiopnrotocoolntopofasimple includesan advancedcockpitsimulator,an ATC networckommunicatiolanyser. simulatora,ndapseudo-aircrcaoftntroal ndsimulation managemestnattion. 3.1 Network architecture 2.Requirements Network communications are handled by standard Transmission Control Protocol/Internet Protocol A distributedsimulationwas neededto place (TCP/IP) layer sockets using full-duplex byte streams. experimenthaul mansubjectosperatingseparatfelight This system was primarily designed to run on and ATC simulatorsin a commonsimulation workstations connected to an Ethernet 10Mbps Local environment.Experimentsdesigned to study Area Network (LAN), although the use of TCP/IP pilot/controllerinteractionsrequire a real-time communications allows applications to be run from simulatiofnacilitycapabloefmodelinagndcoordinatingremote locations that are connected to the Internet. This representatioonfsweathearndtrafficbetweetnhepilot network implementation is simplified by relying on high andcontrollesrubjectVs.oicecommunicatioanmsonagll bandwidth, reliable connectivity. At the hardware level, participanitns thesimulatioanrenecessatroyfacilitate however, network integrity and bandwidth are sensitive theverbailnteractionusndeirnvestigatioAn.centralizedto other hosts that are not part of the distributed meanosfrecordindgataandvoicecommunicatiofrnosm simulation, but are still connected to the same LAN thesimulatioinsnecessafroyranalysisa,ndtheabilityto segment. Network traffic or errors from these hosts recalal ndplaybacpkreviously-recorsdiemdulatiornuns degrade the performance of the distributed simulation isneedetdofacilitatethesubsequednetbriefinogftest unpredictably during a simulation execution. Large subjectsF.inally,a flexiblearchitecturisedesirablseo simulations, which use all of the network bandwidth, thatnewsimulationobjectcsanbeeasilyimplementedmay require computers participating in the simulation to andmodified. be isolated to an independent LAN segment. Additionatloolsarealsorequiredforgeneratintgest The network architecture follows a client-server model, scenariosand managingair traffic in real time as illustrated in Figure 1, which centralizes at the throughouthtesimulationH.umanfactorsexperimentssimulation host the collection and distribution of oftenattemptto studyspecificinteractionbsetween simulation data. Client applications, which may be flight humanasndautomatioonrotherhumansS.cenariothsat simulators, ATC simulators, weather services (e.g., the placehumansubjectisnsituationrsequiringaresponseTotal Atmosphere Ocean Space (TAOS) system [3]) or mustbedesignedandcoordinatetdo stimulatethese interactionsIn. orderto generatesuchscenariosa, scenarimoanagemeanptplicatiomnustbeabletosetthe initialstatefsorallaircraftinthescenarioW.hilethese initialstatesarethesameforeachexecutionofthe simulationth,eactionosfthehumansubjectwsillvary. ,, % Thereforael,laircrafntotundetrhecontroolfhumantest Simulation Host I II I I I I I I subjectms ustbecontrolledin realtimeduringan experimentot emulateeachaircraft'sresponsteo its [% ii1 I environmeinntarealisticmanner. %?X _ II Ii % tl 3.Distributed SimulationArchitecture x ,,_ psecodOCA?[eC_a ft__ ,, Therequiremenfotsradistributesdimulatioanppropriate Simulation data path Audio data path forhumanfactorrsesearcmhotivatetdhedevelopmeonft a customsimulationarchitecturethat could be Figure 1. Network architecture for the distributed implementeadndtailoredmoreeasilythanexisting simulation. distributedsimulationarchitecturess,uch as DIS otherapplicationcsa,nenterorleavethesimulatioant direction of audio data only. The host may continue to anytimebyconnectinogrdisconnectifnrgomthehost. receive and log the audio data, but it is not responsible Thenumbeorfsimultaneocuosnnectionssupportebdy for the distribution of the data. Participation in voice thehostworkstations'systemkerneloftenlimitsthe communications is therefore limited to clients that are numbeorfclientsthatmayconnectotahostb,utnoother explicitly declared at the outset of the simulation, because limitationasreimposebdythehosstoftware. live communication streams must be established between all of the client computers. One advantage of 3.2 Simulation architecture decentralizing the voice communication is that multiple communication groups, analogous to different The host application controls all of the airspace frequencies in radio communications, can be defined. A information, such as the locations of airports and client may be programmed to participate in multiple navigational aids, which are sent to client applications communication groups at once, allowing the.client when requested, usually when the client first connects. operator to "tune" to a different communications channel Once connected, clients