Scholarly article on topic 'An experimental analysis of situation awareness for cockpit display interface evaluation based on flight simulation'

An experimental analysis of situation awareness for cockpit display interface evaluation based on flight simulation Academic research paper on "Medical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Chinese Journal of Aeronautics
OECD Field of science
Keywords
{"Cockpit display interface" / "Design evaluation" / "Flight simulation" / "Heart rate" / SAGAT / "Situation awareness"}

Abstract of research paper on Medical engineering, author of scientific article — Hengyang Wei, Damin Zhuang, Xiaoru Wanyan, Qun Wang

Abstract Aircraft cockpit display interface (CDI) is one of the most important human–machine interfaces for information perceiving. During the process of aircraft design, situation awareness (SA) is frequently considered to improve the design, as the CDI must provide enough SA for the pilot to maintain the flight safety. In order to study the SA in the pilot-aircraft system, a cockpit flight simulation environment is built up, which includes a virtual instrument panel, a flight visual display and the corresponding control system. Based on the simulation environment, a human-in-the-loop experiment is designed to measure the SA by the situation awareness global assessment technique (SAGAT). Through the experiment, the SA degrees and heart rate (HR) data of the subjects are obtained, and the SA levels under different CDI designs are analyzed. The results show that analyzing the SA can serve as an objective way to evaluate the design of CDI, which could be proved from the consistent HR data. With this method, evaluations of the CDI design are performed in the experimental flight simulation environment, and optimizations could be guided through the analysis.

Academic research paper on topic "An experimental analysis of situation awareness for cockpit display interface evaluation based on flight simulation"

Chinese Journal of Aeronautics, 2013,26(4): 884-889

JOURNAL OF

AERONAUTICS

Chinese Society of Aeronautics and Astronautics & Beihang University

Chinese Journal of Aeronautics

cja@buaa.edu.cn www.sciencedirect.com

An experimental analysis of situation awareness for cockpit display interface evaluation based on flight simulation

Wei Hengyang, Zhuang Damin *, Wanyan Xiaoru, Wang Qun

School of Aeronautic Science and Engineering, Beihang University, Beijing 100191, China

Received 30 March 2012; revised 7 June 2012; accepted 18 July 2012 Available online 30 April 2013

KEYWORDS

Cockpit display interface; Design evaluation; Flight simulation; Heart rate; SAGAT;

Situation awareness

Abstract Aircraft cockpit display interface (CDI) is one of the most important human-machine interfaces for information perceiving. During the process of aircraft design, situation awareness (SA) is frequently considered to improve the design, as the CDI must provide enough SA for the pilot to maintain the flight safety. In order to study the SA in the pilot-aircraft system, a cockpit flight simulation environment is built up, which includes a virtual instrument panel, a flight visual display and the corresponding control system. Based on the simulation environment, a human-in-the-loop experiment is designed to measure the SA by the situation awareness global assessment technique (SAGAT). Through the experiment, the SA degrees and heart rate (HR) data of the subjects are obtained, and the SA levels under different CDI designs are analyzed. The results show that analyzing the SA can serve as an objective way to evaluate the design of CDI, which could be proved from the consistent HR data. With this method, evaluations of the CDI design are performed in the experimental flight simulation environment, and optimizations could be guided through the analysis.

© 2013 Production and hosting by Elsevier Ltd. on behalf of CSAA & BUAA.

1. Introduction

The aircraft cockpit is a highly complex human-machine interaction system, and the aircraft cockpit display interface (CDI) is one of the most important devices for pilot-aircraft

* Corresponding author. Tel.: +86 10 82339026. E-mail addresses: keanu131@163.com (H. Wei), dmzhuang@buaa.e-du.cn (D. Zhuang).

