Document
RUNWAY INCURSION PREVENTION FOR GENERAL AVIATION
OPERATIONS
Denise R. Jones and Lawrence J. Prinzel, III National Aeronautics and Space Administration, Hampton, VA of technology, infrastructure, procedural, and training interventions [1]. None of these initiatives
Abstract
involve technology solutions for the aircraft.
A Runway Incursion Prevention System (RIPS) and additional incursion detection algorithm The National Transportation Safety Board were adapted for general aviation operations and (NTSB) also considers runway incursions to be a evaluated in a simulation study at the National serious aviation safety hazard, listing runway Aeronautics and Space Administration (NASA) incursion prevention as one of their “most wanted” Langley Research Center (LaRC) in the fall of transportation safety improvements [2]. The NTSB 2005. RIPS has been designed to enhance surface specifically recommends that the FAA implement situation awareness and provide cockpit alerts of technology that “provides a direct warning potential runway conflicts in order to prevent capability to flight crews” [3].
runway incidents while also improving operational NASA developed a Runway Incursion capability. The purpose of the study was to Prevention System (RIPS) for commercial aviation evaluate the airborne incursion detection algorithms operations to improve airport safety by providing and associated alerting and airport surface display supplemental surface situation awareness concepts for general aviation operations. This information and guidance cues, and alerts of paper gives an overview of the system, simulation runway conflicts and route deviations directly to the study, and test results.
flight crew. The system was evaluated in several flight tests and simulation studies [4][5][6].
Introduction
RIPS was adapted for GA operations and Runway incursions are a serious aviation integrated with NASA’s GA synthetic vision safety hazard, particularly for general aviation (GA) system. A piloted simulation study was conducted operations. According to the Federal Aviation at NASA LaRC to evaluate RIPS for GA operations Administration (FAA) [1], during the four year focusing on analysis of the incursion detection period from fiscal year (FY) 2001 through FY algorithms and display concepts. This paper will 2004, there were approximately 257 million aircraft present an overview of the system, description of operations and 1,395 runway incursions reported at the simulation study, and reporting of test results.
United States towered airports – approximately 5.4 runway incursions for every one million operations.
System Description
General aviation accounted for 74 percent of these incursions but only 57 percent of the operations.
Seventy-six percent of the most severe incursions Simulation Facility (114 of 150 incursions) involved at least one GA Flight Deck Simulator aircraft. Five incursions resulted in collisions, with NASA LaRC’s Integration Flight Deck (IFD) four of these collisions involving GA aircraft.
transport category fixed-based high-fidelity flight These statistics do not consider incidents that occur simulator was used for this study because of its at non-towered airports.
visual, tactile, and audio capabilities. The IFD was adapted to emulate a Cessna 206 (C-206) GA Current FAA initiatives are targeting aircraft (herein referred to as the ownship). A basic reductions in the severity, number, and rate of six degree of freedom non-linear simulation model runway incursions by implementing a combination of the C-206 and representative control force and Research Displays braking models were used. In order to avoid This simulation study was geared for low end distractions, extraneous displays and gauges were GA aircraft; therefore, standard round dial covered or turned off during data collection. This instrumentation was used in conjunction with configuration was used successfully in a previous various airport surface map formats and alerting.
GA simulation [7].
The round dial displays were shown on the RD located in front of the evaluation pilot (EP) (see As shown in Figure 1, an electronic research Flight Deck Simulator section). When required, an display (RD) was installed on the instrument panel airport surface map was displayed on the EFB directly in front of the left seat and control yoke.
located to the left of the round dial display. The The RD was composed of two 10.4” liquid crystal surface map was generated using a Reno displays and was capable of displaying two separate International (RNO) airport geographic database digital displays, side-by-side. An electronic flight developed to RTCA standards [8]. The map scale bag (EFB) display, located to the left of the RD, was set to 2.5 nm for the airborne scenarios and 1.5 was used to present the airport surface map display nm for the ground based scenarios. Audible alerts concepts described below. This display was 10.4” sounded through the flight deck speaker system.
diagonal with a resolution of 1280 x 1024 pixels.
The collimated out-the-window scene provided a Seven display conditions were evaluated 200 degree horizontal by 40 degree vertical field- during the course of the study as described below.
of-view at 26 pixels per degree resolution.
Baseline (B) Traffic position data was “broadcast” at a 1 Hz The Baseline (B) display condition consisted rate. No additional latency in traffic position was of a set of six GA instruments (airspeed, attitude, used. Ownship position data was updated at a 20 altitude, vertical speed, directional gyro, and turn Hz rate. Positional error was not introduced into coordinator) plus manifold pressure and instrument these data.
landing system data. All instruments were three inches in diameter and configured on the RD as shown in Figure 2.
