Ref: C0236

Development of a tractor driving simulator to research er-

gonomics of agricultural machines

Danny Mann, Behzad Bashiri, Aadesh Rakhra and Davood Karimi

Department of Biosystems Engineering, University of Manitoba, Winnipeg, MB R3T 5V6,
Canada

Abstract

Designers of agricultural machines have introduced new automation features to agricultural

machines even though the impact on operators is not always known. As early as a decade
ago, there was anecdotal evidence from sprayer operators that lightbar guidance systems
were “hard to use” even though they were intended to partially automate the guidance task
by providing the operator with guidance information. In the past decade, increasingly sophis-
ticated guidance technologies have been introduced and the role of the operator in the ope-
rator-machine system continues to change. To address this void in the literature, research
has been conducted in the Agricultural Ergonomics Laboratory in the Department of Biosys-
tems Engineering at the University of Manitoba in an effort to better understand the impact of
automation technology on the workload of the tractor operator. The key feature of the Agri-
cultural Ergonomics Laboratory is a tractor driving simulator (TDS) that has been used to
simulate various tractor-machinery systems. The TDS is a unique research tool for studying
the ergonomics of mobile agricultural machines (MAMs). This paper describes the TDS and
explains how this research tool is being used to develop the knowledge necessary to design
MAMs that minimize the impacts, both physical and mental, on the operator.

Keywords: driving simulation, semi-autonomous agricultural machines, ergonomics,

mental workload, driving tasks

1 Introduction

Any visit to an exhibition of agricultural machinery will confirm that these machines are
becoming increasingly complex due to the incorporation of new technologies. Sensors are
being added so that the operator has a better understanding of how the machine is function-
ing. It is good to have this information, but the quantity of information available inside the
operator’s station can be overwhelming. Technology has also enabled some of the functions
to be automated. Although this may seem to be an obvious benefit to the operator, the ulti-
mate impact on the operator cannot be so easily predicted. If not designed carefully (from a
human factors perspective), information overload can be a real problem.
Automated systems were first designed to relieve the human of repetitive or continuous
manual tasks. It has been observed, however, that automation often redistributes workload
rather than reducing it because the human is forced to assume a supervisory role (Sarter et
al. 1997). As a supervisor, it is important that the operator have an awareness of what is
happening. This awareness of one’s surroundings is referred to as situation awareness (SA).
Endsley (1988) defined SA as “the perception of the elements of the environment within a
volume of time and space (level 1), the comprehension of their meaning (level 2) and the
projection of their status in the near future (level 3).” When human intervention is required

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 1/8

(and SA is lacking), the operator does not know how to respond because knowledge of the
steps leading to the crisis is missing (referred to as the out-of-the-loop syndrome) (Endsley et
al. 2003). The mental workload (MW) necessary to solve the problem will be significant.
Thus, it can be correctly concluded that automation changes the MW of the operator, but
does not necessarily reduce it. In fact, caution must be exercised that the introduction of au-
tomation does not increase operator workload [described as “clumsy” automation by Wiener
(1988) and Parasuraman et al. (2000)].
Driving simulators have been developed and used in the automobile industry since at
least the mid-1960s (Weir 2010). As research tools, they provide unique opportunities in
terms of experimental control, flexibility, cost, and safety. Over the last decade, they have
been used to research different aspects of automobile driving including human perception
and control (Kemeny and Panerai 2003), human factors aspects of driving (Rakauskas et al.
2004), and the design of vehicles and roadways (Kawamura et al. 2004). Driving simulators
have also been developed for other vehicles such as construction vehicles (Son et al. 2001),
cranes (Huang and Gau 2003), motorcycles (Ferrazzin et al. 1999), and bicycles (Kwon et al.
2001). This paper briefly reviews the main issues relevant to driving simulation and describes
a simulator that has been developed to simulate a tractor-machinery system.