Peer review under responsibility of Editorial Committee of CJA.

interaction. The pilot perceives flight information mainly from the CDI with his senses, and forms a holistic picture of the present flight environment based upon his knowledge. This process has been widely recognized as the construct of situation awareness (SA). During the process of aircraft design, SA is frequently considered to improve the design, as the CDI must provide enough SA for the pilot.1 Current researches indicate that there is a direct relationship between the pilot's SA and flight safety: the higher the SA degree is, the more effectively the pilot manipulates the flight, thus the safer the flight is. If the CDI design does not provide enough SA, the flying pilot's performance will be degraded.2 To maintain flight safety, it is necessary to evaluate the interface during the design process, and the degree of pilot's

1000-9361 © 2013 Production and hosting by Elsevier Ltd. on behalf of CSAA & BUAA. http://dx.doi.org/10.1016/j.cja.2013.04.053

SA has always been one of the important evaluation indicators.

Along with the rapid development of aeronautical techniques, aircraft instrument and its display interface are greatly improved. And new requirements for pilot SA are proposed. In order to study the SA under specific CDI design with actual flight mission, acquiring and analyzing data in the real flight environment are needed. However, due to the complexity of flight activities in the real aircraft, such studies are often expensive, dangerous, or even not feasible. The common solution for this is to simulate the flight scenarios on the ground, just as the widely use of flight simulators in the commercial pilot training. With the help of flight equations, it is possible to calculate the full envelope flight parameters. Then these data can be used to drive the virtual instrument panel and flight visual displayed for the pilot. Based on this, theoretical researches can be developed in the flight simulation environment, and the study of SA is possible through reproduced flight scenarios and mission simulations. Some valuable researches have been performed on pilot's mental workload and situation awareness in different kinds of simulators.3-5 In particular, Endsley et al. developed the situation awareness global assessment technique (SAGAT), an SA measurement method especially for the simulation environment.2 The results show that such method could be a feasible way to study SA, and it is easy to join into the aircraft design process for CDI evaluation.

Therefore, when we use the simulation method to evaluate aircraft CDI, how to obtain a comprehensive and accurate analysis of pilot's SA becomes important. In this investigation, a virtual cockpit environment for flight simulation is built up, including a virtual instrument panel, a flight visual display and the corresponding control system. In the simulation environment, a human-in-the-loop experiment based on the SAGAT is designed to measure the SA under different CDIs. Through the experiment, the SA levels and pilot heart rate (HR) data of the subject are analyzed, which could be a way to evaluate and optimize the existing design of aircraft CDI.

2. Flight simulation environment setup

The flight simulation environment consists of the virtual interface display system and the flight data control system. The virtual interface display system includes a virtual instrument panel and a flight visual display. The flight data control system is composed of a flight parameter simulation and electronic instrument driving system, as well as an experimental control and data recording system. All these systems are achieved in two DELL T1600 servers with an IBM Think Vision 22' dis-

Fig. 1 Constitution of cockpit flight simulation environment.

Fig. 2 Virtual instrument panel generation process.

play, and the data are communicated through UDP networks, as shown in Fig. 1.

In order to display flight information, a virtual instrument panel is used for simulating real cockpit instruments. Based on the data provided by the flight manual, the virtual instrument model can be developed with GL Studio from DiSTI, Microsoft Visual Studio. Net 2008 and Adobe Photoshop. This process is shown in Fig. 2.

The instrument simulation cannot get away from the actual aircraft type. Considering the current situation in civil aviation of China, the flight manuals of Boeing 737-800 from the Boeing Commercial Airplane Company, Airbus A320 from the Airbus Company and ARJ21 (''Xiang Feng'') from the Commercial Aircraft Corporation of China Ltd. are referred to generate virtual instrument panels. With appropriate simplification and abstraction, three virtual instrument panels are built up based on the research needs, as shown in Fig. 3.

The flight visual displays the virtual scene based on the MultiGen-Paradigm Vega Prime, as shown in Fig. 4.

The experimental control and data recording system is used to control the experiment of starting, suspension and termination. The SAGAT questions are displayed on the experimental interface and the answers of subjects are saved automatically in files, as shown in Fig. 5.

