HF/E Audit of a Locomotive Remote Control Device

Abstract

Conducting a complete human factors/ergonomics (HF/E) audit of a unique technological device can be complicated. Designing an instrument to aid HF/E practitioners in this process can help to ensure that organizational and operator needs are taken into consideration, relevant HF/E principles are followed, and future audits for similar devices are streamlined. Our team designed such an instrument to evaluate a locomotive remote control device. We provide illustrations of the process cycle and time requirements for each phase to be used as a planning tool for HF/E practitioners who are faced with a similar challenge.

Keywords

railway surface transportation task analysis link analysis human factors evaluation audit human factors principles and standards

Having an advocate for human factors/ergonomics in the company smoothes the way for recommended HF/E improvements to device design.

Human factors/ergonomics (HF/E) professionals are frequently involved in various parts of the design cycle – studying the needs of the user, planning designs that balance the abilities of the user with the intent of the technology, and evaluating the usability of products. Conducting scientific, in-depth, and efficient HF/E evaluations under time constraints is a challenging task and calls for efficiently incorporating multiple methodologies.

CSX Transportation tasked the Cognitive Ergonomics Laboratory with performing an HF/E audit of the Cattron Accuspeed™ Operator Control Unit. The control unit (see Figure 1) is used in remote-controlled freight rail switchyard operations. In place of a locomotive engineer operating the engine, remote control operators (RCOs) use the control unit to control movement of the locomotive while assembling (i.e., “building”) a train.

Figure 1.

Top of the control unit with all controls labeled.

The purpose of this article is to describe the method we used to conduct an HF/E evaluation of the control unit. Additionally, we describe challenges that arose throughout the process and how we addressed those challenges.

HF/E Audit Overview

At the time the audit was assigned to our team, a new version of the control unit was in the redesign phase and set to begin production within months. Therefore, the audit had to be performed expeditiously to ensure that any HF/E concerns could be identified and conveyed to designers before the new control unit began production. Given that the rail industry only recently began integrating remote control technology into day-to-day operations, the amount of HF/E research on the control unit was limited (Reinach & Viale, 2006).

In the early stages of our project, we identified a body of technical reports commissioned by the Federal Railroad Administration on control units. This research was conducted by Reinach and colleagues and made up nearly all the remote control unit studies in the United States. They completed a multiphase investigation on the safety of remote control operations (Reinach, 2004; Reinach & Acton, 2006; Reinach, Fadden, Gamst, Acton, & Bartlett, 2006; Reinach & Viale, 2006). Compared with these previous studies, our analyses were part of a more focused consideration of one control unit model and were conducted during a much shorter period: 5 months.

We wanted to identify sources giving rise to HF/E issues with the control unit that originated at the organizational, task, RCO, or device level. To accomplish this in a systematic way, we outlined a set of stages that consisted of methods typically leveraged by HF/E practitioners during the evaluation process.

We began by familiarizing ourselves with the control unit (Phase 1) and conducting a needs analysis for the RCO’s job (Phase 2). Once we better understood how the control unit was used and the job demands involved in its implementation, we developed a master list of HF/E standards and principles from multiple sources, including technical standards documents, principles from textbooks, and other sources (Phase 3). We then eliminated principles that did not relate to the control unit and added “future research needs” to record desired subsequent analyses.

For example, to address the principle “Keep the flow of work moving along smooth curves over space, minimizing backtracking, extra movement, delays and needless inventory” (Lehto & Buck, 2008), we opted to conduct a time-and-motion study and link analysis. This modified list of principles and future needs served as our primary evaluation instrument for the control unit and enabled us to identify gaps in our initial analysis (Phase 4). Finally, we obtained feedback from our industry contact to tailor the document for CSX decision makers (Phase 5).