2 Driving Simulation

A driving simulator makes it possible to operate vehicles with no actual movement, but
in realistic conditions (Kappler, 2008). In fact, a driving simulator provides an intelligent en-
vironment in which a human driver can perceive and control the operation of a virtual vehicle.
If the driving simulator is to reflect real situations, it must invoke the same driving behaviors
from drivers as they would exhibit in real-world driving. To achieve this goal, the driving simu-
lator must have the same appearance and dynamics as the real vehicle and provide the sa-
me information to the driver. It also must provide the same input devices for control of the
system. Some driving simulators provide only visual feedback, but most high fidelity driving
simulators provide motion, haptic, and auditory feedback which allow the driver to interact
with the vehicle and the environment in a multisensory fashion (Kemeny and Panerai, 2003).
Extensive research has shown that, depending on the driving task being simulated, non-
visual cues are necessary to provide realistic simulation (Siegler et al., 2001).
For research purposes, experimental control is the greatest advantage offered by dri-
ving simulators. A driving simulator enables control of many extraneous variables which can-
not be controlled in real driving. It is also less costly to conduct experiments with a driving
simulator compared to experiments in an instrumented car in a real environment. The safety
of the driver in the test is another important advantage of using a driving simulator. This fac-
tor is most significant when studying issues such as driver fatigue or driving during low visibi-
lity conditions. Finally, it is easier to measure driving performance variables and other para-
meters, such as physiological and psychological responses of the driver, in a driving simula-
tor than in a real vehicle (Horiguchi and Suetomi, 1995).
Despite the advantages, driving simulators have certain shortcomings. No driving simu-
lator can perfectly reproduce the real driving experience. Models of vehicle dynamics and
environmental disturbances can be made increasingly accurate, but can never be perfect.
Providing visual feedback that has the same field-of-view, resolution, and depth cues as tho-
se of a real visual scene is extremely difficult, if not impossible. Also, even in the most ad-
vanced driving simulators, certain motion cues are not possible to render, because no driving
simulator has an unlimited motion range. Direct rendering of simple vehicle maneuvers (such
as a long brake) requires large motion systems that are unrealistic. Engineers have develo-
ped special techniques such as motion washout filtering, tilt coordination, and motion scaling
that can render most vehicle motions, but these techniques do not completely resolve the
existing problems. Transport delay is another major issue; there is always a delay between
the subjects’ action and the simulator’s response. This is due to the time required for the ac-
quisition of the subject’s commands, computation of the appropriate response, and the delay
in the visual and motion subsystems. Not only should these delays be small, but all simulator
subsystems should be synchronized, a requirement that is difficult to achieve (Horiguchi and
Suetomi, 1995; Kemeny and Panerai, 2003).

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 2/8

3 Development of a tractor-air seeder driving simulator

3.1 Need for a tool to research ergonomics of agricultural machines

Designers of agricultural machines have introduced new automation features to agricul-

tural machines even though the impact on operators is unknown. As early as a decade ago,
anecdotal evidence was obtained from sprayer operators that lightbar guidance systems we-
re “hard to use” even though they were intended to provide the operator with useful guidance
information. In the past decade, increasingly sophisticated guidance technologies have been
introduced and the role of the operator in the operator-machine system continues to change.
It is becoming increasingly evident that the impact on the operator needs to be considered
when designing operator-machine systems such as agricultural machines. A long-term goal
of the researchers in the Agricultural Ergonomics Laboratory at the University of Manitoba is
to develop the knowledge necessary to design mobile agricultural machines (MAMs) that
minimize the impacts, both physical and mental, on the operator. Driving simulation is being
used to investigate changes to the operator’s workload associated with the introduction of
new guidance technologies in a controlled laboratory setting. Apart from the work being done
at the University of Manitoba, there are only a few recent articles that address automation of
MAMs (Marzani et al. 2009; Lang et al. 2009; Schmitz 2010).
During the past 15 years, two simulators have been developed and used in the Agricul-
tural Ergonomics Laboratory. Until 2008, research was being done using a simulator which
was designed to mimic the tasks associated with operating an agricultural sprayer. This
“sprayer driving simulator” (S-DS) was initially developed because guidance automation first
appeared in agricultural sprayers. Two previous PhD students contributed to improving the
fidelity of the S-DS. Displays and controls were added in response to an in-field task analysis
of agricultural sprayer operators (Dey and Mann 2009). Further improvements (i.e., simulator
yaw motion, steering torque feedback, auditory feedback, and a video projection system)
were made by a PhD student who researched the role of sensory cues on the physical and
behavioural validity of the S-DS (Karimi & Mann 2008a,b,c; Karimi et al. 2008a,b). In 2008,
Case New Holland (CNH) donated a late-model tractor cab to the University of Manitoba and
plans for the second simulator began. It has been designed to simulate a tractor-air seeder
system (TAS-DS).

