Pilot Interaction with TCAS and Air Traffic Control - Semantic Scholar
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS
Pilot Interaction with TCAS and Air Traffic Control Amy R. Pritchett, Elizabeth S. Fleming, William P. Cleveland, Jonathan J. Zoetrum, Vlad M. Popescu, Dhruv A. Thakkar Cognitive Engineering Center Georgia Institute of Technology, Atlanta ABSTRACT
persistent Traffic Situation Display (TSD) as well as the occasional traffic advisory (TA) and resolution advisory (RA) that it provides to the pilot. [6] Further, TCAS adds the dynamic of its TAs and RAs requiring information search and assessment, decision making and control actions. This dynamic is coupled with the dynamic inherent to the pilot’s interaction with air traffic control, including their ability to overhear party-line communications. [7,8,9] Thus, we hypothesized that pilots’ interactions with air traffic control may impact their response to TCAS advisories; similarly, TCAS may impact pilot maneuvering relative to air traffic instructions and pilot communications with air traffic controllers.
The Traffic alert and Collision Avoidance System (TCAS) is often framed as separate to -- or opposing -- air traffic control, i.e. as a backup, redundant system that invokes resolution advisories (RAs) when all other methods of separation assurance fail. A flight simulator experiment examined pilot interaction with TCAS in the context of a full air traffic environment. This paper will describe how pilot responses to RAs are impacted by immediately-prior interactions with the air traffic controller (notably whether they were provided a call-out of the traffic that then caused a traffic event, could overhear party line information relevant to the build of a traffic event, or received air traffic control instructions just before an RA that conflicted with the RA's advised maneuver). Likewise, this paper will describe how features of the RA impact subsequent pilot communications with the air traffic controller.
This paper describes an experiment with airline pilots in an integrated air traffic control and flight simulator facility, examining a range of factors impacting pilot interaction with TCAS. A flight simulator experiment examined pilot interaction with TCAS in the context of a full air traffic environment. An air traffic simulator replicated 'real' operations into Dallas-Fort Worth airport during busy arrival flows. Sixteen airline pilots, each acting as captain and pilot flying of a B747-400 in a medium-fidelity flight simulator in instrument meteorological conditions, interacted as he would in real operations with an experimenter acting as air traffic controller. The surrounding traffic was visible to the pilot on the TCAS traffic situation display. The controller also interacted with these other aircraft as in real operations, with another experimenter acting as the voice of these other aircraft so that the participant-pilot could also over-hear the full 'partyline.’
Categories and Subject Descriptors
J.2 [Physical Sciences and Engineering]: Aerospace. General Terms
Performance, Reliability, Experimentation. Keywords
TCAS, Collision Compliance. 1
avoidance,
human
factors,
ATC,
INTRODUCTION
The Traffic alert and Collision Avoidance System (TCAS) has been in operational use for approximately 20 years. While this system has been extremely successful in reducing the risk of midair collisions, human factors concerns remain with pilot interactions with TCAS, as reflected in lower rates of pilot compliance to TCAS advisories than assumed in safety studies, and other impacts on interaction with air traffic control. [1,2,3,4,5]
This paper starts by overviewing pilot responses to TCAS overall, and then specifically examines how pilot responses to RAs are impacted by immediately-prior interactions with the air traffic controller (whether they were provided a callout of the traffic that then caused a traffic event, could overhear party line information relevant to the build of a traffic event, or received air traffic control instructions just before an RA that conflicted with the RA's climb maneuver. Likewise, this paper describes the timing of pilot communications before, during and after TCAS advisories.
TCAS is often framed as separate to -- or opposing -- air traffic control, i.e. as a backup, redundant system that invokes resolution advisories (RAs) when something within the system has failed. However, rather than being separate, TCAS may also be viewed as introducing another dynamic into the flight deck. This additional dynamic includes the information about traffic provided by TCAS, including its Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Re-publication of material on this page requires permission by the copyright owners. ATACCS’2012, 29-31 May 2012, London, UK. Copyright 2012 IRIT PRESS, ISBN: 978-2-917490-20-4
117
2 DESCRIPTION OF SIMULATOR STUDY 2.1 Participants
Sixteen airline pilots participated. All of the participants were male, ranging in age from their mid-20’s up to 59 years old. Eight held the rank of Captain in their airline, seven were ranked as first officers, and one did not respond
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS
Audio Communications (Aviation Intercom)
SideStick
Simulation Architecture
B747-400 Simulator (RFS)
PFD
First Officer
ND
HLA Connection TCAS Alerts Aircraft State
TCAS Logic
Experimenter
Charts & Checklists
Touch screen
PartyLine
Homing Guidance for Intruder Aircraft
Captain
TSD PFD ND
Air Traffic Simulator (TGF)
Air Traffic Transcripts
Relative Waypoints to Create RA
Eyetracker
ATC
VGA ATC->TSD
Coded Log of Flights Figure 1. Schematic of the simulator setup, including both flight simulator and air traffic simulator to the question. Eight of the sixteen of the pilots reported having received some form of military training in their aviation career. Eight of the pilots reported being very familiar with the airport used for this study (Dallas- Fort Worth), seven of the pilots had some familiarity with the airport, and one pilot reported having no familiarity.