Although we understood that the process would be iterative, had we realized the true extent of the iterative nature of the process, we might have excluded certain tasks and allocated more time to accomplishing others. For example, to construct the task analysis, we would have spent less time using the control unit simulator and proceeded directly to the interviews and cognitive walk-throughs. Although we originally thought we could create a reasonable task analysis by using the simulator alone, it was beneficial only in familiarizing us with overall RCO duties needed to understand certain tasks. Adding cognitive walkthrough and verbal protocol analysis techniques was much more informative when fleshing out the task analysis.

The following sections describe each phase of our process in detail. Table 1 lists the phases along with the percentage of time allocated to each to illustrate how resource intensive each step of an evaluation process can be. Subphases are listed below each phase.

Table 1.

Percentage of Time Spent Working on Each Phase of the Audit

Phase/Subphase	Percentage of Time
Familiarization	21
Review control unit and RCO materials
Interview SMEs
Training simulator walk-throughs
Needs analysis	28
Organizational analysis
Person analysis
Task analysis
Job analysis
Evaluation instrument design	9
Preliminary analysis
Selection of standards and principles
Master list development
Evaluation instrument implementation	34
Team consensus of control unit compliance with master list
Identify principles requiring further analysis
Feedback	8
Detailed interviews with SMEs
Time-and-motion study
Enhanced task analysis
Developed matrix of identified human factors issues

Note. RCO = remote control operator; SME = subject matter expert.

Phase 1: Familiarization

To prepare us for the assignment, CSX provided our team with a TrainMaster™ Rail Operation Simulation system complete with 14 train yard scenarios and a replica control unit. As none of our team members is a certified RCO and therefore could not participate in live remote operations, this simulator enabled us to gain firsthand experience of train yard procedures and operations.

Phase 2: Needs Analysis

We conducted a multipronged analysis to better understand the needs of the RCOs and the organization. This analysis was derived from a compilation of previous methodologies and encompassed an evaluation of the needs of the organization, the needs of the users of the device, and the nature of the task in which the device was used (Salas, Wilson, Priest, & Guthrie, 2006).

At the beginning of the needs analysis phase, CSX appointed a senior RCO instructor as a confederate for the team. This subject matter expert (SME) also had multiple years of experience as an RCO and was available to answer questions throughout the process. Establishing a close relationship with an experienced SME enabled our team to learn about the resources and constraints surrounding the introduction of new technology into train yard operations. The SME also was able to express struggles with integrating technology into operations and identify areas where support for this transition could be improved.

Organizational analysis

We used a variety of resources to gain an understanding of the context and organizational culture in which the RCOs work. A review of the CSX training literature, safety manuals, and general information about CSX Transportation via its Web site, as well as interviews with CSX management, enabled us to learn about system-wide elements that influence RCO behaviors and attitudes. To better understand the general guidelines with which RCOs were expected to work, we reviewed the policies, procedures, and operating rules.

Task analysis

The first iteration of hierarchical task analyses (Kirwan & Ainsworth, 1992) started during Phase 1 of the process. Researchers used the TrainMaster Rail Operation Simulation system to identify important jobs on the switchyard.

As we progressed, we used interviews, cognitive walk-throughs, and talk-throughs with SMEs to explore each job in greater detail. Involving the SMEs at this stage was useful not only in checking the task analyses for completeness but also for identifying features of the control unit that could potentially create difficulty for some users. For example, controls that the RCOs were instructed to use for nearly all their tasks were located on the right side of the control unit. This feature might make it difficult for left-handed RCOs to effectively use the control unit.

The iterative nature of the audit process became particularly apparent to our team during the needs analysis phase. For example, the goal of the interviews during this subphase was to gather more information regarding the task analysis. However, what we learned in this subphase made the researchers aware of gaps in organizational analysis that needed to be addressed. When we revisited the organizational analysis to fill those gaps, we discovered more details that were helpful in building a more complete task analysis.