3.2 Fundamental Characteristics of Driving Simulation

Driving simulation can be either stationary (i.e., a fixed base) or allow some movement
(i.e., rotation or vertical displacement). In motion-based driving simulation, actuators are
used to move the simulator in ways that can produce the sensations of movement. The moti-
on system enhances realism of simulation using characteristics such as pathway roughness,
bump encounters, vehicle-centered vibrations, and acceleration. It is the most difficult, sensi-
tive and expensive part of developing a driving simulator. If motion is not a necessity for the
research being envisioned, the simulator can be stationary. Weir (2010) has listed the prima-
ry driving tasks and maneuvers that can be done using fixed-base simulation (i.e., lane regu-
lation or path following and speed maintenance on a nominally straight roadway, easy ac-
celeration and braking tasks, gradual turn maneuvers, and on-center steering control tasks).
These driving tasks and maneuvers are a good match with operating a tractor-machine sys-
tem in the field (which typically involves following straight passes back and forth across a
field). A fixed-base driving simulator was deemed to be sufficient to represent the in-field mo-
vement of a tractor-machine system.
The visual scene for a tractor simulator is less complicated compared to the visual sce-
ne for the simulation of on-road vehicles. An automobile driver visually interacts with many
objects (i.e., other vehicles on the road, road edges, road signs, and other nearby objects).
Tractor driving does not involve such interactions, except for situations when a tractor opera-
tor uses an object on the field boundary as an aiming cue to facilitate driving on a straight
line. Therefore, the field boundary can be considered to be a long distance away from the
tractor, making the image-generating program simpler.

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 3/8

3.3 Design of the Tractor-Air Seeder Driving Simulator
In order to incorporate proper features in the simulator, it was first necessary to deter-
mine the tasks of operating a tractor air-seeder system. A complete function-oriented task
analysis was conducted to identify the required functions, tasks, and subtasks associated
with operating a tractor-air seeder system. A complete description of this task analysis pro-
cedure and its results can be found in Bashiri et al. (2011). The task analysis revealed that
operators use a GPS guidance system (lightbar) as the main source of information for stee-
ring the tractor-air seeder system. Operators allocate a substantial portion of their time (i.e.,
anywhere from 10-50%) to controlling the air seeder. They also scan other displays in the
cab such as i) a GPS mapping system (which shows a bird’s-eye view of the field with the
tractor’s position in the field and the area where air seeding has been completed) and ii) an
application display (which provides such information as the forward speed, seeding depth,
fan rotational speed, and the amount of seed and fertilizer in the air-seeder tank).
Figure 1 shows a complete block dia-
Back monitor Projector Curved screen
gram representation of the TAS-DS that has
been developed. Use of an actual tractor cab Control unit Steering wheel
provides a realistic environment for the opera-
tor (Fig. 2). The TAS-DS includes a cab moni-
tor situated to the right of the operator’s seat
that displays status information for the follo-
wing air seeder parameters: amount of
seed/fertilizer in the air seeder tank, seed Console
application rate, fertilizer application rate, trac-
tor forward speed, fan rpm, tool pressure, Implement information display