for flaps, gear and speedbrakes; the pilot could also interact with these systems through a mouse or trackball provided on the throttle quadrant. The pilot was also provided with an approach checklist and all appropriate charts. Additionally, an air traffic control station used the traffic display and traffic emulation provided by the FAA’s Traffic Generation Facility (TGF), populated with traffic flows recorded from real operations. Additionally dynamic models of intruder aircraft steered intruder aircraft relative to the participant’s aircraft. Through a priori analysis, their relative trajectories were pre-determined such that, barring significant deviations from the participants’ expected flight profile, traffic events would create specific TCAS advisories of interest: only the pre-cautionary TA, or both a TA and an RA of a pre-determined type.
2.2 Apparatus
As shown in the schematic in Figure 1, the simulator apparatus included several components. Most notable to the participant, the ‘Reconfigurable Flight Simulator’ software provided an emulation of a two-crew flight deck in which the participant sat in the left seat to act as Captain and Pilot Flying. The flight displays and underlying models of the aircraft dynamics and aircraft systems were based on a B747-400, and had been used in prior studies with airline pilots. [10,11,12] A basic physical mockup of the flightdeck mounted computer monitors in approximately the correct location for each, an provided the captain with a side-stick at his left hand and mockup of a throttle quadrant’s structure on his right. The monitor in front of the pilot provided an accurate representation of the B747-400 Primary Flight Display (PFD) and Navigation Display (ND); this was mirrored on the first officer’s die. A large touch screen in front of the throttle quadrant provided the Control Display Unit (CDU), Mode Control Panel (MCP), control’s for the captain’s ND, Engine Indication and Crew Alerting Display (EICAS), and levers
An experimenter acting as air traffic controller interacted with the pilot according to proper air traffic procedures via a voice channel established using an aviation intercom. Another experimenter acted as ‘party-line,’ i.e. providing the voice communications of all other aircraft, aided by pre-established transcripts. Both the captain (participant) and first officer (experimenter) wore aviation headsets with a ‘hot’ intercom within the flightdeck and push-to-talk button to broadcast on the air traffic frequency. Thus, the apparatus afforded realistic air traffic communications.
118
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS
TCAS was emulated according to the standards required of the Minimum Operational Performance Standards (MOPS) for TCAS [13]. The full alerting and advisory-generation logic for TCAS Version 7.1 was implemented from the pseudo-code provided by the MOPS for two-aircraft conflicts. The resulting advisories were portrayed to the pilot via the required aural call-outs (note: the historic ‘Adjust Vertical Speed Adjust’ aural was used in place of the new ‘Level Off’ because the participants were not yet familiar with the new aural). The ND displayed “TRAFFIC” in yellow/red for the TA/RA respectively, and the PFD provided the RA maneuver guidance both in terms of a red-arc on the vertical speed tape and a ‘trapezoid’ on the attitude indication indicating the pitch attitude corresponding to the vertical speed red arc.
other aircraft on the shared voice frequency. While the FO was responsible for handling air traffic communications, they could also be overhead by the captain and the captain could call for, or make himself, any call to air traffic control as desired. The FO was responsible for ensuring the captain understood air traffic instructions directed to their aircraft, but otherwise was not to coach the pilot about the traffic situation as shown by the TSD and by party line communications. Before some events the FO and captain were given an explicit ‘call-out’ by the air traffic controller (e.g. “GT123, traffic your 3 o’clock, 1000 feet below”) which the FO would repeat to the participant if necessary. Before other events the air traffic controller communicated with the aircraft causing the traffic event, providing ‘party-line information’ which could be useful in predicting the development of the traffic event, but which may not be apparent to the participant without careful monitoring of the voice communications and which was not repeated or discussed to the participant by the FO. Thus, the pilot’s tasks included monitoring disparate sources of traffic information and evaluating potential maneuvers as displayed by TCAS or instructed by air traffic control.