Job/person analysis

To gain a better understanding of the RCO and the job, we conducted a series of interviews with the SME, who served as the stand-in human resources representative for RCO sourcing and development. We learned about RCO selection criteria and key demographics. The physical rigors of the job, in addition to competencies required to efficiently interact with the control unit, became apparent.

For example, we learned that RCOs often operate the control unit with one hand while riding on the side of a train (see Figure 2). We noted that performing this action while simultaneously using controls on the control unit that are contralateral to the free hand might prove challenging for some users.

Figure 2.

An operator poses to illustrate how the control unit is used to control a train while riding on the side of a box car. Although situations requiring the operator to reach across the control unit are rare, activating appropriate controls in this manner might be challenging for some operators.

Phase 3: Instrument Design Evaluation

The control unit was a combination of display panels and controls. Our team composed a master list of principles and standards related to these categories (e.g., Lehto & Buck, 2008; Sanders & McCormick, 1993; Wagner, Birt, Snyder, & Duncanson, 1996). At least two, and usually three, members of our team reviewed each principle or standard (i.e., item) on the list, discussed the relevance of the item with regard to remote control devices, and reached a unanimous decision whether to keep the item because it was relevant to the control unit, combine it with a similar item, or consider the item as not applicable. For example, control principles from the master list that related to slider controls were determined to be irrelevant because the control unit did not have these types of controls.

Additionally, our team identified Federal Aviation Administration standards that were similar to HF/E design principles (and similar design principles across HF/E handbooks), so these items were combined into one item on the evaluation instrument. After we eliminated all irrelevant principles and combined similar items, there were a total of 370 applicable items on our evaluation instrument against which we analyzed the control unit.

Phase 4: Instrument Implementation Evaluation

We used each relevant principle to guide an in-depth analysis of the control unit against that principle, noting features that were consistent with the principle or, conversely, violated it. Whenever possible, the team made recommendations about changing features on the control unit to comply with the principles. These recommendations were intended to inform designers of potential HF/E issues specific to the device (see Table 2).

Table 2.

Selected Items From the Evaluation Instrument Used to Guide Analysis of the RCOs’ Remote Control Unit