working depth, and blocked seed distribution Figure 1: Overhead view of the TAS-DS.
units (Fig. 3). The display reflects the states of
these parameters in a suitable format (i.e.,
with the use of graphical images, pictorials, or
text). In addition, the monitor displays a map
of the “field” showing the portion that has be-
en covered by the TAS-DS. Also on the right
side of the operator’s seat is a control panel
(Fig. 4) that is used by the operator to make
adjustments to the air seeder parameters as
required. A lightbar with 23 light-emitting dio-
des (LEDs) is used to provide guidance in-
structions to the simulator operator. It con-
sists of three green LEDs in the center with
10 red LEDs on each side. The details of Figure 2: Placement of the TAS-DS in the Ag-
modeling straight line driving with this lightbar ricultural Ergonomics Laboratory.
as a guidance aid can be found in Karimi et al.
(2008a).

Figure 3: Information display used in the TAS-DS.

3
Item 1 represents the forward motion of the tractor
(green stripe) in the field and item 2 shows the for-
4 5 ward speed of the tractor (and acceptable range).
Items 3 to 11 are related to the air seeder system.
1 Item 3 shows the full status of the air seeder tank.
Items 4 and 5 show the seed and fertilizer applicati-
2 6 7 on rate. Items 6 and 7 display fan RPM and tool
pressure. Items 8 and 9 show working depths of
8 9 tools. Item 10 indicates whether any seed distributi-
on boots are blocked. Item 11 is a message box
that is used to provide necessary information regar-
10 11
ding air seeder parameters.
.

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 4/8


Figure 4: Control panel installed insi-
de the TAS-DS on the right-hand
side of the operator’s seat.
The task of operating a tractor-air seeder sys-
tem also includes monitoring of the air seeder that is
mounted behind the tractor. To mimic this charac-
teristic, two computer monitors were situated behind
the cab (one to rear-left of the operator’s seat and
one to the rear-right of the operator’s seat) (Fig. 5).
Various images are displayed on the monitors when
the simulator is in use. The images either depict
normal operation of the seeding equipment or one of
several malfunctions (i.e., improper seeding depth,
seeding tool plugged with crop residue, or seed spil-
lage). The computer is programmed to randomly dis-
play malfunctions; it is the operator’s mandate to
watch the computer monitors to detect the occur-
rence of malfunctions and make the necessary cor-
rective action using the control panel. The rear-
monitoring task can be completed with manual turn-
ing, with the use of rear-view mirrors on either side of
the cab, or with the use of a camera-based monitoring
system that displays images inside the cab. The simu-
lator computer records the occurrence of each mal-
function as well as the time interval until the correct
action is taken to fix the malfunction.

Figure 5: A rear monitor used to

simulate the air seeder portion of the
TAS-DS.

Thirty-two images create a panoramic view that forms the field boundary for the TAS-
DS (Fig. 6) . A photo of a level field surface for seeding has been used to create the field
texture of the visual scene. The image-generating program receives computed vehicle mo-
tions from the main program and renders them in two translational and rotational motions.
Translational motion, which is the movement in the y-direction, is only applied for the field
surface. In this position, the field boundary only moves along with the driver’s virtual position
in the scene. Rotational motion is applied to both field surface and boundary. A desktop
computer generates the visual scene and synchronizes it with the simulator controller's data.
Figure 7 shows a view of the visual scene when the simulator is running.

Figure 6: Schematic illustrating how the visual scene

is simulated in the TAS-DS. Photo of tractor and see-
der is taken from a sales brochure.

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 5/8

 Figure 7: Simulated visual scene from the driver’s
seat in the TAS-DS.