may declare objects (at present ("frequency") when appropriate, although this capability limited to aircraft and ATC types) that will be controlled has not been implemented in existing client applications. by the client in the simulation. Clients may declare new objects (e.g., aircraft taking off) or remove existing 4. Client Applications objects under their control (e.g., aircraft landing) at any time during the simulation. There are no software limitations to the number of objects that may be declared In the following discussion, the screen captures from the by a client application. different client applications that appear in Figures 2, 3, 4, and 5 were taken simultaneously during a simulation The simulation host is responsible for keeping the execution. During the discussion, note how the same simulation time. Updates of the simulation time are weather cell and air traffic are perceived from the transmitted to client applications only when the client different client applications. first connects, when the simulation time is disruptedm 4.1 Advanced Cockpit Simulator (ACS) such as when the simulation is paused---or when a client explicitly requests an update. The advanced cockpit simulator (Figure 2) is a part-task flight simulator that was developed to study human The host application is also responsible for maintaining a performance issues associated with advanced cockpit log of the simulation execution. For analysis, the host systems. The simulator emulates the Electronic Flight may be restarted in a playback mode to replay the Instrument System (EFIS), Flight Management previously recorded simulation. Clients can then connect to the host to observe the simulation. For example, the flight simulator can connect using the same aircraft identifier string as any of the original aircraft in the simulation, and the cockpit simulator's attitude, trajectory, and alerting displays will reflect those of the original aircraft, even if that aircraft was a pseudo- aircraft. 3.3 Voice communications Voice communications are also sent over the network, but the audio data is sent separately from the simulation data directly between the client computers in order to prevent transmission delays and to reduce the network load on the host computer. The dashed lines in Figure 1 represent the path of voice communications. While these are also full-duplex byte streams, voice data is sent in only one direction and is acknowledged in the return Figure 2. Advanced Cockpit Simulator (ACS) display, direction. The arrows on these paths indicate the including prototype air traffic and weather displays. Compute(FrMC)a,ndautoflighstystemfoundinmodern used at most en route ATC centers in the United States. "glass-cockptitr"ansporatircraftsuchastheBoeing The PVD displays radar tracks and full data blocks for all 757/76o7r747-400E.ntryofflightpathinformatioinnto tracked aircraft in the simulation within its assigned theFMCisaccomplishtehdroughareplicaoftheBoeing airspace sector, along with sector adaptation data such as 757/76C7ontroalndDisplaUynit(CDU).Theautoflight airports, navigation aids, and airways. Although aircraft systemiscontrolletdhroughaBoeing737-20a0utopilot position updates are received continuously, target ModeControPl ane(lMCP).Directflightcontrolsare positions are updated once every 12 seconds on the PVD availableusinga side-stickcontrollerandthrottle to emulate the update rate of the actual ATC equipment. quadranat,lthoughthesearenottypicallyusedwhen Trackball inputs and/or alphanumeric keyboard evaluatinoguter-loopc,ognitive-leviseslueswhereit is commands may be used to display supplementary assumethdataircraftcontrowl ouldbeperformeudsing information such as a target's current trajectory, filed theautoflighstystems. flight plan, or position history. The same input devices may be used to zoom or offset the plan view display. All Thecockpitsimulatofreaturesadvanceadlertingand data entry keyboard/mouse input sequences emulate those displaysystemfosrtraffict,errainandweatheAr.Traffic of the real DEC. In addition, a new NEXRAD-based alertandCollisionAvoidancSeystem(TCAS)provides weather display prototype has been integrated into the advancedwarningof potentialconflictswith other ATC simulator to support ongoing research into air- aircraftin the simulationA. n EnhancedGround ground datalink systems. In Figure 3, which shows the ProximityWarningSystem(EGPWSi)ncludesplan-, ATC simulator display, flight plan information and a 6- profile-a,ndperspective-dispolafyssurroundintegrrain. mile segmented circle are displayed for the subject Awindsheaarlertingsystemindicatetshepresencaend aircraft being simulated by the ACS. locationofdetectemdicrobursatctivityI.nadditionn,ew traffic and