The TCAS TSD was provided by a 7” display mounted in the flight simulator above the ND, with its range slaved to the ND range selected by the participant. However, it was driven by the computer providing the air traffic simulator and intruder generation, which had knowledge of the location of the other aircraft. The TSD also followed all conventions required by regulatory material [13,14,15]. 2.3 Experimental Task
The pilot’s task was to fly a Standard Arrival Route (STAR) and to perform all required checklists, beginning with the approach briefing and checklist at the start of the flight. The participant acted as the Captain, an correspondingly sat in the left seat of the flight simulator. A researcher familiar with the controls of the aircraft simulator posed as the First Officer (FO). The FO provided the duties of the ‘Pilot Not Flying’ in airline operations, which focus on managing the aircraft systems and interacting with air traffic control. The pilot was briefed on the specific simulator’s systems in initial training runs, and was also informed that the FO was proficient on all aspects of the aircraft’s systems and could be called on at any time.
2.4 Independent Variables
The independent variables described the traffic events experienced by the pilots: whether they received air traffic callouts from the controller, could overhear partyline information, or received air traffic instructions shortly before an RA that conflicted with the RA’s advised maneuver; traffic density (and correspondingly how ‘busy’ the shared voice frequency was); and the intruder trajectory relative to the ownship and the TCAS advisory it created (an ‘corrective’ climb or descend RA displaying a vertical maneuver away from the intruder, a ‘crossing’ RA displaying a vertical maneuver through the intruder’s altitude when the closure rate prevented a maneuver away, or a TA alone without an RA).
Typically, the flights began around an altitude of 10,000 to 20,000 feet and lasted 15 minutes. The flights ended during the approach intercept, i.,e. when the aircraft was within ‘one dot’ of the localizer beam indicating the approach course. The weather was calm with no wind. However Instrument Meteorological Conditions (IMC) applied for the duration of the flight, as there were no out-the-window visuals provided by the simulator; thus, the pilot could only reference a traffic situation display and air traffic communications for information about the traffic situation, and could not visually acquire a target.
Two special cases also reflected events pilots encounter in realistic air traffic operational environment. One of the special cases involved an aircraft under Visual Flight Rules (VFR) crossing 500’ below (legal, but can trigger TCAS advisories) below the pilot’s aircraft. The second event involved advisories while the pilot was intercepting his final approach course.
In each flight, pilots encountered two traffic events. Some of events resulted only in TCAS TAs and required no maneuvering. The more severe events resulted in TCAS RAs which displayed a vertical maneuver to the pilot.
Altogether each pilot flew eight ‘data’ flights described by the letters A through H with two traffic events in each, as summarized in Table 1. The experiment design used an 8th order Latin-Squares design to assign the order of the eight flights to participants. Thus, run order was fully balanced within each of two sets of eight participants.
An air traffic controller was controlling the pilot’s aircraft as well as other aircraft in the vicinity of the airport, and the pilot could hear the party-line communications to the
119
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS Table 1. Description of Traffic Events
Flight / Event Advisory Type and Intruder Trajectory
Traffic Information Traffic Density
A-Event 1
TA
Partyline
Heavy
A-Event 2
Climb RA
Callout
Heavy
B-Event 1
Climb RA
Conflicting
Light
B-Event 2
TA- While on intercept approach
None
Light
C-Event 1
Descend RA
Partyline
Heavy
C-Event 2
Climb RA, caused by VFR traffic
Callout
Heavy
D-Event 1
Descend RA
Partyline
Light
D-Event 2
Preventive RA while on intercept approach
None
Light
E-Event 1
Crossing Descend RA
Callout
Heavy
E-Event 2
TA
Partyline
Heavy
F-Event 1
Crossing Descend RA
Callout
Heavy
F-Event 2
TA
Partyline
Heavy
G-Event 1
TA- Caused by VFR traffic
Partyline
Light
G-Event 2
TA
Callout
Light
H-Event 1
TA
Partyline
Heavy
H-Event 2
Descend RA
Callout
Heavy
2.5 Dependent Variables
as a time relative to the important points of the traffic advisory: before the TA, between the TA and RA, during the RA, or after “clear of conflict.”