Figure 2. Baseline Display Condition Figure 1. IFD Flight Deck Baseline with Surface Map and Ownship (BMO) Air Traffic Control Simulation The BMO display condition consisted of the An air traffic control (ATC) simulation was baseline round dials displayed on the RD with the created to establish a realistic terminal area addition of a plan view surface map displayed on environment. Approach and tower ATC the EFB. This version of the surface map displayed instructions and traffic requests and replies were an airport layout along with ownship position.
prerecorded using different voices. The messages Traffic was not shown. Incursion alerting was not were then played through the flight deck speaker part of this condition.
system when the ownship and simulated traffic reached specified locations.
Baseline with Surface Map and Ownship Baseline with Perspective Surface Map and Traffic (BMOT) (BRIPS) The BMOT display condition was equivalent The BRIPS display condition consisted of the to the BMO condition with the addition of traffic on baseline round dials displayed on the RD with a the surface map (Figure 3). Traffic was indicated by perspective surface map displayed on the EFB. The dark blue chevrons when on the ground and cyan map showed a perspective track-up airport layout chevrons when airborne. Incursion alerting was with ownship and traffic locations, ATC not part of this condition. instructions (including the approved taxi route and hold short locations), and incursion alerts (Figure 4). Audible incursion alerts were also sounded.
Audible route deviation and crossing hold alerts were also generated. Route deviation alerts were generated if ownship left its assigned path during taxi. Crossing hold alerts were generated if ownship crossed a hold line without clearance.
Figure 3. Plan View Surface Map Baseline with Audible Incursion Alerts (BA) The BA display condition consisted of the baseline round dials displayed on the RD and audible runway incursion alerts sounded in the flight deck when a potential conflict was detected.
Figure 4. Perspective Surface Map Baseline with Surface Map and Ownship and Audible Incursion Alerts (BAMO) The BAMO display condition was equivalent Runway Incursion Alerting to the BMO condition with the addition of audible The runway was monitored for potential runway incursion alerts.
incursions any time the ownship was to enter the Baseline with Surface Map and Ownship, runway - during final approach and landing, take- Traffic, and Incursion Alerts (BAMOT) off roll, and taxi crossing. If an incursion was The BAMOT display condition was equivalent predicted, audible and graphical alerts were to the BMOT condition with the addition of both generated and presented to the flight crew. These audible and graphical runway incursion alerts. The aircraft-generated alerts could also be data linked to alert implementation is described in detail below in ATC so the pilots and controllers have the same the Runway Incursion Alerting section.
awareness. Currently, conflict resolution advisories are not provided.
Two different incursion detection algorithms textual forms of these alerts were presented on the were evaluated during the simulation study. surface map. Also, the traffic symbol representing the incurring aircraft was enlarged, changed color The Runway Safety Monitor (RSM) incursion (yellow for Caution and red for Warning), and was detection algorithm [9] uses a generic approach for highlighted by a target designator box. The detecting and generating incursion alerts and is not identification tag was also highlighted. In the event designed to detect only specific incursion scenarios.
the incurring traffic symbol was not shown because The RSM monitors traffic that enters a three- of the display scale, a symbol was pegged on the dimensional virtual protection zone around the edge of the display in the direction of the traffic on runway that is being used by the ownship.
the perspective surface map only. The distance to Incursion detection is based on the operational state conflict was shown beneath the ownship symbol.
of the ownship and traffic, as well as other criteria Figure 4 shows and example of a Warning alert.
(separation and closure rate), to avoid false alerts.
Identification, position, and altitude data is used to track the traffic in the protection zone. Traffic data
Test Method
projections are calculated within RSM since, from Data collection occurred for several different flight test experience, reliable position updates are scenarios and test conditions as described below.
not received at consistent intervals. RSM generates a Warning alert, which occurs when a runway incursion is detected and evasive action is required Test Scenarios to avoid a potential collision. Information provided Five incursion scenarios were evaluated during with each alert includes identification of the the simulation. The EP was trained to abort if a incurring traffic and separation distance to potential warning alert was given during departure, go- conflict. RSM was developed for NASA by around if a warning alert was given on approach, Lockheed Martin.
and stop if a warning alert was given during taxi.