weatherdisplayprototypeshavebeen 4.3 Pseudo-Aircraft Controller integrateindtothecockpistimulatotrosupporotngoing researcinhtoair-grounddatalinksystems. The pseudo-aircraft control station (Figure 4) manages simulation scenarios for human factors experiments in a 4.2 Air Traffic Control Simulator distributed environment. This application allows for the creation and coordination of scenarios designed to place The Air Traffic Control (ATC) part-task simulator human subjects in predetermined situations, so that the emulates the Plan View Display (PVD), Computer response of the human subject to the situation can be Readout Display (CRD), and Data Entry Control (DEC) studied. The client software enables a human operator to quickly control and manage the simulated air traffic in real time during an experiment. This application also simulates the flight dynamics of all pseudo-aircraft under its control. (For large simulations, this task may be distributed among multiple workstations running this client application, each controlling a subset of the pseudo-aircraft traffic.) Many existing pseudo-aircraft control applications require the pseudo-pilot to use mouse clicks and alphanumeric commands to effect changes in flight paths or flight plans of the simulation pseudo-aircraft [4,5]. This control scheme requires the pseudo-pilot to quickly alternate between the mouse and keyboard. While this may be acceptable for small numbers of pseudo-aircraft or infrequent clearance changes from ATC, it quickly becomes unmanageable in the high-density, high- workload environments that are of primary interest in Figure 3. Air Traffic Control Simulator display, including current Air Traffic Management (ATM) research. a new NEXRAD-based weather display prototype. attitudea,irspeeda,ltitude,headinga,ndflightcontrol modea,swellasitscommandesdtates(Figure5).The pseudo-pilmotayalsodisplaythecurrenwt aypoinftosr the selectedaircraft,bothtextuallyin a list and graphicalolynthePVD. If an aircraft object is under the pseudo-pilot's control (as distinguished by its blue color; other aircraft appear red on the display), the pseudo-pilot may change the aircraft's commanded states by using the second mouse button to click in the appropriate area of the screen (using the second mouse button rather than the primary mouse button prevents the pseudo-pilot from inadvertently changing the commanded state of a pseudo- aircraft). When the mouse cursor is in the PVD, a heading cue is displayed at 5-degree increments on the compass rose surrounding the selected aircraft and is also Figure 4. Pseudo-Aircraft Controller display. Figure 6. Compass rose surrounding a pseudo-aircraft with heading control cue displayed. The small set of crosshairs appearing to the right of the navigational aid is the mouse pointer. shown numerically (Figure 6). This cue aids the pseudo- pilot in determining the heading to a navigational aid or Figure 5. "Pseudo-Cockpit" display. a heading clear of weather. This heading can be commanded by clicking the second mouse button. Similarly, a target altitude or airspeed is selected by An interface to the pseudo-aircraft control application clicking on the appropriate tape indicator. Flight control was developed to enable real-time control of pseudo- modes are set by clicking on the flight mode aircraft by providing the pseudo-pilot with an intuitive annunciators shown in Figure5. Using just these "point-and-click" interface to exercise outer-loop control controls, a pseudo-pilot is able to perform most of the of an aircraft's autoflight systems. The pseudo-pilot is routine tasks necessary to manage the air traffic during a provided with a plan view display of the air traffic, which simulation. (Note that in Figure 5, the subject aircraft is continuously updated during the simulation. The being simulated by the ACS is selected as the active pseudo-pilot can click on any aircraft to display that aircraft, so it cannot be controlled from the pseudo- aircraft's "pseudo-cockpit", showing the aircraft's current aircraft control station.) Scenargioeneratioanndmoresophisticatmedanipulation 5. Execution Example ofpseudo-aircraft--sauscphrogramminagndmodifying waypoinotsrchangintgheactuasltatersatherthanthe This distributed simulation facility is currently in use to commandesdtatesof an aircraft--areaccomplished support a study of the effect of shared information (via usingacommanldineinterfaceT.hisinterfaciencludes air-ground datalink) on pilot/controller shared situation commandfosrcreatingn,aminga,ndremovinagircraft; awareness and re-route negotiation. The experiment pairs manipulatinagndcopyingaircrafwt aypointasn;dsaving a commercial airline pilot subject with an en route air andrestoringscenarioswhichprovidethe initial