Noise and vibration levels are usually higher in tractors compared to automobiles. Pre-
recorded tractor noise has been incorporated into the simulator code to include auditory sti-
mulation. Although vibration may be important to realistically simulate the agricultural tractor,
this characteristic has not been considered to be essential for the research studies envisio-
ned to date. The TAS-DS does not currently include any mechanism for creating vibration.

4 Research Enabled by the TAS-DS

The TAS-DS has been used as a tool in the search for knowledge to inform the design
of mobile agricultural machines (MAMs). Research over the past few years has shown that
the ergonomic impacts of guidance systems cannot be considered in isolation because au-
tomation of the guidance task can change the nature of the remaining tasks. With the overall
objective of considering the design of automation for MAMs, the TAS-DS is currently being
used to research two issues. First, research is being done, using function allocation theory,
to determine an appropriate automation design for a MAM. Four different levels of automati-
on can be achieved: information acquisition, information analysis, decision selection, and
action implementation (Parasuraman et al. 2000). Experimental work has been completed
with the intent of determining whether system performance varies with the level of automati-
on implemented. Second, research is being initiated to understand the impact of display de-
sign on the level of situation awareness achieved by an operator of a semi-autonomous
MAM. Future research will investigate whether an external stressor (such as noise or re-
duced visibility) changes the preferred level of automation for a semi-autonomous MAM.

5 Conclusions

Industry continues to develop new technologies for “improving“ mobile agricultural ma-
chines. Unfortunately, the impact of the new technology on the operator’s mental workload is
not always positive. Over the past 15 years, driving simulation has been a valuable technique
to research the ergonomics of mobile agricultural machines. This paper describes a simulator
that has been designed to mimic a system consisting of a tractor and air seeder (TAS-DS).
The current driving simulator is being used to research various issues related to the design of
automation for mobile agricultural machines.

6 Acknowledgements

The authors acknowledge the technical assistance of Matt McDonald, Dale Bourns,
and Robert Lavallee, the tractor cab donation by Case New Holland (CNH), and the financial
assistance of the Natural Sciences and Engineering Research Council of Canada (NSERC).

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 6/8

7 References

Bashiri, B., Mann, D.D., & Karimi, D. (2011). Hierarchical Task Analysis of Driving a Tractor
Air Seeder System (TAS): Determining TAS Simulator Requirements. 2011 CSBE/SCGAB
Conference, 10-13 July 2011, Winnipeg, MB, Canada.
Dey, A. K., & Mann, D.D. (2009). A complete task analysis to measure the workload
associated with operating an agricultural sprayer equipped with a navigation device. Applied
Ergonomics 41, 146-149.

Endsley, M.R. (1988). Design and evaluation for situation awareness enhancement. In
Proc. HFES 32nd Annual Meeting, pp 97-101. Santa Monica, CA: HFES.

Endsley, M.R., Bolté, B., & Jones, D.G. (2003). Designing for Situation Awareness: An Ap-
proach to User-Centered Design. London, UK: Taylor & Francis.
Ferrazzin, D., Salsedo, F., & Bergamasco, M. (1999). The MORIS simulator. In Proc. of the
1999 IEEE International Workshop on Robot and Human Interaction, 136-141. Pisa, Italy.
Horiguchi, A., & Suetomi, T. (1995). A Kansei Engineering approach to a driver/vehicle sys-
tem. International Journal of Industrial Ergonomics, 15, 25-37.
Huang, J.Y., & Gau, C.Y. (2003). Modeling and designing a low-cost high-fidelity mobile cra-
ne simulator. International Journal of Human-Computer Studies, 58, 151–176.
Karimi, D., & Mann, D.D. (2008a). Role of motion cues in straight-line driving of an agricultu-
ral vehicle. Biosystems Engineering, 101(3), 283-292.
Karimi, D. & Mann, D.D. (2008b). Role of visual cues in driving an agricultural vehicle. The
Ergonomics Open Journal 1, 54-61.