Human-Centered Design Principles	Outcomes and Recommendations
Physical or physiological aspects
6-11. Work should not be performed consistently at or above the level of the heart; even occasional performance of tasks where forces are exerted above heart level should be avoided. When light handwork must be performed above heart level, rests for the upper arms are required. (Also FAA 7.4.1.1.1) Further Analysis Needs: What are the upper hooks on the vest used for? [Interview with SME]	The vest worn by RCOs allows them to attach the control unit at one of two positions. Although the lower setting places the top of the control unit at abdomen level, some RCOs preferred to attach the control unit at the higher, above-the-heart, setting. Recommended: Remove the clasps at the top of the vest so that it would not be possible to mount the control unit at the higher setting. As a result, CSX managers are working with RCOs to design a vest with more appropriate mounting options.
Perceptual aspects
18-1. The most appropriate sensory modality of a display depends on its intended function, sensory demands of the background task, and sensory capabilities of its intended audience. Further Analysis Needs: Field observation Task analysis [Immediacy of alert & workload limitations] Time-and-motion study on multiple RCOs Do they wear any protective gear? [Interview with SME] Season-specific equipment? [Interview with SME] Would integrating vibrotactile aids into the design be feasible? [Interview with CSX management and control unit stakeholders]	While conducting field studies in the train yard, we noticed that RCOs were periodically subjected to a great deal of auditory noise that was unrelated to their task. Additionally, we found that the control unit alerted RCOs to certain task-relevant events exclusively through auditory means. Recommended: Using vibrotactile aids to alert the RCO to the aforementioned vigilance time-out event. However, along with this recommendation, the team pointed out that such technologies consume battery power very quickly. Our team presented both sides of this design trade-off to stakeholders, who will make the ultimate decision.
18-10. Symbols and alphanumeric characters should subtend a visual angle of at least 12 minutes of arc. When legibility is a primary concern, both should subtend visual angles of at least 16 to 25 minutes of arc. Characters should be in sans serif fonts, with character width-to-height ratios of 0.6:1 to 1:1. Further Analysis Needs: Control unit display label height and width in minutes of visual arc	See Table 3 for width measurements of selected control unit labels. Some of the control labels on the control unit panel fell below the minimum required size for legibility. However, when considered in conjunction with our task analyses, these inconsistencies with HF/E principles were not found to be a concern because spatial coding was inherent in the design of the control unit (i.e., controllers mapped the location of controls with their respective functions), and the tasks for which the controls were used were not time critical. Therefore, it was determined that RCOs could complete the tasks in a reasonable amount of time without looking at the control unit. For other time-critical tasks, such as applying an emergency break, redundant coding for spatial and tactile characteristics were in place, so the findings that labels for these controls were above the minimum size but below recommended size for legibility was also not a concern. Again, our team’s decision to consider issues other than human factors engineering practices (e.g., evaluating a device against established principles) was critical to conducting a comprehensive evaluation.
Psycholinguistics aspects
18-14. Display instructions should, whenever possible, be stated in a positive manner. Further Analysis Needs: Task analysis [Include all tasks involving status display messages] Control unit user manual	Our team identified one inconsistency with this principle. Specifically, the control unit status display presented the message, “Speed change without vigilance press,” when the control sequence required to move a locomotive from a stopped position was executed incorrectly. The best way to communicate a control sequence error is to alert the RCO to the error and state the appropriate actions in the order in which they should be carried out. Recommended: Change the message to “Press vigilance before changing speed.” This recommendation was passed along to the control unit manufacturer.
RCO mental model aspects
18-20. The direction of movement of an indicator on a display should be compatible with the direction of movement in the RCO’s mental model of the variable whose change is indicated. (Also 18-19) (FAA 7.1.3.10) (FAA 7.1.2.8) Further Analysis Needs: Task analysis Cognitive walkthrough Verbal protocol analysis Field observation When is the CHARGE setting on automatic brake typically used? [Interview with SME]	While conducting field observations, our team identified an inconsistency with this principle that arose when RCOs were facing one direction while the train was moving in another. Specifically, if the direction control is set to forward, the RCO cannot depend on the “pictorial realism” of the device to indicate a forward motion. Instead, the RCO must consult his or her mental model of the direction in which the train is facing relative to his or her own and determine if a mental rotation is required to move the train in the desired direction. As is true with any remote control device used in the real world, this mental rotation introduces cognitive load, and RCOs might be likely to avoid that load by relying on the literal interpretation of the device setting. There are no established design recommendations available to address this matter.
Control features aspects
18-22. Many coding schemes other than color - including shape, size, texturing, shading, and intensity - can be effective. These coding schemes are often not as good as color for speeding search performance. As with color, these coding schemes can convey a limited set of meanings in certain contexts.	Our team pointed out that the physical characteristics of the toggles were the same, but their functions were different. This did not strike us as a major problem, as the toggles could be distinguished by spatial location. Recommended: Further discriminability could be achieved by coding the controls based on size and/or color as long as no more than three sizes and no more than nine colors were used. (People have a hard time discriminating among more than three sizes and more than nine colors.)
Control and Display Standards (From the FAA)	Outcomes and Recommendations
Mode awareness aspects
FAA 7.4.1.1.5. Characteristics of common controls. The characteristics of different potential controls should be considered in the selection of a control for a given use. Further Analysis Needs: Task analysis Simulator scenario run through	Our team identified one toggle switch that served two functions: (a) Pressing the toggle switch away from the RCO extended the time in which RCOs were allowed to bend over to work on objects close the ground without the control unit applying the emergency brake to the locomotive. (b) Pulling the toggle switch toward the RCO enabled him or her to view current system settings on the digital status display. If RCOs were unaware that the control unit was in “system status mode” and pressed the toggle away from them with the intent of extending the time allowed to bend over (i.e., RCOs make a mode awareness error), then they might easily exceed the allotted time, and the locomotive would be thrown into emergency braking. Recommended: Design two separate controls for the time extension and system status functions. This recommendation was passed along to the control unit manufacturer.
Control features aspects
FAA 7.4.4.10.1. Toggle switch specifications. Further Analysis Needs: Remote control unit toggle switch measurements	See Table 4 for control unit toggle switch measurements. Spacing and physical characteristic measurements for control unit toggles ranged from meeting requirements to meeting requirements but only without gloves. This information was passed forward.