The flight parameter simulation and electronic instrument driving system makes use of the flight equations to calculate the 6-DOF (six degree of freedom) parameters of the aircraft in the current environment. It can completely simulate the full envelope flight, including the takeoff, climb, cruise, descending and landing. To ensure the comparability of the results, the same flight scenario of about 12 min is used in the experiment. Then the flight data is processed and converted into the avionics display data, which can directly drive the virtual instrument panel and visual display. As the above systems are running on two separated servers, data communication is via the UDP network and shared memory. The network automatically synchronizes the shared memory between the two severs, and the data are shared through the shared memory on the same sever.

3. Cockpit SA measurement method

As Ref. 6 described, SA refers to ''the perception of the elements in environment within a volume of time and space, the comprehension of their meaning and the projection of their status in the near future''. From the perspective of cognitive psychology, SA can be divided into three relative levels in the information processing model: perception of the elements

L\n . .X.. E-j

V24 12s»"

(c) ARJ21 Fig. 3 Virtual instrument panels.

in the current situation, comprehension of the current situation and projection of the future or the ability to predict what will happen next based on the current situation.

At present, the SA assessment methodology is mainly divided into four kinds: physiological measurement, memory probe measurement, performance measurement and subjective measurement. The physiological measurement has a long history in mental workload researches; however, its applications in SA are few, as there are still problems to be validated. The memory probe measurement is the most consistent with Endsley's definition of SA.6 This methodology can be divided

Fig. 4 Flight visual display.

Answer the following questions based on current flight circumstance:

What's the current phase of the flight? Takeoff Cruise Climb

Confirm Reset

Strat Esc

Fig. 5 SA experimental interface.

into three subcategories: retrospective measurement, simultaneous measurement and freeze measurement. The freeze measurement can not only solve the time problem from the retrospective measurement, but also eliminate the interference from the simultaneous measurement. Therefore, it has been widely applied in the study of SA. The performance measurement uses tasks to speculate the SA and employs the hypothesis that the better a task is completed, the higher the SA is. But some studies show that the high degree of SA does not necessarily mean good performance. The subjective measurement is a practical low-cost method, and it can be used in both simulated situations and actual tasks, but it is difficult to ensure the consistency between subjects.7

The SAGAT is a computerized memory probe measurement. First, it simulates tasks on the screen and freezes the interface at a randomly selected time. Then, with all task-related information removed, subject is asked to answer questions related to the tasks. This requires the subject to correctly assess the current environment, which directly relates to the degree of SA. Finally, the correct rate of answers can objectively reflect the current degree of the SA.

Some researchers suggested that the SAGAT may interrupt the process of recognition, and thus affect the performance of the task, and even break the process of obtaining SA. Additionally, the SAGAT requires the subject to answer the questions with all task-related information removed, which may serious rely on the short-team memory.8 However, Jones and Endsley carried out relevant researches which validated the SAGAT of an effective measurement of SA.9 Endsley also confirmed that the SAGAT is highly effective in the measurement of SA. He found that in the multi-task situation, the SAGAT

freeze causes no interruptions to the performance of the operating subject.10 In the previous studies, Endsley found that while using SAGAT, freeze has no significant negative impact on the performance of the simulation task for a long time.11 This approach is practicable in measuring the SA under different mental workloads. Therefore, this study chooses the SA-GAT to measure the SA in the flight simulation environment.

In addition, Byrne proposed the use of physiological measurement to measure SA by analyzing whether the subject received some important information in the experimental environment.12 These physiological indexes include the event related potentials (ERP), event related desynchronization (ERD), HR and skin electrodermal activity (EDA). Researches indicated that there is a certain degree of correlation between HR and SA.13 We recorded the HR of the subject during the experiment, expecting that the physiological parameters could be the evidence of the SA measurement results.