The EP was not required to take evasive action The PathProx™ detection algorithm [10] is when a caution alert was issued.
designed to handle over 40 specific runway incursion scenarios. Alerts are issued based on the A traffic pattern was established to emulate a states of the ownship and traffic and on conditions representative GA traffic flow at RNO. Alternating including position, speed, and track angle.
arrivals and departures were on Runway 34R with PathProx™ generates two types of alerts analogous interleaving departures on Runway 25. Traffic to the Traffic Alert and Collision Avoidance traveled through the intersection of runways 25 and System (TCAS) approach. A Caution alert informs 34R every minute. The incurring traffic was the flight crew of a potential incursion or an interleaved into this traffic flow.
incursion where the conflict does not yet require Every effort was made to produce similar evasive action. The crew can take evasive action, timing for the scenarios; however, a certain amount however, at their discretion. PathProx™ also of variability was naturally introduced due to the generates Warning alerts when immediate evasive maneuvering conducted by the EP (i.e., approach action is required. Information provided with each speed, taxi speed, etc.).
alert includes identification of the incurring traffic, the associated runway, and separation distance Scenario 1 – arrival/take-off hold between the traffic and ownship. PathProx™ was Scenario 1 tested the incursion situation where developed by Rannoch Corporation.
an aircraft was in position and holding for departure clearance while another aircraft was approaching For this test, the incursion alerts were the same runway for a landing (Figure 5). The presented to the flight crew both visually on the ownship was approaching Runway 34R, 3 nm from surface map and audibly, if required by the current the threshold at 1010’ above field level (AFL) and display condition. The alert phrases were designed at an indicated airspeed of 90 kts. The EP was to provide descriptive information regarding the instructed to land. The traffic was initially at the location of the incurring traffic, e.g. “Warning, 34R hold line nearest to the runway threshold. The Traffic Departing 25”, “Caution, Traffic 34R”. The traffic was cleared to taxi into position on the active Once the ownship began its departure (on runway runway when the ownship was 2 nm from the heading and traveling greater than 10 kts), the threshold and held in that location awaiting traffic taxied across the hold line and entered departure clearance. Runway 34R without clearance.
Scenario 3 – arrival/departure Scenario 3 was a crossing runway scenario with one aircraft landing at the same time another aircraft was departing on an intersecting runway.
The ownship was approaching Runway 34R, 3nm from the threshold at 1010’ AFL and at an indicated airspeed of 90 kts. The EP was instructed to land and stop on the runway. The traffic was initially on Taxiway L near the Runway 25 hold line at the threshold. The traffic was cleared to taxi into position on Runway 25 while the ownship was on approach to Runway 34R. When the ownship crossed the threshold, the traffic began its departure on Runway 25 without receiving clearance.
Scenario 4 – departure/departure Scenario 4 was another crossing runway scenario where two aircraft were departing simultaneously on crossing runways. The ownship was initially on Taxiway C approximately 500’ from the Runway 34R threshold traveling at 8 kts.
The traffic was initially on Taxiway L near the Runway 25 hold line at the threshold. The traffic was cleared into position on Runway 25 while the ownship taxied on C. The EP was cleared for departure on Runway 34R. Once the ownship began its departure (on runway heading and traveling greater than 10 kts), the traffic began its departure without receiving clearance.
Scenario 5 – taxi crossing/departure Scenario 5 was designed to evaluate the taxi crossing situation. The ownship began on the ramp at the Mercury Air Center facing Taxiway L. The Figure 5. Scenario 1 Configuration traffic began on Taxiway L at the Runway 25 hold line at the runway threshold. The traffic was cleared to taxi into position while the ownship Scenario 2 – departure/intersection departure taxied out of the ramp. The ownship was cleared to Scenario 2 tested the case where one aircraft taxi to Runway 34R, with instructions to cross Runway 25 (without stopping). When the ownship was departing as another aircraft entered the runway for an intersection departure. The ownship crossed the centerline of Taxiway L, the traffic was initially on Taxiway C approximately 500’ began its departure without receiving clearance.
from the Runway 34R threshold traveling at 8 kts.