traffic controller subject in a real-time simulated air conditionfosradistributesdimulation. traffic environment. The availability of shared traffic and weather information is manipulated as an independent The pseudo-aircracfotntrolleralso includessome variable in the experiment. elementsof a robustsituationgeneratioanpproach developebdyJohnson [6]. Robust situation generation is Test scenarios intentionally bring the goals of the pilot a method of automating pseudo-aircraft trajectories using and controller into conflict in re-routing situations. state feedback to generate specific aiT traffic situations. Subjects interact within the simulation environment to For example, an experiment may require a collision resolve traffic and weather conflicts. Of particular hazard situation if no action is taken by the experimental interest are indications of each subject's recognition of subjects. The ability to reliably generate this situation is the other's constraints, anticipation of needs and/or sensitive to the unexpected actions of the human subjects desires, willingness to comply/cooperate, and persistence (e.g., an unrelated course deviation requested by ATC in pursuing an alternate solution. The experiment will long before the desired conflict). To make the situation provide input in terms of the potential for shared more robust, the pseudo-aircraft can be set to adjust its information to effect more collaborative or competitive speed to arrive at the desired conflict location at the interaction between pilots and controllers. appropriate time. Only some elements of the robust situation generation implementation could be included In this experiment, each pilot/controller pair participates for use in the pseudo-aircraft control software, because in six scenarios. The discussion that follows focuses only many of the actions that pseudo-aircraft must take to on one run of the distributed simulation executed during reliably generate a situation require ATC clearance. this experiment as an example of the performance typically achieved by the distributed simulation facility. Finally, due to its real-time display and control interface, This particular scenario contained one subject aircraft an instance of the pseudo-aircraft controller client simulated using the ACS, 16 pseudo-aircraft controlled running idly (i.e., controlling no pseudo-aircraft) is ideal by a single execution of the pseudo-aircraft control for observation of the simulation by those not actively application, one air-traffic controller, and a weather participating. It may also be used to view playbacks of front, which provided the impetus for re-route the simulation. This is especially useful during the negotiation. In this case, both the ACS and the debriefing portion of a human factors experiment, when simulation host application were run on an SGI Indigo 2 it may be beneficial for the test subjects to review the workstation. The ATC simulator was run on another SGI simulation with all weather and traffic information Indigo 2 workstation and the pseudo-aircraft control revealed. station was run on an SGI Octane workstation. The audio logging function was separated from the simulation host 4.4 Weather Application (TAOS) and run on an SGI Indigo workstation. For the demonstration and experiment described herein, The 16 pseudo-aircraft which comprised the surroundin,, NEXRAD-based weather was integrated into the cockpit, air traffic were managed by a single pseudo-pilot who controller and pseudo-aircraft displays statically (see was also responsible for accepting and responding to Figures 2, 3, and 4). The data was collected and archived radio calls from the air traffic controller. The number of by a tool like TAOS (Total Atmosphere Ocean aircraft that a single pseudo-pilot can manage using the Space [3]), and then a series of static images were pseudo-aircraft controller is dependent on the pseudo- distributed off-line to the simulation suite. There was no pilot's experience, so an upper limit to this number could link to real-time dynamic weather during the simulation. not be determined. Figure7showstheairtrafficandweathefrrontasseen fromthe pseudo-aircracfotntrolstationduringthis 200 simulationexecutionD.uringthisexecutionb,oththe subjecptilotandthecontrollehradaccestsoairtraffic ,=.. ra_ andtheweatherradarinformationT.omaintainaircraft 150 separationand avoidthe hazardouwseather,the controllerissued17 route amendmenotsver the / 100 executiontw'selve-minudteurationE.ightofthepseudo- aircraftwereforcedtodeviateoffcoursetoavoidthe I,,= " 50 weathefrrontand/oortherairtraffic.Thepseudo-pilot wasabletonegotiataendsuccessfuallcycomplisahll14 ATCclearanccehangedsirectedtowardthepseudo- 0 0 200 400 660 aircrafitnrealtime. Simulation