Karimi, D. & Mann, D.D. (2008c). Torque feedback on the steering wheel of agricultural ve-
hicles. Computers and Electronics in Agriculture 65, 77-84.
Karimi, D., Mann, D.D., & Ehsani, R. (2008a). Modeling of straight-line driving with a
guidance aid for a tractor-driving simulator. Applied Engineering in Agriculture, 24(4), 403-
408.
Karimi, D., Mondor, T., & Mann, D.D. (2008b). Application of auditory signals to the operati-
on of an agricultural vehicle. Journal of Agricultural Safety and Health, 14(1), 71-78.
Kappler, W.D. (2008). Smart Driver Training Simulation: Save Money. Prevent. Springer-
Verlag Berlin Heidelberg.
Kawamura, A., Maeda, C., Shirakawa, T., Ishida, T., Nakatsuji, T., & Himeno, K. (2004). Ap-
plicability of a driving simulator as a new tool for the pavement surface evaluation. In Pro-
ceedings of the SIIV (Italian Society for Transportation) 2004 International conference. 52, 1-
10. Florence, Italy.
Kemeny, A., & Panerai, F. (2003). Evaluating perception in driving simulation experiments.
Trends in Cognitive Sciences, 7(1), 31-37.
Kwon, D.S., Yang, G.H., Lee, C.W., Shin, J.C., Park, Y., Jung, B., Lee, D.Y., Lee, K., Han,
S.H., Yoo, B.H., Wohn, K.Y., & Ahn. J.H. (2001). KAIST interactive bicycle simulator. In Pro-
ceedings of IEEE International Conference on Robotics and Automation 3, 2313- 2318.
Lang, T., Gores, T., Junemann, D., Vollrath, M., Werneke, J., & Huemer, A.K. (2009). Ana-
lysis of human factors on agricultural machines. Landtechnik 64(1), 58-60.
Marzani, S., Tesauri, F., Minin, L., Montanari, R., & Calefato, C. (2009). Designing a control
and visualization system for off-highway machinery according to the adaptive automation
paradigm. FAC’09 Proc. 5th Intl. Conf. Foundations of Augmented Cognition, Neuroergono-
mics and Operational Neuroscience, pp 42-50.

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 7/8

Parasuraman, R., Wickens, C.D., & Sheridan, T. (2000). A model for types and levels of hu-
man interaction with automation. IEEE Trans. Sys, Man & Cybernetics 30, 286-297.
Rakauskas, M.E., Gugerty, L.J., & Ward. N.J. (2004). Effects of naturalistic cell phone con-
versations on driving performance. Journal of Safety Research. 35, 453-464.
Sarter, N.B., Woods, D.D., & Billings, C.E. (1997). Automation Surprises. In Handbook of
Human Factors and Ergonomics, 2nd Edition. Wiley & Sons: New York, NY.
Schmitz, B. (2010). Ergonomics and automation – safe manipulation of complex systems.
Landtechnik 65(3),167-169.
Siegler, G., Reymond, G., Kemeny, A., & Berthoz, A. (2001). Sensorimotor integration in a
driving simulator: contributions of motion cueing in elementary driving tasks. Driving Simula-
tor Conference DSC. 21-32. Sophia Antipolis, France.
Son, K., Goo, S.H., Choi, K.H., & Lee, W.S.Y. (2001). A driving simulator of construction ve-
hicles. International Journal of the Korean Society of Precision Engineering, 2(4), 12-22.
Weir, D.H. (2010). Application of a driving simulator to the development of in-vehicle human-
machine-interfaces. IATSS Research, 34(1),16-21.
Wiener, E.L. (1988). Cockpit automation. In E.L. Wiener & D.C. Nagel (Eds.), Human fac-
tors in aviation (pp.433-459). San Diego, CA: Academic Press.

Proceedings International Conference of Agricultural Engineering, Zurich, 06-10.07.2014 – www.eurageng.eu 8/8