Sei sulla pagina 1di 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327869301

Eye Tracking for Assessment of Mental Workload and Evaluation of RVD


Interface: Proceedings of the 18th International Conference on MMESE

Chapter · January 2019


DOI: 10.1007/978-981-13-2481-9_2

CITATION READS

1 110

5 authors, including:

Yu Tian Chunhui Wang


China Astronaut Research and Training Center astronaut center of china
27 PUBLICATIONS   41 CITATIONS    30 PUBLICATIONS   111 CITATIONS   

SEE PROFILE SEE PROFILE

Shanguang Chen
Astronaut Center of China
101 PUBLICATIONS   404 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Neuroprotection of herbal plants View project

China manned space mission View project

All content following this page was uploaded by Yu Tian on 10 March 2019.

The user has requested enhancement of the downloaded file.


Eye Tracking for Assessment
of Mental Workload and Evaluation
of RVD Interface

Yu Tian, Shaoyao Zhang, Chunhui Wang, Qu Yan


and Shanguang Chen

Abstract Eye tracking is increasingly employed to evaluate operators’ mental


workload and the usability of interfaces. In this paper, eye tracking data were
recorded for ten operators performing the simulated manually controlled ren-
dezvous and docking (manual RVD) of two space vehicles. Indices such as blink
rate, blink duration, percent eyelid closure (PERCLOS) were calculated to assess
the mental workload and fatigue level of the operators. Fixation measures were
analyzed to investigate the attention allocation of the operators on the different
information areas on the display. Results showed that the workload of the RVD task
was generally acceptable. However, workload increased in the accurate control
stage (the last 20 meters’ approaching) compared to the tracking control stage (the
more distant approaching). The fixation measures showed that human eyes were
mostly fixed on the area of the spacecraft image, while numerical display areas
provided compensatory information. The present study revealed that the design of
the RVD interface supported human perception and task completion.


Keywords Eye tracking Manually controlled rendezvous and docking

Workload Interface evaluation

1 Introduction

There is a hypothesis that the eye movements of a person are mainly directed by his
visual attention and are closely related to cognitive processes in the brain [1]. Eye
movements can be recorded accurately by the eye trackers. Measures of the eye
activities, such as the fixations, the scan paths, the blink rate, the pupil diameter, can

Y. Tian (&)  S. Zhang  C. Wang  Q. Yan  S. Chen


National Key Laboratory of Human Factors Engineering, China Astronaut
Research and Training Center, Beijing 100094, China
e-mail: cctian@126.com

© Springer Nature Singapore Pte Ltd. 2019 11


S. Long and B. S. Dhillon (eds.), Man-Machine-Environment
System Engineering, Lecture Notes in Electrical Engineering 527,
https://doi.org/10.1007/978-981-13-2481-9_2
12 Y. Tian et al.

be extracted from the eye tracking data and have been applied for assessing
interface usability, mental states of the operator, as well as human–computer
interactions [1–4].
Manually controlled rendezvous and docking (manual RVD) of space vehicles is
a complex human–computer interaction (HCI) task in space [5–7]. The manual
RVD simulators provide facilities to train the astronauts to master necessary skills
on the ground. Experiments with eye tracking of the operators in the manual RVD
simulators were employed to evaluate the design of task interface and the mental
states of operators in the task process.

2 Methods

2.1 Subjects

Ten male technicians aged between 26 and 31 years from Astronaut Research and
Training Center of China participated in the experiments. Each of them had an
education background with at least a bachelor’s degree and work experience related
to manned spaceflight projects.

2.2 Equipment

The prototype of the manual RVD simulator on the ground is shown in Fig. 1 [6].
The video image of the target spacecraft obtained from cameras is displayed on the
monitoring interface, with the numerical data of the relative position and altitude of
the two spacecrafts obtained from the sensors overlaid on the corners of the
interface. The operator observes the target and numerical data on the monitoring
interface and manipulates the controllers to control the chaser space vehicle [7].
Measures such as the control time and the fuel consumption are recorded by the
system.

Fig. 1 Prototype of the manual RVD simulator on the ground [6]


Eye Tracking for Assessment of Mental Workload … 13

The eye movements of the operators in the manual RVD tasks are recorded by
the H6 head-mounted eye tracker produced by Applied Science Laboratories
(ASL). The H6 head-mounted eye tracker system records the location and diameter
of the pupil at a sampling rate of 60 Hz. The system is designed to track gaze
direction over approximately a 30°–35° vertical visual angle and a 40°–45° hori-
zontal visual angle.