time Is] Figure 8. The data rate of the simulation data transmission plotted as a function of time during a single execution of the simulation. required for the simulation data was 156 Kbytes/s. 75 audio transmissions were made during the simulation, each lasting an average of 3.9 seconds. The data rate for the audio data was 16 Kbytes/s, increasing the network load by an additional 48 Kbytes/s during each transmission. Because data must be repeated to each client application subscribing to the data, the bandwidth requirements for the simulation execution scale linearly with the number of clients connected. The bandwidth requirements do not necessarily scale linearly with the number of objects in the simulation, because the update rate for each object in the simulation depends on the speed of the computer controlling that object. Although the voice communications functioned normally during this execution of the simulation, some runs that were of comparable complexity as the one described above experienced interruptions and delays in the audio transmissions. Voice communications, which are more sensitive to network delays than the simulation data transmissions, may have been interrupted by an increase Figure7.Airtrafficandweathefrrontasviewedfromthe in the load on the network that was observed during these pseudo-aircrcaoftntrolstationseveraml inutesintoa executions (while no attempts were made to completely simulatioenxecution. quantify these delays, the network latency measured during these executions was on the order of a second, Figure8 showsthedataratesexperiencedduringthis compared to the millisecond latency experienced during executioonfthesimulationn,otincludingthebandwidthnormal network operations). It is likely that transmission requirebdythevoicecommunicatioTnhse.sevaluewsere of the simulation data was similarly delayed during these obtainefdromthesimulatiolnogfilesbyaveragintghe executions, although this was not noticeable to the amounotfdatabeingtransmittedduringeachseconodf human subjects. As discussed in Section 3.1, future thesimulationT.hereforeth,esevaluedsonotreflectthe simulation exercises may require that participating actualinstantaneoutrsansmissionratesexperiencedcomputers be isolated to an independent LAN segment. duringthesimulationIn. thiscaset,heaveragdeatarate Thisdistributedsimulationarchitecturheasalsobeen 7.Future Work validateidnaremotesimulatioenxecutioinncorporating simulatorfacilitiesat MIT, locatedin Cambridge,Inordertotakeadvantagoefreal-timeweathedratat,he Massachusetatsnd TASC, locatedin Reading,airtrafficmanagemesnimt ulatiocnouldtransitiotnoDIS MassachuseTttsA.SCinstalledthehostsoftwareand orHLA,whichwouldallowittomakeuseoftheweather actedasthesimulatiosnerverT.ASCalsoinstalledand andeffectsservecrapabilitieosfTAOS.TAOSprovides executetdheATCclientapplicationandthepseudo-consistent, tactically significant, high-fidelity aircraftcontrollearpplicationw,hileMITexecutetdhe environmendtaaltaondemantdodistributesdimulation advancecdockpistimulatoTr.hesimulatioanppearetod federationTsA.OSenvironmendtaaltaservicperovideas functionnormallya,lthougvhoicecommunicatiownesre detaileddynamicdescriptionof the combined notattempteindtheremotseimulation. atmosphere-ocean-linttaotruarlaelnvironmeunsting4-D grids(threespatiadlimensionpslustime)toprovidea 6.Conclusion commornepresentatioofntheenvironmentbaalsefields andembeddefedaturesB.asefieldsdescribteheambient Adistributerdeal-timesimulatioonfthecivilairtraffic conditionss,uchasatemperatuorerwindfield,while environmendtevelopedto supporthumanfactors embeddefdeaturesarefine-scalelocalizedprocesses, researcihnadvanceadirtransportatiotenchnologhyas suchascloudsorduststormsT.AOSprovidelsinkstoa beenpresenteTdh.edistributeednvironmeinstbaseodn widevarietyofexternadlatasourcersa,ngingfromlive acustomsimulatioanrchitectudreesignefdorsimplicity observationasnddatafieldsfromoperationaslources andflexibilityinhumanexperiments. (e.g.,commerciaraldarfeedsandAWN,Automated WeatherNetwork),to authoritativegriddedforecast Severalclientapplicationsmincludianng advancedproductsprovidedby DMSO's MEL (Master cockpistimulatora,nenrouteATCsimulatora,nda EnvironmenLtaiblraryo)rpublicInternestites. pseudo-aircrcaoftntrosltationmhavbeeendevelopetod supporrteal-timeexperimenwtsithhumanisntheloop. Futuredevelopmeonftthissimulationfacilitycallsfor Thepseudo-aircrcaoftntrosltationinparticulaernablestheintegratioonfreal-timeweathemr odelsto, include thecreatioonfscenariothsatgovernahumanexperimentfour-dimensionwailnd,temperaturteu,rbulenceic,ing, ina distributeednvironmenOt.ncethesimulationhas andconvectiveweatherphenomenaT.heseweather begunth,epseudo-aircrcaofnt trosltationenableassingle elemenatsrecriticatloarealisticsimulatioonfairtraffic