The traffic was located at the 34R hold line on Evaluation Pilots Taxiway L for an intersection departure. (Taxiway Sixteen GA pilots served as participants in the L is approximately 3000’ from Runway 34R experiment with an equal distribution of flying threshold.) The ownship was cleared for departure.
experience used to represent a cross-section of the Procedure Part 91 pilot population: low-time (<400 hours) Each EP participated in an extensive briefing visual flight rules (VFR), high-time (>400 hours) and training session prior to data collection that was VFR, low-time (<1000 hours) instrument-rated, and principally designed to mask the focus of the high-time (>2000 hours) instrument-rated.
experiment (runway incursion prevention) for the rare event testing. The EP was trained on the incursion alerting system prior to the rare event Test Matrix testing only if the display condition evaluated The testing was conducted in two phases. A included alerting; otherwise, the training was “rare event” study evaluated pilot reaction to a conducted before the usability study. Before each runway incursion event with a given display run, the pilots were briefed on the run conditions, condition and incursion detection algorithm without e.g. approach or departure, visibility, alerting expectation on the part of the subject as to the true system selected (for usability study), and displays intent of the study. Four display conditions (BMO, available. Post-run, post-block, and post-test BMOT, BAMO, BAMOT) were evaluated across questionnaires were administered when required.
subjects (each EP was given only one display Audio, video, and digital data were also recorded.
condition). Eighteen runs were randomly flown that consisted of six different approaches in varying day time visual meteorological conditions (VMC)
Results
and instrument meteorological conditions (IMC) A summary of quantitative and qualitative visibility conditions. The intent of these runs was results is presented for the rare event testing and to provide sufficient variety and task demands to usability study. All data is referenced from the hide the true focus of the study. The final run in the center of gravity of the aircraft.
block was the runway incursion event (Scenario 1 above) conducted in VMC conditions (3 nm visibility and 1000’ ceiling). Scenario 1 – the Rare Event Testing arrival/take-off hold incursion – was selected As described above, each EP received 18 because of its prevalence in runway incursion various approach tasks before being presented with incident and accident statistics. A wind profile was the runway incursion event (Scenario 1).
introduced to add workload to the tasks.
The FAA runway incursion severity rating [1], Following the rare event study phase, a described below, was used to categorize the runway usability study evaluated the display concepts incursion incident data from this study.
effectiveness for runway incursion prevention.
• Category A – Separation decreases, extreme Each EP evaluated all five incursion scenarios with action taken to narrowly avoid collision, or the seven display conditions. The first run for each collision occurs; scenario group used the baseline display condition.
• Category B – Separation decreases, All of the runs that provided alerting used the RSM significant potential for collision; incursion detection algorithm as the alert source. In • Category C – Separation decreases, ample this study phase, the subjects had an expectation for time and distance to avoid collision; the study intent (i.e., runway incursions) but they • Category D – Little or no chance of collision did not know the scenario before the initial run.
Another purpose of the usability study was to but meets definition of runway incursion.
evaluate the RSM and PathProx™ incursion Fourteen scenarios resulted in the less detection algorithms for GA operations. Each EP evaluated both algorithms using all five incursion hazardous Category C and D incursions, one scenarios and the BRIPS display condition. All resulted in a Category A incursion, and one resulted runs in the usability study were conducted in VMC in a Category B incursion.
conditions (3 nm visibility and 1000’ ceiling) The 14 less hazardous Category C and D without winds.
incursions were mitigated by the EPs by conducting a go-around and gaining separation from the traffic.
The Category A runway incursion occurred between the PathProx™ Warning alert and RSM with the EP flying the BMOT display concept. Warning alert.
Despite the traffic indications on the surface map Of the 16 runs, eight EPs initially became and out-the-window visuals, the EP demonstrated aware of the incursion traffic before the alert would no awareness of the runway traffic and over-flew have occurred by viewing out the window; five the traffic and landed.
visually acquired the traffic out the window after an The Category B incident occurred when the EP incursion alert, if used, would have occurred; one over-flew the runway traffic (at 146’ AFL) before saw the traffic on the surface map well before the conducting a go-around. The EP was aware of the alert occurred; and two EPs did not see the traffic at incursion after having received an audible alert all (the category A and B incursions described (BAMO display concept) but continued to descend above). As shown in Figure 6, the incursion traffic to visually acquire the traffic to confirm the alert. was acquired sooner when the EP was provided This incident would have been classified as a with a traffic display on the surface map and/or Category D incursion if the EP had initiated the go- incursion alerts, but the differences were not around at first awareness of the alert. statistically significant at the α = 0.05 level. The data for the two EPs that did not acquire the traffic No statistically significant differences were and the two EPs that headed to the incorrect runway found between the display concepts for the distance were omitted from Figure 6.
to the incurring traffic when a go-around was initiated by the pilots or for the EP’s reaction time 6000 35 from the incursion event occurrence (based on a Multivariate Analysis of Variance (MANOVA) test p > 0.05).
Data on both incursion detection algorithms 15 Feet Seconds were collected during the incursion events; 10 1000 Distance to Traffic however, only the alerts generated by the RSM 5 Time to Traffic algorithm were displayed to the EP, when the 0 0 display condition included alerting. For all incursion runs, the RSM issued an alert when the BMO BMOT BAMO ownship was an average distance of 4060’ and BAMOT approximately 24 seconds from the traffic.