Note. Numeric categories follow Lehto and Buck (2008). RCO = remote control operator; FAA = Federal Aviation Administration; SME = subject matter expert.

Table 3.

Human Factors Design Principle for Label Character Size and Control Unit Label Width Measurements

Label	Width of Character: Requirement –Recommendation (in minutes of visual arc)	Width of Character: Actual (in minutes of visual arc)	Comments
General labels	16 – 25	21.60	The current general label character widths are above the minimum required size for legibility but below the recommended size for legibility.
Low battery (Label) (Detailed task analysis Task Numbers 1.1.4.4 and 1.3.4.3.)	16 – 25	14.32	The current low battery label character widths are below the minimum required size for legibility.
Radio signal (Label “TX”) (Detailed task analysis Task Numbers 1.1.4.3.1 and 1.3.4.2.1)	16 – 25	10.74	The current radio signal label (“TX”) character widths are below the minimum required size for legibility.

Source. Lehto and Buck (2008) 18-10: Width of label characters.

Table 4.

Human Factors Design Standard for Toggle Switch Specifications and Control Unit Toggle Measurements

FAA 7.4.4.10.1. Toggle Switch Specs	Spacing/Measurement Requirement	Actual Measurements	Comments
Spacing between toggles	3/4 in. (0.75 in.)	7/8 in. (0.875 in.)	The current spacing allows for 1/8 in. additional room for gloves
Spacing between toggle and obstruction	1/2 in. (0.5 in.)	1/2 in. (0.5 in.)	The current spacing does not allow for additional room for gloves
Length of toggle arm	1/2 in. (0.5 in.) Use with heavy hand wear: 1.5 in.	9/16 in. (0.5625 in.)	Requirements for toggle length while wearing heavy gloves is 1.5 in., so the current toggle length would not be adequate while wearing heavy gloves
Width of toggle tip	13/100 in. (0.13 in.)	1/4 in. (0.25 in.)	Meets requirement

Note. FAA = Federal Aviation Administration.

This phase also highlighted the iterative nature of the audit. In other words, if it was not possible to immediately evaluate a principle upon inspection of the control unit, the team identified methodologies (e.g., additional task analyses, time-and-motion studies, SME interviews) from previous phases that would be necessary to gather the requisite information for the audit.

For example, we discovered a problem with one of the displays when applying HF/E principles indicating that information displayed across like devices should be consistent with regard to the order and amount of information that is presented (Wagner et al., 1996). The status display, a scrolling display used to provide the RCO with low-level information about system states, provided different information categories across control units. We noted that the inconsistency in status information across control units might create a challenge for RCOs.

At one point in the audit, we noted the need for a time-and-motion study. We conducted a link analysis of SMEs’ interactions with the control unit using the RCO training simulator to evaluate the control unit against control sequence principles, such as “If controls are usually activated in a definite sequence, arrange them to follow that sequence” (Lehto & Buck, 2008, p. 713).

Figure 3 (page 11) shows a link analysis of the control unit interactions during flat rail yard operations with an emergency brake scenario included in the simulation. Our link analyses showed the proportion of control activations made by the RCO across the duration of the simulation and provided a graphical depiction of the sequence of control activations encountered during principal tasks. Each node corresponds to a specific control, and the node diameter indicates the total proportion of control activations made during the simulation. The links between nodes represent the order of control activations, with arrows indicating the direction of the activation sequence. The width of each link corresponds to the proportion of control activations for a particular sequence.