Of interest to this paper, objective measures recorded continuously throughout the flights included the simulator’s record of aircraft location, attitude, and pilot’s control actions. During a traffic event, the time of TA and RA were also recorded, as was the maneuver displayed by a TCAS RA through time (which could include ‘strengthening’ or ‘weakening’ maneuver strengths). All audio communications were recorded and analyzed digitally for when the push-to-talk button indicated that the FO was transmitting to air traffic control; while such communications could be routine, in response to air traffic control, during traffic events such transmissions were frequently initiated by the captain (participant) to investigate the status of the intruder-aircraft, announce an avoidance maneuver, or ask for further instructions.
Several other measures were collected, including experimenter coding of pilot responses to TCAS advisories, pilot opinions gathered through questionnaires, and eyetracker logs. However, their analysis is outside this paper’s focus on pilot interaction with TCAS and air traffic control, and will be documented elsewhere. 3
RESULTS
Analysis used SPSS statistical software package. ANOVA used a general linear model allowing for repeated measures. This analysis treated pilots and run order as random effects and examined for fixed effects due to ATC interaction (callout, party-line information or conflicting air traffic instructions), traffic density, and RA type.
Post-hoc categorization provided several measures of the pilots’ responses to the TCAS advisory, as shown in Figure 2. Compliance at any instant to a TCAS RA was defined as, after allowing for a five second reaction time, the pilot meeting (or exceeding) the vertical rate displayed by TCAS and the pilot not performing any horizontal maneuver (defined when the aircraft’s roll angle exceeded 5 degrees).
3.1 Compliance and Maneuvering in Response to RAs
Examining compliance percentage, Table 2, neither run order, nor ATC interaction, nor traffic density had significant effects. Between-pilot variance was significant. As shown in Figure 3, significant differences were found between the four RA types created in the experiment “(Corrective) Climb,” “(Corrective) Descend,” “Crossing Descend,” and “Preventive” RA’s; significant differences were found between these RA types. (Although a Levene test suggested that the data violates the assumption of homogeneity of variances, robust tests of equality of means confirmed significant differences, p
ATACCS’2012 | RESEARCH PAPERS
Pilot Interaction with TCAS and Air Traffic Control Amy R. Pritchett, Elizabeth S. Fleming, William P. Cleveland, Jonathan J. Zoetrum, Vlad M. Popescu, Dhruv A. Thakkar Cognitive Engineering Center Georgia Institute of Technology, Atlanta ABSTRACT
persistent Traffic Situation Display (TSD) as well as the occasional traffic advisory (TA) and resolution advisory (RA) that it provides to the pilot. [6] Further, TCAS adds the dynamic of its TAs and RAs requiring information search and assessment, decision making and control actions. This dynamic is coupled with the dynamic inherent to the pilot’s interaction with air traffic control, including their ability to overhear party-line communications. [7,8,9] Thus, we hypothesized that pilots’ interactions with air traffic control may impact their response to TCAS advisories; similarly, TCAS may impact pilot maneuvering relative to air traffic instructions and pilot communications with air traffic controllers.
The Traffic alert and Collision Avoidance System (TCAS) is often framed as separate to -- or opposing -- air traffic control, i.e. as a backup, redundant system that invokes resolution advisories (RAs) when all other methods of separation assurance fail. A flight simulator experiment examined pilot interaction with TCAS in the context of a full air traffic environment. This paper will describe how pilot responses to RAs are impacted by immediately-prior interactions with the air traffic controller (notably whether they were provided a call-out of the traffic that then caused a traffic event, could overhear party line information relevant to the build of a traffic event, or received air traffic control instructions just before an RA that conflicted with the RA's advised maneuver). Likewise, this paper will describe how features of the RA impact subsequent pilot communications with the air traffic controller.
This paper describes an experiment with airline pilots in an integrated air traffic control and flight simulator facility, examining a range of factors impacting pilot interaction with TCAS. A flight simulator experiment examined pilot interaction with TCAS in the context of a full air traffic environment. An air traffic simulator replicated 'real' operations into Dallas-Fort Worth airport during busy arrival flows. Sixteen airline pilots, each acting as captain and pilot flying of a B747-400 in a medium-fidelity flight simulator in instrument meteorological conditions, interacted as he would in real operations with an experimenter acting as air traffic controller. The surrounding traffic was visible to the pilot on the TCAS traffic situation display. The controller also interacted with these other aircraft as in real operations, with another experimenter acting as the voice of these other aircraft so that the participant-pilot could also over-hear the full 'partyline.’