PathProx™ computed that a Caution alert was Figure 6. Initial Traffic Awareness necessary when ownship was 5836’ and approximately 35 seconds from the traffic and a For those displays that had alerting (BAMO, Warning alert was necessary when ownship was BAMOT), there were no significant differences in 4234’ and approximately 25 seconds from the timeliness of the alerting in terms of being able to traffic. (Data from two EPs were omitted from take evasive action. However, when pilots were these averages because the pilots were initially asked to rate all four display concepts on the heading to the wrong runway and turned toward perceived efficacy of the alerts (F(3,16) = 10.948) Runway 34R within 1.1 nm. Even though alerts and the additional safety value added (F(3,16) = were generated the results were skewed.) A 8.814) an ANOVA revealed a significant effect MANOVA showed a significant effect between the between the displays (p < 0.01). Subsequent post- incursion detection algorithms tested (F(4,42) = hoc Student Newman Kuels tests showed that pilots 190.163, p< .0001). A subsequent ANOVA on the reported that the BMO display condition was dependent variables revealed that the PathProx™ significantly poorer than the other three display Caution alert was generated significantly earlier conditions which were not significantly different (F(2,42) = 6.839) and at a greater distance from the from each other.
incursion aircraft (F(2,42) = 7.302) than from either the PathProx™ Warning alert or the RSM Warning alert. There were no significant differences occurred before the PathProx™ alerts were Usability Study generated. Detailed PathProx™ analysis was not All test runs conducted during the usability possible because specific alerting criteria is study included incursion events. During each test proprietary.
run, data were simultaneously collected on the performance of both incursion detection algorithms; Scenario 1 arrival/take-off hold results – For however, only one method was chosen for display.
the Baseline display (B) condition, all EPs acquired the incursion traffic out the window for Scenario 1 The EPs evaluated all five incursion scenarios when approximately 1 nm from the runway with the seven display conditions to determine the threshold. As a result, the average go-around effectiveness of the display for runway incursion initiation point was 443’ AFL and 5240’ prevention. The RSM was used as the alert source (approximately 28 seconds) from the traffic. The for this testing phase. Although the EPs were aware EP initiated a go-around after the PathProx™ that each run would contain an incursion event, they Caution alert would have occurred on average (475’ were not told the type of incursion before the initial AFL and 5641’ – approximately 34 seconds – from run for each scenario. The initial run evaluated the the traffic) but before either the RSM (409’ AFL baseline display condition. This was done to and 4272’ – approximately 24 seconds – from the determine if the EP would visually acquire the traffic) or PathProx™ (392’ AFL and 4117’ - incursion traffic out the window since the surface approximately 25 seconds – from the traffic) map and alerting were not available.
Warning alert would have occurred.
The EPs also evaluated the RSM and Post-run data analysis on all Scenario 1 test PathProx™ incursion detection algorithms for GA runs during the usability study showed that the operations using the five incursion scenarios and PathProx™ Caution alert was generated when the the BRIPS display condition.
ownship was at approximately 608’ AFL and 5756’ Quantitative Results (approximately 36 seconds) from the incurring A total of 612 test runs were completed. Since traffic. Both the RSM (488’ AFL and 4121’ – the BMO and BAMO display conditions provided approximately 25 seconds - to traffic) and similar alerting information to the B and BA display PathProx™ (492’ AFL and 4173’ - approximately conditions, a limited number of BMO and BAMO 25 seconds - to traffic) Warning alerts were test runs were conducted in the interest of time.
generated at essentially the same moment. It should Data was not analyzed for 16 runs due to missing be noted that during the test runs in which alerts data files, yielding a total of 596 runs. Alerts were were provided, the EP was asked to continue the required for display on 432 test runs. During these maneuver until the alert was received for evaluation 432 runs, the RSM was chosen as the alert source purposes.
80 percent of the time (347 runs).