Figure 3.

Link analysis depicting control unit activations made by a remote control operator during flat yard operations, including a simulated emergency scenario.

Most large links in our link analysis were the loops indicating consecutive activations of the same controls, which were obviously not a concern. However, one large link that initially appeared to be out of sequence was the one connecting the speed control (located on the right side of the control unit) to the left vigilance control (located on the left side of the control unit) and back again. However, two features of the process suggested that the control unit design was fine in this regard. First, the operators typically use two-handed operations (left hand for vigilance, right hand for speed), and second, the right vigilance switch (which is functionally identical to the left vigilance switch) is located next to the speed control and therefore complies with the principle, should the RCO choose to use this redundant control, as when hanging on the side of a car.

Product of evaluation instrument

We arranged the identified HF/E issues in a matrix comprising 15 overarching human factors concerns and 21 control unit features. Some issues had been identified by Reinach and colleagues (Reinach & Viale, 2006) as occurring in other control units, but the vast majority of potential issues we identified using the Cattron control unit had not been articulated previously.

We reported all violations, prioritized them according to perceived importance, and made change recommendations for some. We did not view some violations as particularly problematic. For example, one standard states that lines associated with each control setting should be placed at the base of a rotary control device to better indicate the current setting of the control. Although the control unit did not have these demarcations, LED lights next to the control values, along with tactile information about the position of the knob, provided redundant feedback.

Some issues were particularly important to our CSX liaison, so we provided possible solutions for those issues as well. For example, he was interested in the possible ergonomic concerns resulting from the position and range of rotation of the rotary control devices. We investigated this issue further by referencing additional ergonomics standards, computing force measurements, and making suggestions regarding alternative control options.

Phase 5: Feedback

After reviewing our deliverable, the CSX liaison provided suggestions on tailoring the document for the industry reader. As a result of this feedback, we became more sensitive to legal issues, wording, document organization, and use of scientific jargon while writing subsequent versions. We also became aware of the effectiveness of photographs to accompany our description of the HF/E issues. For example, we used Figure 2 to illustrate the difficulty that some RCOs might face when using the control unit while riding the train.

As a proponent of HF/E research, our CSX liaison was influential in setting up a meeting with the manufacturer of the control unit. As a result of the meeting, we were asked to evaluate the new prototype that was nearing production deadlines; consequently, production was delayed in order for CSX to consider some of our suggestions. Because of time constraints, we were provided with images of the prototype in place of the actual model. However, in a matter of days, we were able to effectively use the evaluation tool we created to produce a short report identifying possible HF/E issues with the new design.

Final Thoughts

Even though several features of the control unit were not consistent with HF/E design principles, these findings were only one aspect of our overall human factors audit. A number of other considerations – for example, the way in which tasks are carried out by a legacy workforce using a legacy product, and the cost of implementing new technology – played a big role in deciding which design changes to recommend to stakeholders.

For example, one standard related to lifting items close to the body to avoid injuring the back was identified as a possible issue when the control unit was evaluated against the evaluation instrument in conjunction with our task analysis. However, follow-up interviews with RCO managers revealed that company policies prohibited the RCO from lifting objects while wearing the control unit, and procedures were in place to communicate these needs to other team members. This example illustrates the importance of the multifaceted nature of our audit. Had we not considered organization-level factors and simply relied on a human factors engineering approach, we would have reported an unnecessary recommendation.

In summary, we adopted a multiphase audit to evaluate a remote control device used in the rail industry and developed an audit tool by consolidating numerous HF/E principles and standards. We then applied this tool to the analysis through an analytical and empirical approach under significant time constraints.