Categories and Subject Descriptors
J.2 [Physical Sciences and Engineering]: Aerospace. General Terms
Performance, Reliability, Experimentation. Keywords
TCAS, Collision Compliance. 1
avoidance,
human
factors,
ATC,
INTRODUCTION
The Traffic alert and Collision Avoidance System (TCAS) has been in operational use for approximately 20 years. While this system has been extremely successful in reducing the risk of midair collisions, human factors concerns remain with pilot interactions with TCAS, as reflected in lower rates of pilot compliance to TCAS advisories than assumed in safety studies, and other impacts on interaction with air traffic control. [1,2,3,4,5]
This paper starts by overviewing pilot responses to TCAS overall, and then specifically examines how pilot responses to RAs are impacted by immediately-prior interactions with the air traffic controller (whether they were provided a callout of the traffic that then caused a traffic event, could overhear party line information relevant to the build of a traffic event, or received air traffic control instructions just before an RA that conflicted with the RA's climb maneuver. Likewise, this paper describes the timing of pilot communications before, during and after TCAS advisories.
TCAS is often framed as separate to -- or opposing -- air traffic control, i.e. as a backup, redundant system that invokes resolution advisories (RAs) when something within the system has failed. However, rather than being separate, TCAS may also be viewed as introducing another dynamic into the flight deck. This additional dynamic includes the information about traffic provided by TCAS, including its Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Re-publication of material on this page requires permission by the copyright owners. ATACCS’2012, 29-31 May 2012, London, UK. Copyright 2012 IRIT PRESS, ISBN: 978-2-917490-20-4
117
2 DESCRIPTION OF SIMULATOR STUDY 2.1 Participants
Sixteen airline pilots participated. All of the participants were male, ranging in age from their mid-20’s up to 59 years old. Eight held the rank of Captain in their airline, seven were ranked as first officers, and one did not respond
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS
Audio Communications (Aviation Intercom)
SideStick
Simulation Architecture
B747-400 Simulator (RFS)
PFD
First Officer
ND
HLA Connection TCAS Alerts Aircraft State
TCAS Logic
Experimenter
Charts & Checklists
Touch screen
PartyLine
Homing Guidance for Intruder Aircraft
Captain
TSD PFD ND
Air Traffic Simulator (TGF)
Air Traffic Transcripts
Relative Waypoints to Create RA
Eyetracker
ATC
VGA ATC->TSD
Coded Log of Flights Figure 1. Schematic of the simulator setup, including both flight simulator and air traffic simulator to the question. Eight of the sixteen of the pilots reported having received some form of military training in their aviation career. Eight of the pilots reported being very familiar with the airport used for this study (Dallas- Fort Worth), seven of the pilots had some familiarity with the airport, and one pilot reported having no familiarity.
for flaps, gear and speedbrakes; the pilot could also interact with these systems through a mouse or trackball provided on the throttle quadrant. The pilot was also provided with an approach checklist and all appropriate charts. Additionally, an air traffic control station used the traffic display and traffic emulation provided by the FAA’s Traffic Generation Facility (TGF), populated with traffic flows recorded from real operations. Additionally dynamic models of intruder aircraft steered intruder aircraft relative to the participant’s aircraft. Through a priori analysis, their relative trajectories were pre-determined such that, barring significant deviations from the participants’ expected flight profile, traffic events would create specific TCAS advisories of interest: only the pre-cautionary TA, or both a TA and an RA of a pre-determined type.
2.2 Apparatus
As shown in the schematic in Figure 1, the simulator apparatus included several components. Most notable to the participant, the ‘Reconfigurable Flight Simulator’ software provided an emulation of a two-crew flight deck in which the participant sat in the left seat to act as Captain and Pilot Flying. The flight displays and underlying models of the aircraft dynamics and aircraft systems were based on a B747-400, and had been used in prior studies with airline pilots. [10,11,12] A basic physical mockup of the flightdeck mounted computer monitors in approximately the correct location for each, an provided the captain with a side-stick at his left hand and mockup of a throttle quadrant’s structure on his right. The monitor in front of the pilot provided an accurate representation of the B747-400 Primary Flight Display (PFD) and Navigation Display (ND); this was mirrored on the first officer’s die. A large touch screen in front of the throttle quadrant provided the Control Display Unit (CDU), Mode Control Panel (MCP), control’s for the captain’s ND, Engine Indication and Crew Alerting Display (EICAS), and levers
An experimenter acting as air traffic controller interacted with the pilot according to proper air traffic procedures via a voice channel established using an aviation intercom. Another experimenter acted as ‘party-line,’ i.e. providing the voice communications of all other aircraft, aided by pre-established transcripts. Both the captain (participant) and first officer (experimenter) wore aviation headsets with a ‘hot’ intercom within the flightdeck and push-to-talk button to broadcast on the air traffic frequency. Thus, the apparatus afforded realistic air traffic communications.