Scenario 2 departure/intersection departure RSM generated alerts on 533 runs. RSM did results - Six EPs (37.5 percent) did not acquire the not alert on 60 runs due to the maneuver taken by incursion traffic visually for Scenario 2 or saw the the EP. For example, the EP may have acquired the traffic too late to abort the departure and actually traffic out the window and conducted a go-around over-flew the runway traffic when using the before the alerting criteria were met. RSM did not Baseline (B) display condition.
alert on three runs due to the scenario timing (no Figure 7 shows when the EP began to abort the incursion event). For the PathProx™ algorithm, departure to avoid a conflict with the incursion Caution alerts were only possible on 243 runs and traffic. A greater safety margin resulted when alerts of these, alerts were generated on 157 runs. Of the were provided (BA, BAMO, BAMOT, and BRIPS).
possible 596 test runs, PathProx™ generated The EP also aborted the departure for all of these warning alerts on 352 runs. The 244 runs in which runs. The departure was aborted later when alerting PathProx™ did not alert were generally due to the was not provided (B, BMO, and BMOT), although maneuvering performed by the EP. For instance, still with enough time to stop prior to reaching the when RSM alerting was provided, the pilot traffic. Also, as discussed above, the EP actually executed an aircraft maneuver and this generally took off on six runs and over-flew the traffic, using Scenario 4 departure/departure results – Five the Baseline display. EPs (31.3 percent) saw the incursion traffic visually and aborted the departure when using the Baseline For this departure scenario, the RSM Warning (B) display condition. Nine EPs (62.5 percent) did alert (ownship at 28 kts ground speed and 2638’ - not see the incursion traffic visually or saw the approximately 21 seconds - from Taxiway L) was traffic too late to abort the departure. For these generated before and at a lower ground speed than runs, the ownship came within an average distance the PathProx™ Warning alert (41 kts ground speed of 458’ (range from 179’ to 795’) from the and 2452’ - approximately 19 seconds - from incursion traffic. The timing was early for two of Taxiway L); however, both provided ample time to the runs resulting in no incursion event; therefore, abort and stop before reaching the incurring traffic.
these data are not included in the analysis.
PathProx™ Caution alerts are not generated on departure. Generally, a greater safety margin resulted when alerts were provided (BA, BAMO, BAMOT, and BRIPS) as shown in Figure 8. For runs in 3000 20 which the EP acquired traffic, the departure was aborted later when alerting was not provided (B and BMOT); however, with still enough time to stop before the crossing runway.
1500 10 Feet Seconds 6 Distance to L 2500 18 Time to L 0 0 B BA BMO BMOT BAMO BRIPS BAMOT 8 Feet Seconds Distance to 25 Figure 7. Scenario 2 Abort Initiation Time to 25 0 0 Scenario 3 arrival/departure results – For the B Baseline (B) condition, 13 EPs (81 percent) did not BA BMO acquire the incursion traffic visually for Scenario 3.
BMOT BAMO BRIPS BAMOT Due to the scenario design; however, 14 EPs landed and stopped before reaching the crossing runway.
Figure 8. Scenario 4 Abort Initiation One EP landed but taxied through the intersection as the traffic was departing from crossing Runway 25. One EP conducted a go-around.
The RSM Warning alerts were generated before the PathProx™ Warning alerts for this On average, the RSM Warning alert (1303’ - departure scenario. When RSM alerts were approximately 11 seconds - to Runway 25) was presented, the EP typically reacted based on those generated slightly before the PathProx™ Warning alerts; therefore, PathProx™ alerts were sometimes alert (1181’ - approximately 9 seconds - to Runway not generated or generated after the abort maneuver.
25). Both algorithms alerted at or slightly before For an accurate assessment of PathProx™ touchdown. PathProx™ Caution alerts were not performance, only the results from the runs in generated for this scenario.
which the PathProx™ alerts were shown to the EP For all Scenario 3 runs (124 total), a go-around will be presented. For these runs, the PathProx™ was conducted just before touchdown (7’ AFL) on Warning alerts were generated when the ownship only 11 percent (14) of the runs.
was going 48 kts ground speed and was 1891’ and approximately 15 seconds from Runway 25. The RSM Warning alerts were generated when the ownship was going 29 kts ground speed and was Table 1. Scenario 5 RSM Alert Generation 2216’ and approximately 18 seconds from Runway EP Ground Distance to 25. Again, PathProx™ Caution alerts are not Speed (kts) hold line (feet) generated on departure.
1 10.7 -86 Scenario 5 taxi crossing/departure results - 1 9.7 -59 Fourteen EPs acquired the incursion traffic out the 2 9.7 -48 window when using the Baseline (B) condition and 3 9.2 -59 stopped before reaching Runway 25. One EP did 4 7.9 12 not see the traffic and actually crossed Runway 25 5 7 6 in front of the departing traffic. One EP saw the 6 8.4 -56 departing traffic out the window but chose to cross 7 6 7 Runway 25 anyway.