4. Experimental design of cockpit SA measurement

Thirty subjects (aged from 22-year to 28-year and on an average of 24-year, 11 females and 19 males) from Beihang University participated in the experiment. They had been trained in flight simulator for a long time. All participants have 20/20 or corrected to normal vision. There are three different CDIs for SA measurement, as shown in Fig. 3. Based on single factor randomized complete block design, each subject is involved in all of the three interface simulation tasks, and there is 1 week interval between two tasks in order to eliminate the interrupts of biological rhythms and memory. The experiments are carried out at the same time each week, and the experimental order of different interfaces is cross-balanced. The time of each experiment takes about 12 min, including the whole process

Fig. 6 SA measurement experiment.

of takeoff, climb, cruise, descent and landing. The subject is asked to closely monitor the instrument panel. At random time, the flight simulation will freeze and the instrument panel will be covered with the interface in Fig. 5. The subjects have to answer the SAGAT questions by recalling information in the panel, e.g. the airspeed, altitude, roll, pitch, and heading of the airplane. There are 10 s left for the subject to answer the question. Either answer confirmed or timeout, the interface will automatically disappear. After that, the flight simulation continues until the next freeze.14 This requires the subjects to form a complete acknowledgment of the flight from takeoff to landing. The experimental control and data recording system will automatically record the answers. A TH-P physiological tester from Beijing Tongfangshenhuo Union Technology Ltd. records the HR data of the subjects during the experiment, as shown in Fig. 6.

The SAGAT questions are obtained from the question bank restored in the experimental control and data recording system. According to the theoretical model of SA, the questions can be divided into three different levels.15 At the level of perception, information such as pitch angle, airspeed or altitude is directly asked. At the level of comprehension, questions of understanding the current environment are asked, i.e. whether the current speed or altitude is above or below the plan value. And at the level of projection, the future state is asked, such as whether the scheduled flight will climb or decline at the next moment.16 These questions are tested in preliminary experiments to guarantee the availability that they can be used for investigating the understanding of the flight information. To avoid the memory of the questions, in the formal experiment, 24 questions are randomly extracted from the bank and presented to the subject. In order to make a complete measurement of the SA, the selection of 24 questions is controlled to cover the three levels of SA. That means in the 24 questions, there is at least one question for each level of SA. After the experiment, the experimental control and data recording system will decide the corrigendum of the 24 questions. The overall correct rate can then be calculated through the percentage of correct answered questions in all the 24 questions. The correct rate of different SA levels can also be calculated, e.g. the correct rate of perception-level SA is the percentage of correctly answered perception-level questions in all the perception-level questions.

5. Experimental results

From the experiment, we can get the correct rate of subjects answering SAGAT questions. Table 1 shows the correct rates of SAGAT questions under different interfaces and SA levels. In the table, M is the mean value, and SD is the standard deviations. For different display interfaces, the overall SA is in the

Table 1 Correct rate (in percentage; with standard deviations) of SAGAT questions under different interfaces and SA levels.

Interface SA level (%)

Perception Comprehension Projection Overall

M SD M SD M SD M SD

Airbus Boeing ARJ21 75.96 83.28 79.30 17.20 11.60 16.19 88.33 8.64 84.17 11.14 93.33 4.68 89.43 92.87 88.13 12.39 12.74 18.43 78.55 84.13 82.49 15.69 11.69 16.29

Table 2 Average heart rate (per minute; with standard deviations) under different interfaces.