118
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS
TCAS was emulated according to the standards required of the Minimum Operational Performance Standards (MOPS) for TCAS [13]. The full alerting and advisory-generation logic for TCAS Version 7.1 was implemented from the pseudo-code provided by the MOPS for two-aircraft conflicts. The resulting advisories were portrayed to the pilot via the required aural call-outs (note: the historic ‘Adjust Vertical Speed Adjust’ aural was used in place of the new ‘Level Off’ because the participants were not yet familiar with the new aural). The ND displayed “TRAFFIC” in yellow/red for the TA/RA respectively, and the PFD provided the RA maneuver guidance both in terms of a red-arc on the vertical speed tape and a ‘trapezoid’ on the attitude indication indicating the pitch attitude corresponding to the vertical speed red arc.
other aircraft on the shared voice frequency. While the FO was responsible for handling air traffic communications, they could also be overhead by the captain and the captain could call for, or make himself, any call to air traffic control as desired. The FO was responsible for ensuring the captain understood air traffic instructions directed to their aircraft, but otherwise was not to coach the pilot about the traffic situation as shown by the TSD and by party line communications. Before some events the FO and captain were given an explicit ‘call-out’ by the air traffic controller (e.g. “GT123, traffic your 3 o’clock, 1000 feet below”) which the FO would repeat to the participant if necessary. Before other events the air traffic controller communicated with the aircraft causing the traffic event, providing ‘party-line information’ which could be useful in predicting the development of the traffic event, but which may not be apparent to the participant without careful monitoring of the voice communications and which was not repeated or discussed to the participant by the FO. Thus, the pilot’s tasks included monitoring disparate sources of traffic information and evaluating potential maneuvers as displayed by TCAS or instructed by air traffic control.
The TCAS TSD was provided by a 7” display mounted in the flight simulator above the ND, with its range slaved to the ND range selected by the participant. However, it was driven by the computer providing the air traffic simulator and intruder generation, which had knowledge of the location of the other aircraft. The TSD also followed all conventions required by regulatory material [13,14,15]. 2.3 Experimental Task
The pilot’s task was to fly a Standard Arrival Route (STAR) and to perform all required checklists, beginning with the approach briefing and checklist at the start of the flight. The participant acted as the Captain, an correspondingly sat in the left seat of the flight simulator. A researcher familiar with the controls of the aircraft simulator posed as the First Officer (FO). The FO provided the duties of the ‘Pilot Not Flying’ in airline operations, which focus on managing the aircraft systems and interacting with air traffic control. The pilot was briefed on the specific simulator’s systems in initial training runs, and was also informed that the FO was proficient on all aspects of the aircraft’s systems and could be called on at any time.
2.4 Independent Variables
The independent variables described the traffic events experienced by the pilots: whether they received air traffic callouts from the controller, could overhear partyline information, or received air traffic instructions shortly before an RA that conflicted with the RA’s advised maneuver; traffic density (and correspondingly how ‘busy’ the shared voice frequency was); and the intruder trajectory relative to the ownship and the TCAS advisory it created (an ‘corrective’ climb or descend RA displaying a vertical maneuver away from the intruder, a ‘crossing’ RA displaying a vertical maneuver through the intruder’s altitude when the closure rate prevented a maneuver away, or a TA alone without an RA).
Typically, the flights began around an altitude of 10,000 to 20,000 feet and lasted 15 minutes. The flights ended during the approach intercept, i.,e. when the aircraft was within ‘one dot’ of the localizer beam indicating the approach course. The weather was calm with no wind. However Instrument Meteorological Conditions (IMC) applied for the duration of the flight, as there were no out-the-window visuals provided by the simulator; thus, the pilot could only reference a traffic situation display and air traffic communications for information about the traffic situation, and could not visually acquire a target.