8 13.4 -17 The RSM algorithm is designed to provide 9 10.3 -70 early warning of an impending incursion during 10 8.4 -48 taxi, based on the aircraft taxi speed. If the ownship 11 8.3 -48 is traveling 8 kts or greater and is not slowing 12 7.5 17 down, the alert will be generated before the aircraft 13 7.4 14 reaches the hold line, providing sufficient distance 15 6.5 7 to stop before crossing the hold line. As the taxi 16 10.6 -8 speed increases, the alert is generated when the ownship is a farther distance from the hold line.
Qualitative The alert is not generated until after the ownship Post-run, post-block, and post-test crosses the hold line when the ownship is traveling questionnaires were administered when required.
less than 8 kts. The 8 kt threshold was used to Ratings for most of the questions were given on a prevent false or nuisance alerts as the ownship taxis scale of 1 (low) to 10 (high).
toward a hold line. (As stated earlier, details on the PathProx™ implementation are proprietary.) When asked to rate the effectiveness of the perspective surface map compared to the plan view As with scenario 4, the RSM alerts were map for prevention of runway incursions, the generated before the PathProx™ alerts for this perspective map was rated slightly more effective scenario. Therefore, for an accurate assessment of (2.9). However, the location of the surface map PathProx™, only the results from runs in which the was suboptimal and should be positioned closer to PathProx™ alerts were given to the EP will be the pilots head-up field-of-view. The traffic presented. For these runs, the RSM Warning alerts presentation on the surface map was considered were generated when the ownship was an average easily discernable (8.4). Most EPs (14 of 16) distance of 27’ before the hold line and 197’ from considered traffic presentation necessary to prevent the edge of Runway 25. Table 1 shows that the runway incursions. The addition of traffic would RSM alert was generated before reaching the hold provide increased (8.6) situation awareness (SA) line (negative value) when the taxi speed was over a surface map with only ownship location.
greater than 8 kts. A positive value indicates the Over half of the EPs (10 of 16) indicated graphical aircraft crossed over the hold line. The RSM presentation of alerts on the surface map was alerting is proactive in preventing an incursion necessary to prevent runway incursions. The (crossing the hold line) in this situation.
addition of graphical alerts would provide increased PathProx™ Caution alerts were generated when the (6.3) SA. The EPs considered the terms used for ownship was an average of 18’ past the hold line the incursion alerts (e.g. “Warning, Traffic and 152’ from the runway edge. PathProx™ Departing 25”) to be very effective (8.9 average Warning alerts were generated when the ownship rating for all phrases).
was 41’ past the hold line and 129’ from the runway edge.
For display conditions with alerting available, the EPs indicated that the incursion event would most likely be brought to their attention first airport surface (during departure and taxi). This through audible alerting, then on the surface map, was particularly true for Scenario 5. Many EPs and lastly out the window. Nine of 16 EPs indicated that the short caution alert had no benefit.
indicated that an audible alert alone would provide The majority of EPs would like to be provided a minimally effective incursion prevention display, with maneuver guidance for conflict resolution in while five of 16 indicated a surface map with addition to runway incursion alerting on final ownship and traffic would be minimally effective.
approach (12 of 16) and when taxiing across a However, all 16 EPs indicated a surface map with runway (nine of 16). Half of the EPs would like ownship and traffic in conjunction with an audible maneuver guidance on departure.
alert would be an optimal incursion prevention display. Table 2 shows, according to averaged EP In general, the EPs felt safer during runway incursion incidents when alerting was provided ratings, that for all alerting display conditions, the (B=2.1, BMO=2.75, BMOT= 5.4, BA=7.2, caution and warning alerting system provided the greatest amount of runway incursion awareness. BAMO=7.4, BAMOT=9.3, and BRIPS=9.2).
Analysis of pilot responses to their perceived safety An ANOVA confirmed this conclusion with significant main effects (p < 0.0001) for display on and runway incursion prevention effectiveness support this conclusion with significant main effects pilot ratings of situation awareness of where they found (i.e. for perceived safety (F(6,90) = 857.390) were located (F (6,90) = 143.956) and where other traffic and hazards were located (F(6,90) = 94.899). and runway incursion prevention effectiveness (F(6, 90) = 188.793)). The addition of traffic was Table 2. Alerting Preference for Incursion marginally beneficial when presented on a moving Awareness and Resulting SA Improvement map display and was only effective when alerting was provided. A possible cause may be that pilots Alerting Warning only Caution & Warning had to transition to out-the-window and were not Type # of EP SA # of EP SA focused on the head-down display. With alerting, the pilot is provided a cue to direct focus and BA 6 6.5 10 4.2 attention to the head-down display to locate the BAMO 6 6.7 10 4.3 incurring traffic. In fact, pilots rated having audible alerts (BA) and having alerts with a map with BAMOT 7 7.7 9 5 ownship but no traffic (BAMO) similarly for runway incursion prevention on almost all BRIPS 7 7.6 9 5.2 dependent variables measured. For the experimental scenarios tested, the moving map display revealed its utility only when traffic AND All EPs indicated that both the RSM and alerting were options.