Interface Average heart rate per minute (%)

Airbus 76.58 1.57

Boeing 82.15 1.46

ARJ21 78.94 2.17

order of Boeing > ARJ21 > Airbus. One-way repeated measures ANOVA shows that the overall SA is significantly affected by the interface (F-test statistic F(2, 58) = 5.89, significance probability p = 0.010). Least-significant difference (LSD) post-hoc tests indicate that the overall SA in the Boeing display interface is significantly higher than the Airbus (p < 0.050). For different SA levels, the perception-level SA is in the order of Boeing > ARJ21 > Airbus, and one-way repeated measures ANOVA shows that the perception-level SA is significantly affected by the interface (F(2, 58) = 3.39, p = 0.031). LSD post-hoc tests indicate that the perception-level SA in the Boeing interface is significantly higher than the Airbus (p < 0.050). The comprehension-level SA is in the order of ARJ21 > Airbus > Boeing, and one-way repeated measures ANOVA shows that the comprehension-level SA is significantly affected by the interface (F(2, 58) = 4.58, p = 0.042). LSD post-hoc tests indicate that the comprehension-level SA in the Airbus display is significantly higher than that in the Boeing display (p < 0.050). The projection-level SA is in the order of Boeing > ARJ21 > Airbus, but ANOVA shows that the projection-level SA is not significantly affected by the interface (p > 0.050).

The physiological tester records the average heart rate under the three interfaces, which is in the order of Boeing > ARJ21 > Airbus, as shown in Table 2. One-way repeated measures ANOVA shows that the HR of the subject is significantly affected by the interface (F(2, 58) = 5.85, p = 0.015). LSD post-hoc tests indicate that the HR of the subject under Boeing interface is significantly higher than that under the Airbus display (p < 0.050).

6. Discussions

From the experimental results, we can see that the correct rates of the three interfaces are all higher than 75%, which indicates that the interface design of the three displays can guarantee the basic access to the flight information. Significant difference between the overall SA of the Boeing and Airbus interfaces could be the result of different design concepts of CDI: for the level of automation, the Airbus tends to be of a higher degree of automation, which means more works are done automatically and the pilot has smaller demand for the flight information, as a lot of important information is processed and displayed integrally; for the man-machine function allocation, the Boeing tends to assign more tasks to the pilot, requiring the subject to participate in the task more actively, which increases the workload but provides further awareness of the situation. The above influencing factors make small differences between the two display interfaces. While investigating the design of ARJ21, its correct rate is in the middle of the two displays,

which may be the fact that it has made reference to the design of two display interfaces, comprehensively surveyed their advantages and disadvantages, and designed the interface with a balanced consideration of automation and human-machine function allocation.

The above analysis can be verified by the three-level SA data under different display interfaces. First, according to the perception-level SA data, the correct rate under the Boeing display interface is significantly higher than the Airbus, which indicates that the design of Boeing display interface requires the pilot to form a higher perception of the environment. Therefore, more attention of the subject is focused on extracting this information, resulting in a higher SA at this level. Second, it is easy to see from the comprehension-level SA data that the correct rate under the Airbus display interface is significantly higher than the Boeing, which indicates that the Airbus has reduced the demand of environment perception. However, due to its integrated information display, the pilot could get an easier understanding of the flight information. Whereas for the Boeing interface, due to the request for more perception, there is a greater mental workload and the effectiveness of flight information acquisition has dropped. Finally, the projection-level SA data shows that the three interfaces have no significant differences, indicating that they are generally properly designed. The data of ARJ21 interface is at the middle in the perception-level SA but the highest in the comprehension-level SA, which probably manifests that its design has taken the problems of the pilot's work load, the level of automation and human-machine function allocation into consideration, and has made a better solution to the problem of SA decrease under high mental workload.

Physiological measurement can also be the corroboration for the above analysis. If subjects under the Boeing display interface are in the need for greater attention allocation, they must increase the perception of the flight environment. Thus they get more stress and mental workload, and their physiological parameters tend to be more active. Therefore, the subject's heart rate should be faster than that under the Airbus. This is consistent with the data and significance analysis of HR, as shown in Table 2.

7. Conclusions

(1) Based on the flight simulation, a flight environment can be fast and flexibly built at the design phase of CDI. And with SAGAT, an objective assessment of the display interface can be provided. The results of SA levels can be used for evaluating the CDI design, thus providing a reference for the aircraft design.