Two special cases also reflected events pilots encounter in realistic air traffic operational environment. One of the special cases involved an aircraft under Visual Flight Rules (VFR) crossing 500’ below (legal, but can trigger TCAS advisories) below the pilot’s aircraft. The second event involved advisories while the pilot was intercepting his final approach course.
In each flight, pilots encountered two traffic events. Some of events resulted only in TCAS TAs and required no maneuvering. The more severe events resulted in TCAS RAs which displayed a vertical maneuver to the pilot.
Altogether each pilot flew eight ‘data’ flights described by the letters A through H with two traffic events in each, as summarized in Table 1. The experiment design used an 8th order Latin-Squares design to assign the order of the eight flights to participants. Thus, run order was fully balanced within each of two sets of eight participants.
An air traffic controller was controlling the pilot’s aircraft as well as other aircraft in the vicinity of the airport, and the pilot could hear the party-line communications to the
119
London, UK, May 29-31, 2012
ATACCS’2012 | RESEARCH PAPERS Table 1. Description of Traffic Events
Flight / Event Advisory Type and Intruder Trajectory
Traffic Information Traffic Density
A-Event 1
TA
Partyline
Heavy
A-Event 2
Climb RA
Callout
Heavy
B-Event 1
Climb RA
Conflicting
Light
B-Event 2
TA- While on intercept approach
None
Light
C-Event 1
Descend RA
Partyline
Heavy
C-Event 2
Climb RA, caused by VFR traffic
Callout
Heavy
D-Event 1
Descend RA
Partyline
Light
D-Event 2
Preventive RA while on intercept approach
None
Light
E-Event 1
Crossing Descend RA
Callout
Heavy
E-Event 2
TA
Partyline
Heavy
F-Event 1
Crossing Descend RA
Callout
Heavy
F-Event 2
TA
Partyline
Heavy
G-Event 1
TA- Caused by VFR traffic
Partyline
Light
G-Event 2
TA
Callout
Light
H-Event 1
TA
Partyline
Heavy
H-Event 2
Descend RA
Callout
Heavy
2.5 Dependent Variables
as a time relative to the important points of the traffic advisory: before the TA, between the TA and RA, during the RA, or after “clear of conflict.”
Of interest to this paper, objective measures recorded continuously throughout the flights included the simulator’s record of aircraft location, attitude, and pilot’s control actions. During a traffic event, the time of TA and RA were also recorded, as was the maneuver displayed by a TCAS RA through time (which could include ‘strengthening’ or ‘weakening’ maneuver strengths). All audio communications were recorded and analyzed digitally for when the push-to-talk button indicated that the FO was transmitting to air traffic control; while such communications could be routine, in response to air traffic control, during traffic events such transmissions were frequently initiated by the captain (participant) to investigate the status of the intruder-aircraft, announce an avoidance maneuver, or ask for further instructions.
Several other measures were collected, including experimenter coding of pilot responses to TCAS advisories, pilot opinions gathered through questionnaires, and eyetracker logs. However, their analysis is outside this paper’s focus on pilot interaction with TCAS and air traffic control, and will be documented elsewhere. 3
RESULTS
Analysis used SPSS statistical software package. ANOVA used a general linear model allowing for repeated measures. This analysis treated pilots and run order as random effects and examined for fixed effects due to ATC interaction (callout, party-line information or conflicting air traffic instructions), traffic density, and RA type.
Post-hoc categorization provided several measures of the pilots’ responses to the TCAS advisory, as shown in Figure 2. Compliance at any instant to a TCAS RA was defined as, after allowing for a five second reaction time, the pilot meeting (or exceeding) the vertical rate displayed by TCAS and the pilot not performing any horizontal maneuver (defined when the aircraft’s roll angle exceeded 5 degrees).
3.1 Compliance and Maneuvering in Response to RAs
Examining compliance percentage, Table 2, neither run order, nor ATC interaction, nor traffic density had significant effects. Between-pilot variance was significant. As shown in Figure 3, significant differences were found between the four RA types created in the experiment “(Corrective) Climb,” “(Corrective) Descend,” “Crossing Descend,” and “Preventive” RA’s; significant differences were found between these RA types. (Although a Levene test suggested that the data violates the assumption of homogeneity of variances, robust tests of equality of means confirmed significant differences, p