HF/E advocates at CSX were absolutely critical in raising the visibility of the research team within the organization and helped to ensure the availability of important resources throughout the iterative process. Such access helped to improve the overall quality of the audit, such that the proposed design of a device would further take into account human capabilities and limitations. Ultimately we were able to take the tools developed over the course of months in the initial audit and apply those issues to the new prototype in a matter of days. Because internal HF/E advocates were able to convince decision makers in the company, some (not all) of our suggestions were incorporated in the new prototype despite production delays.

Footnotes

Joseph D. McDonald, MS, is a graduate student in the engineering psychology PhD program at the Georgia Institute of Technology. His research focuses on understanding how organizational, team-level, and individual-level factors interact with technology to contribute to systemic failures in dynamic environments. He may be reached at joe.mcdonald@gatech.edu.

Sadaf Kazi is a graduate student in the engineering psychology PhD program at the Georgia Institute of Technology. Her research focuses on understanding how cognitive variables, such as memory, decision making, and strategy use, interact with task variables, such as interruptions and time pressure, to influence performance in dynamic environments, such as health care and air traffic control.

Cale M. Darling is a graduate student in the engineering psychology PhD program at the Georgia Institute of Technology in the Cognitive Ergonomics Lab. His research interests include visual attention and situation awareness.

Francis T. Durso is a professor in the engineering psychology program at the Georgia Institute of Technology. He received his BS from Carnegie-Mellon University and his PhD from SUNY at Stony Brook. He is president of the Human Factors and Ergonomics Society and a member of the NRC Board of Human-Systems Integration. His current research interests focus on cognitive factors used to manage dynamic situations in domains such as aviation, rail, and health care. He may be reached at frank.durso@gatech.edu.

Kenneth A. Glover, MBA, MS, RCEP, CSCS, is general manager of Safety Systems and Human Performance for CSX Transportation. He serves as a council member on the Mayor’s Council for Fitness and Wellbeing for the city of Jacksonville, Florida; as a dean’s associate advisory board member for the School of Public Health at Indiana University in Bloomington; and as a board member for the School of Human Sciences at Florida State University.

References

Kirwan

Ainsworth

L. K.

(1992). A guide to task analysis. Philadelphia, PA: Taylor & Francis.

Lehto

M. R.

Buck

J. R.

(2008). Introduction to human factors and ergonomics for engineers. New York, NY: Taylor & Francis.

Reinach

(2004). Robot trains: Results of focus groups with remote control operators in the United States and Canada. In Wilson

Norris

Clarke

Mills

(Eds.), People and rail systems: Human factors at the heart of the railway (pp. 79–87). Farnham, UK: Ashgate.

Reinach

Acton

(2006). Remote control locomotive operations: Results of focus groups with remote control operators in the United States and Canada (DOT/FRA/ORD-06/08). Washington, DC: Federal Railroad Administration.

Reinach

Fadden

Gamst

Acton

Bartlett

(2006). A comparative risk assessment of remote control locomotive operations versus conventional yard switching operations (DOT/FRA/ORD-06/09). Washington, DC: Federal Railroad Administration.

Reinach

Viale

(2006). Application of a human error framework to conduct train accident/incident investigations. Accident Analysis & Prevention, 38, 396–406.

Salas

Wilson

K. A.

Priest

H. A.

Guthrie

J. W.

(2006). The design, delivery and evaluation of training systems. In Salvendy

(Ed.), Handbook of human factors and ergonomics (3rd ed., pp. 472–512). Hoboken, NJ: Wiley.

Sanders

M. S.

McCormick

E. J.

(1993). Human factors in engineering and design (7th ed.). New York, NY: McGraw-Hill.

Wagner

Birt

J. A.

Snyder

M. D.

Duncanson

J. P.

(1996). Human-equipment interfaces. Human factors design guide (HFDG) for acquisition of commercial off-the-shelf subsystems, non-developmental items, and developmental systems (DOT/FAA/CT-96/1). Atlantic City International Airport, NJ: FAA Technical Center.