2.3 Experiment Design and Procedure

The subjects were trained twice to ensure that they were familiar enough with the
manual RVD tasks. In the formal experiments with eye tracking, the eye tracker
was put on the head of the subject, and then calibration work was done for the eye
tracking system. After that, each subject practiced once and then performed two
consecutive trials of manual RVD tasks of the same difficulty level, with around
3 min rest in the middle. In the manual RVD tasks, the subject observed the
monitoring interface and regulated the relative position and posture of the two
spacecrafts by manipulating the control handles. The initial distance of the two
spacecrafts was set to be 100 m, and the control time for one trial of the manual
RVD task was mostly within 10 min.

3 Results

3.1 Data Collected and Indices Calculated

In the experiments, the subjects’ performance data in the manual RVD tasks and
their eye tracking data during the task process were collected. The performance
indices considered in this paper were control time and fuel consumption. Control
time is the time from the beginning to the end of the manual RVD task. Fuel
consumption is the fuel consumed in the whole manual RVD task process.
Eye movement measures in the whole task process, such as the blink rate, the
blink duration, the 80% eyelid closure (PERCLOS) [8], the fixation distribution,
and dwell time in areas of interest (AOIs), were calculated.
Pervious research revealed that manual RVD can be divided into tracking control
stage and accurate control stage. Tracking control stage is from the initial distance to
about 20 m away from the target spacecraft. In this stage, due to the long distance,
subjects mainly track the target by observing the video image of the target spacecraft,
and large deviation, attitude, and translation of the two spacecrafts are eliminated in
this stage. Accurate control stage is 20 m away from the target spacecraft to the end
of the task. In this stage, subjects have to narrow the deviation, attitude, and
translation of the two vehicles to meet the docking access requirements [6].
14 Y. Tian et al.

Averagely, the tracking control stage counts for about 64% of the control time, and
the accurate control stage counts for the remaining 36% of the control time. Eye
movement measures in the two stages were also calculated separately by dividing the
whole eye tracking data into the first 64% of the dataset and the remaining 36%.

3.2 Fixation Distribution and Dwell Time

Statistic data showed that human eyes were fixed on the image of the spacecraft
nearly 80% of the time, while fixations on the numerical display areas counted for
only about 20% of all the fixations. Among the numerical display areas, the velocity
display area attracted about twice more fixations than the deviation display area,
indicating that the velocity information may be more important for subjects per-
forming the manual RVD tasks. The average fixation time of the subjects on the
velocity display area (0.92 s) was much longer than the average fixation time on the
deviation display area (0.49 s), indicating that the velocity information may be
more difficult to extract.

3.3 Comparisons of Eye Status in the Two Stages

Results showed that the blink rate in the accurate control stage was significantly
lower than that in the tracking control stage (p = 0.012, n = 10). The decrease of
eye blink rate in the accurate control stage indicated that there might be an
increased demand involving visual attention in the accurate control stage [9].
PERCLOS and blink duration have been considered predictors of drowsiness
and mental fatigue [3, 8, 10]. Both the PERCLOS and the blink duration in the two
stages were small and show no significant difference, indicating that the subjects
were not drowsy at all during the experiments.

3.4 Eye Activity and Task Performance

Correlations of three eye movement measures with task performance were analyzed
(see Table 1). Blink rate shows a significant correlation with both control time and
fuel consumption. This shows that the eye activities and the task properties are
interconnected: While the increase of task demand causes a decrease in blink rate,
lower blink rate (which may indicate higher concentration level) is accompanied by
better task performance.
Although the PERCLOS is not significantly correlated with the task perfor-
mance, for each of the ten subjects who performed two trials of manual RVD tasks,
a task with longer control time was accompanied by higher PERCLOS (see Fig. 2).
Eye Tracking for Assessment of Mental Workload … 15

Table 1 Spearman’s correlation of eye movement measures with performance indices of manual
RVD tasks
Performance indices Blink rate PERCLOS Blink duration
Control time 0.507* −0.027 0.240
Fuel consumption 0.549* −0.376 0.052
*P < 0.05

Fig. 2 PERCLOS of the subjects in the two trials of manual RVD tasks

The results may indicate that there is a tendency of getting fatigue as control time
increases, although they are not really drowsy. These data validate that PERCLOS
is an effective predictor of mental fatigue.