(2) According to the experiments based on the Boeing, Airbus and ARJ21 display interfaces, it is clear that the CDI design has a significant impact on the SA degree of the subjects. Compared with the analysis of three-level SA data, the differences in the CDI design concept can be found, and the result of physiological measurement offers the corroboration. Through the analysis of SA, evaluations of the specific interface design are available for design optimization and improvement.

Acknowledgment

This study was supported by National Basic Research Program of China (No. 2010CB734104).

References

1. Endsley MR. Toward a theory of situation awareness in dynamic systems. Hum Factors 1995;37(1):32-64.

2. Endsley MR, Selcon SJ, Hardiman TD, Croft DG. A comparative evaluation of SAGAT and SART for evaluations of situation awareness. In: Proceedings of the human factors and ergonomics society annual meeting; 1998. p. 82-6.

3. Perry CM, Sheik-Nainar MA, Segall N, Ma R, Kaber DB. Effects of physical workload on cognitive task performance and situation awareness. Theor Issues Ergon Sci 2008;9(2):95-113.

4. Pope A, Bowles RL. A pogram for assessing pilot mental state in flight simulators. AIAA-1982-257; 1982.

5. Wilson GF. An analysis of mental workload in pilots during flight using multiple psychophysiological measures. Int J Aviat Psychol 2009;12(1):3-18.

6. Endsley MR. Errors in situation assessment: implications for system design. In: Human error and system design and management. Heidelberg: Springer Berlin; 2000. p. 15-26.

7. Yang JZ, Zhang K. Situation awareness: approaches, measures and applications. Adv Psychol Sci 2004;12(6):842-50

[Chinese].

8. Gonzalez C, Wimisberg J. Situation awareness in dynamic decision making: effects of practice and working memory. J Cogn Eng Decis Making 2007;1(1):56-74.

9. Jones DG, Endsley MR. Can real-time probes provide a valid measure of situation awareness. In: Proceedings of the human performance, situation awareness and automation: user centered design for the new millennium conference; 2000.

10. Endsley MR. Direct measurement of situation awareness: validity and use of SAGAT. In: Situation awareness analysis and measurement. Mahwah, NJ: Erlbaum; 2000. p. 147-73.

11. Endsley MR. Measurement of situation awareness in dynamic systems. Hum Factors 1995;37(1):65-84.

12. Wilson GF. Strategies for psychophysiological assessment of situation awareness. In: Situation awareness analysis and measurement. Mahwah, NJ: Lawrence Erlbaum Associates Inc.; 2000. p. 175-88.

13. Byrne EA. Role of volitional effort in the application of psychophysiological measures to situation awareness. In: Proceedings of the international conference on experimental analysis and measurement of ssituation awareness; 1995. p. 147-53.

14. Motz F, MacKinnon S, Dalinger E, Widdel H, Hockel S. Comparison of ship bridge designs using the situation awareness global assessment technique (SAGAT). In: Proceedings of 17th world congress on, ergonomics; 2009.

15. Wickens CD. Situation awareness: review of mica Endsley's 1995 articles on situation awareness theory and measurement. Hum Factors 2008;50(3):397-403.

16. Strater L, Endsley MR. SAGAT: a situation awareness measurement tool for commercial airline pilots. In: Proceedings of the human performance, situation awareness, and automation: user-centered design for the new millennium conference; 2000.

Wei Hengyang is a Ph.D. student at the School of Aeronautic Science and Engineering, Beihang University. He received his B.S. degree from the same university in 2008. His area of research includes cockpit display interface and flight simulation, human factors evaluation, and cockpit environment control.

Zhuang Damin is a professor and Ph.D. supervisor at the School of Aeronautic Science and Engineering, Beihang University, Beijing, China. He received his Ph.D. degree from the Kagoshima University, Kagoshima, Japan. His current research interests are human-machine engineering, computer simulation of human-machine interface and computational fluid dynamics.