PathProx™ alerting provided sufficient time to avoid a potential incursion conflict. Only six of 16 EPs thought providing both caution and warning
Summary
alerts was more effective in preventing runway A Runway Incursion Prevention System incidents than a single warning alert. However, (RIPS) adapted for general aviation operations was nine of 16 EPs indicated that the caution and evaluated in a simulation study in the fall of 2005.
warning system provided greater (4.4) situation The purpose of the study was to evaluate the awareness and provided more reaction time (even airborne incursion detection algorithms and though this was not always the case). Eleven of 16 associated alerting and airport surface display EPs liked the idea of having a caution alert in concepts for general aviation operations using a rare conjunction with a warning to provide more event study followed by a usability study.
evaluation and reaction time, i.e. a greater comfort level. For the scenarios evaluated, the EPs The results indicate that, during the rare event generally felt that providing caution and warning testing, most pilots were able to acquire the alerts on approach was most effective, while a incurring traffic looking out the cockpit windows warning alert alone was sufficient when on the (in VMC conditions), even before incursion alerting was activated. In the usability study, varying out- Demonstration and Testing at the Dallas/Fort th the-window detection of incurring traffic was noted Worth International Airport , Proceedings of the 20 and the importance of incursion detection Digital Avionics Systems Conference.
algorithms was shown.
[5] Jones, Denise R., 2002, Runway Incursion The surface map with ownship (without other Prevention System Simulation Evaluation , st traffic) was rated as being significantly inferior to a Proceedings of the AIAA/IEEE 21 Digital surface map with traffic and/or incursion alerting Avionics Systems Conference.
for perceived safety value added. However, one [6] Jones, Denise R., 2005, Runway Incursion pilot experienced a severe runway incursion risk for Prevention System Testing at the Wallops Flight collision despite having traffic displayed on a Facility , Proceedings of the SPIE Defense & surface map. The addition of traffic was marginally Security Symposium.
beneficial when presented on a surface map display [7] Bartolone, Anthony P., L. J. Glaab, M. F.
and was only effective when alerting was provided.
A possible cause may be that pilots had transitioned Hughes, R. V. Parrish, 2005, Initial Development of a Metric to Describe the Level of Safety Associated to out-the-window and were not focused on the head-down display to locate the incurring traffic. with Piloting an Aircraft with Synthetic Vision Systems (SVS) Displays , Proceedings of the SPIE Pilots reported that the utility of the surface map Defense & Security Symposium.
would be significantly more effective if located higher on the instrument panel closer to the pilot’s [8] RTCA SC-193, 2001, User Requirements for head-up field-of-view.
Aerodrome Mapping Information , RTCA/DO-272.
The results generally matches past research on [9] Green, David F., February 2006, Runway Safety commercial operations that the incursion alerts Monitor Algorithm for Single and Crossing Runway provided sufficient time to avoid a potential Incursion Detection and Alerting , NASA CR-2006- incursion conflict. Departures were generally 214275.
aborted sooner when alerts were provided, resulting [10] Cassell, Rick, C. Evers, J. Esche, 2003, Safety in greater safety margins. A surface map with Benefits of PathProx – A Runway Incursion ownship and traffic along with audible alerts was Alerting System , Proceedings of the AIAA/IEEE considered an optimal incursion prevention display, nd 22 Digital Avionics Systems Conference.
while an audible alert alone was considered a minimally effective display. Over half of the pilots would like maneuver guidance for conflict th resolution in conjunction with incursion alerting. In 25 Digital Avionics Systems Conference general, the pilots felt substantially safer during runway incursion incidents with alerting onboard. October 15, 2006
References
[1] FAA Office of Runway Safety, August 2005, FAA Runway Safety Report.
[2] National Transportation Safety Board, November 2005, Most Wanted Transportation Safety Improvements , www.ntsb.gov/recs/mostwanted/index.htm [3] National Transportation Safety Board, July 2000, Safety Recommendation, Letter to the FAA Administrator , A-00-66.
[4] Jones, Denise R., C. C. Quach, S. D. Young, 2001, Runway Incursion Prevention System –