4 Discussion

The eye tracking data show that some areas (such as the velocity display area from
the numerical displays) attract more fixations than other areas. More fixations on a
particular area indicate that this area is more important or more noticeable [1]. As
all the numerical displays are in a parallel relation, the velocity display area is not
more noticeable, so the only explanation is that the area is more important for the
operators performing the manual RVD tasks. The velocity display area also attracts
longer fixation dwell time, indicating that information presented in the area may be
16 Y. Tian et al.

more difficult to extract. If velocity display is designed to be more noticeable, the


overall visual workload of the operators may decrease. And from the decrease of
the blink rate in the task process, it can be inferred that the visual workload in the
second stage of the task process (referred to as accurate control stage) may be
significantly higher than the first stage (referred to as tracking control stage). So in
the second stage, if numerical information is still needed, it should be brief and
noticeable in order not to further increase the visual workload of the operator.
Two eye movement measures, the PERCLOS and the blink duration, do not
show significant changes in the two control stages. These two measures are pre-
dictors of drowsiness [3, 8, 10], and the results may indicate that the operators are
not drowsy at all in the task process. Although PERCLOS tends to increase in tasks
with longer control time, the absolute value is still low, which shows that the fatigue
level of the operators is unlikely to increase sharply during the task process.
Although the eye activities of the subjects show some similar characteristics, the
individual differences in eye movements between participants were also found. For
example, the fixation patterns of the subjects were moderately different. So gen-
erally, it is not proper to use a within-participants design to make valid performance
comparisons. But some researchers have illustrated that the eye movement mea-
sures may predict the task performance in a complicated way [11]. In our study, the
blink rate shows a significant correlation with the performance of the manual RVD
task. One possible explanation is that blink rate may be connected with a con-
centration level of the operator; the more concentrated the operator, the lower the
blink rate tends to be. And when the operator is more concentrated, he is more
likely to perform well in the task.

5 Conclusion

The present study demonstrates that the eye movement activities in the manual
RVD task can provide useful information for interface evaluation and mental states
assessment. Eye tracking data indicate that the monitoring interface of the manual
RVD simulator generally provides sufficient information for the task, and the
workload of the operator is acceptable. Some suggestions for small modifications of
the interface are also proposed based on eye tracking analyses.

Acknowledgements This study is supported by National Key Research and Development


Program (No. 2017YFB1300305), Foundation of Key Laboratory of Science and Technology for
National Defense (No. 6142222030301, No. 614222204020617), Foundation of National Key
Laboratory of Human Factors Engineering (No. SYFD170051802), and National Basic Research
Program of China (No. 2011CB711000).
Compliance with Ethical Standards The study was approved by the Logistics Department for
Civilian Ethics Committee of China Astronaut Research and Training Center. All subjects who
participated in the experiment were provided with and signed an informed consent form.
All relevant ethical safeguards have been met with regard to subject protection.
Eye Tracking for Assessment of Mental Workload … 17

References

1. Strandvall T (2003) Eye tracking in human-computer interaction and usability research.


Minds Eye 2(3):573–605
2. Orlov PA, Apraksin N (2015) The effectiveness of gaze-contingent control in computer
games. Perception 44(9):1136–1145
3. Tsai YF, Viirre E, Strychacz C, Chase B, Jung TP (2007) Task performance and eye activity:
predicting behavior relating to cognitive workload. Aviat Space Environ Med 78(5)
4. Singh H, Singh J (2018) Real-time eye blink and wink detection for object selection in HCI
systems. J Multimodal User Interfaces 12(1):55–65
5. Du X, Zhang Y, Tian Y, Huang W, Wu B, Zhang J (2015) The influence of spatial ability and
experience on performance during spaceship rendezvous and docking. Front Psychol 6:955
6. Jiang T, Wang C, Tian Z, Xu Y, Wang Z (2011) Study on synthetic evaluation of human
performance in manually controlled spacecraft rendezvous and docking tasks. In:
International conference on digital human modeling, 2011, pp 387–393
7. Wang C, Tian Y, Chen S, Tian Z, Jiang T, Du F (2014) Predicting performance in manually
controlled rendezvous and docking through spatial abilities. Adv Space Res 53(2):362–369
8. Bergasa LM, Nuevo J, Sotelo MA, Barea R, Lopez ME (2006) Real-time system for
monitoring driver vigilance. IEEE Trans Intell Transp Syst 7(1):63–77
9. Wilson GF (2002) An analysis of mental workload in pilots during flight using multiple
psychophysiological measures. Int J Aviat Psychol 12(1):3–18
10. Galley N, Schleicher R, Galley L (2003) Blink parameter as indicators of driver’s sleepiness—
possibilities and limitations
11. Van Orden KF, Jung TP, Makeig S (2000) Combined eye activity measures accurately
estimate changes in sustained visual task performance. Biol Psychol 52(3):221

View publication stats

Potrebbero piacerti anche