usetromanagmeultipleaircrafetmulatintgheairtraffic managemeAnltt.houghthissetofweatheprarameteirss observebdythehumansubjects. slightlydifferentthanthedatasetprovidedduringthe STOW'97ACTD(SynthetiTcheateorfWarAdvanced This distributedsimulationfacility has been ConcepTtechnologDyemonstratioann)dUSACOM's demonstrateind a studyof pilot/controllerre-route (U.SA. tlanticCommanedx)erciseU,nifiedEndeavoUrE negotiationthat is evaluatingalternativedatalink 98-1,TAOScanprovidethe additionapl arameters concepts.The experimentsuccessfullyjoined describingturbulencaendicing.Howevert,hereare pilot/controllerpairs in a distributedairspaceissuesto beaddressewdiththetemporaalndspatial environment,although some difficulties were resolutioonfthedatarequirefdorairtrafficmanagement encounterwediththevoicecommunicatioPnsre.liminaryscenariothsattypicallyruninasmalleprlaybo(xonthe resultsfromthisstudyindicatethatsharedinformationorderof severahlundrednauticaml iles,withgreatest improvesthe situationawarenesosf pilots and interesitntheareasurroundinagnairport)andovera controllersW. hilethereisevidencferomthisstudythat muchshortetrime(ontheordeorfminuteosrhours). improvedsituationawarenesesnablespilots and controllertso workmorecollaborativeilnyre-routing 8.Acknowledgments situationsth,ereis alsoevidencferomthisstudythat improvedsituationawarenescsausesmistrustor Thisworkis supportebdyTASCaspartoftheFAA frustratiownhenthegoalsofthepilotandthecontroller CenteorfExcellencineOperationRsesearcahndbythe areinconflictT.hedistributesdimulatiofnacilitywillbe NationaAl eronauticasndSpaceAdministration/Ames usedtoexplorethesehumanfactorsissuemsorefullyin ResearcChenteurndegrrantNAG2-716T.AOSworkis futureexperiments. supportebdytheDefensAedvanceRdesearcPhrojects AgencayndtheDefensMeodelinagndSimulatioOnffice throughitsModelingandSimulatioEnxecutivAegents TODD FARLEY is a graduate student at MIT in the for the Natural EnvironmentT. he U.S. Army Department of Aeronautics and Astronautics, where he is TopographEicngineerinCgentesrerveasstheDARPA a research assistant at the MIT International Center for agenftorSynthetiEcnvironmenatnsdasthecontracting Air Transportation. organizatiofonrthiswork. R. JOHN HANSMAN is a Professor at MIT in the 9. References Department of Aeronautics and Astronautics, where he is Head of the Humans and Automation Division and [1] E.C. Hahn and R.J. Hansman, Jr.: "An Experimental Director of the International Center for Air Study of the Effects of Automation on Pilot Transportation. He conducts research in several areas Situational Awareness in the Datalink ATC related to flight vehicle operations and safety. His current Environment" ASL-92-1, Department of research activities focus on advanced cockpit information Aeronautics and Astronautics, Massachusetts systems, including Flight Management Systems, Air- Institute of Technology, May 1992. Ground Datalink, Advanced Alerting Systems, and [2] A.H. Midkiff and R.J. Hansman, Jr.: "Identification Flight Crew Situational Awareness. of Important 'Party Line' Informational Elements and the Implications for Situational Awareness in JOHN LADIK is a principal member of the technical the Datalink Environment" ASL-92-2, Department staff at TASC involved in systems engineering, of Aeronautics and Astronautics, Massachusetts mathematical modeling, and statistical analysis. Institute of Technology, May 1992. [3] D.A. Whitney, R.A. Reynolds, D.Z. Sherer, P.S. DANA SHERER is a senior member of the technical Dailey, M.L. Driscoll, MAJ M. Zettlemoyer, CDR R. staff at TASC. Her expertise is in systems engineering Schultz, and I. Watkins: "Impacts of the with a focus in simulation and modeling, environmental Environment on Warfighter Training: STOW 97 simulation systems, distributed computing, data analysis Experiences with TAOS" 98S-SIW-224 Spt4ng and scientific data visualization. Simulation Interoperabilio' Workshop, March 1998. [4] R.A. Weske and G.L. Danek: "Pseudo Aircraft Systems: A Multi-Aircraft Simulation System for Air Traffic Control Research" Proceedings of the AIAA Flight Simulation Technologies Conference, Monterey, California, August 1993. [5] J. Klinge, K. Smith, and P.A. Hancock: "DATIDS: The University of Minnesota Distributed Air-Traffic Information Display Simulator" Proceedings of the 9th hlternational Symposium on Aviation Psychology, Columbus, Ohio 1997. [6] E.N. Johnson" "Multi-Agent Flight Simulation with Robust Situation Generation" Master's Thesis, Department of Aeronautics and Astronautics, MIT, 1995. Author Biographies KEITH AMONLIRDVIMAN is an undergraduate student at MIT in the Department of Aeronautics and Astronautics, where he is a research assistant at the MIT International Center for Air Transportation.