2005-2006 Assessment of the Army Research Laboratory (2007)

The Panel on Soldier Systems reviewed a sample of programs of the Human Research and Engineering Directorate (HRED) of the Army Research Laboratory (ARL) during April 27-29, 2005, and May 1-3, 2006. HRED is organized as two divisions to conduct research and development efforts to enhance soldier performance. The Soldier Performance Division conducts a broad program of soldier-centered basic and applied scientific research and technology development, directed toward maximizing battlefield effectiveness. The Human Factors Integration Division conducts analyses and applied studies to ensure that soldier performance requirements are adequately considered in technology development and system design. Tables A.1 and A.2 in Appendix A show the funding profile and the staffing profile for HRED.

The framework for the assessment, as presented by the acting director, was HRED’s dual objectives of providing science and technology to enable transitional capabilities for the smaller, smarter, lighter, and faster future force, while also seeking opportunities to accelerate the insertion of technologies directly into the current force. Within this framework of dual objectives, the reviewed projects demonstrated a high level of scientific merit and relevance to the Army’s needs. The professionalism and enthusiasm of the HRED staff in presenting information and answering questions about individual programs and projects were also at very high levels. Although critical in the past about insufficient outreach, the Board is very impressed with the growing participation of HRED in the scientific and technical community and with the open dissemination of research results.

HRED has continued to adjust its programs to meet changing requirements for the Army’s transformation to a smaller, smarter, lighter, and faster future force. For example, as the Army moves toward

increasing levels of automation to meet its mission requirements, HRED has initiated efforts to address the enhancement of soldier performance in automation-related functions. A series of projects has been initiated in response to an Army Technology Objective in human-robotic interaction to address critical operator performance issues in the development and deployment of semiautonomous robots, such as unmanned air and ground vehicles. Studies have now been conducted to assess factors affecting control performance and situation awareness, operator performance and workload relative to the number and types of assets controlled, and the effects of team size on robot controller performance.

Continuing improvements were observed in the quality of, and the rationale for, experimental designs employed by the HRED researchers. One notable advance was the more extensive use of soldier-centered modeling tools during the initial phase of the research to analyze the work domain, operationalize hypotheses, and/or systematically select parameter values. This is evidence that the model>test>model approach that has been advocated by the Board is finding its way into the planning and conduct of experimental studies—that is, the development and use of soldier-centered models to assess the work domain and plan studies, to clarify and extrapolate the experimental results, and then to integrate the empirical results to further enhance the models. The use of this approach can become more widespread at HRED if those who conduct experimental studies develop a better understanding of the value and operation of available models.

The latest review revisited HRED efforts in the development, application, and refinement of soldier-centered design tools and analysis. The goal of these modeling efforts has been to develop and apply computer simulation methods of the type that can enable the analysis of human-hardware interactions in a variety of anticipated scenarios. These simulations can ensure the early insertion of human systems integration considerations, constraints, and measures into the acquisition cycle. The tools and their enhancements resulting from these efforts have also been useful in supporting HRED research efforts by mapping work domains to ensure that the most relevant tasks are addressed in research efforts and by integrating research results to enhance and further their application to operational systems. The core (and workhorse) of the soldier-centered design and analysis efforts has been the task-based Improved Performance Research Integration Tool (IMPRINT). Recent usability and redesign efforts have resulted in the development of two enhanced versions of the tool: IMPRINT Standard for the typical model user, and IMPRINT Pro for the sophisticated model user. The earlier gap analysis and the current stressors analysis have guided efforts to develop more effective approaches to adding IMPRINT functionality. Moreover, input from outside contractors and the modeling community has been employed effectively in the process of upgrading the model. These efforts have also been very responsive to Board suggestions, such as the need for improving visualization during model applications and increasing the extent of formal collaboration with academic, tri-service, and other government groups and laboratories.

The Board applauds recent HRED efforts to extend the utility and applicability of soldier-centered design tools by integrating them into a powerful, easy-to-use modeling platform for exploring systems concepts and designs. The effort consists of integrating two models with IMPRINT and developing the needed middleware and a unified interface, the Human Behavior Architecture (HBA), to facilitate both cognitive and manual task modeling. One model, Command, Control and Communications Techniques for Reliable Assessment of Concept Execution (C3TRACE), depicts human tasks with an information flow and net-centric focus. It is designed to evaluate the effects of organizational structure and

information technology effects on human-system performance. The other model, Adaptive Control of Thought-Rational (ACT-R), represents human cognitive functioning in simulations and assessments. The IMPRINT and ACT-R interaction and control are particularly critical to this effort, as is the need to model a proactive information-gathering capability. However, caution should be exercised with the ACT-R integration so as not to assume that this model is now a complete, unified theory of cognition.

In support of the Vice Chief of Staff, Army, Holistic Helmet Study Group, HRED conducted a set of quick-turnaround studies to assess survivability issues relative to helmet design, fit, wear, and compatibility factors. This study was notable, inasmuch as HRED completed four high-quality, definitive studies within a month that provided an extensive amount of revealing and useful data to the Army. Evaluation of the two helmets now in Army use showed a significant impact of design on the capability of sound localization, an important capability for threat identification and targeting and for maintaining situation awareness in the field. In addition, the two helmets differed greatly, relative to the amount of pain experienced from extended wear. Added to these results were findings that relatively few soldiers have helmets that fit and also that few soldiers wear their helmets properly, whether they fit or not. How the helmets fitted and how they were worn were also shown to affect the ability to sight weapons and use goggles. The extensive, detailed data collected in these studies supported in-depth analyses to answer the many questions raised by the principal results. The proposed utility analysis presented for review should provide a transparent approach to needs analyses and the future evaluation of helmet design alternatives, particularly when coupled with sensitivity analysis and within a holistic, systems engineering framework.

Research on auditory awareness and speech communication continues to be an exemplary program at HRED. The program demonstrates an insightful anticipation of future Army needs and the development of applied research that is cast in terms of basic research requirements in which research questions are effectively formulated and research designs are appropriate to well-articulated needs and objectives. Moreover, the program is being conducted within the context and understanding of related work that is being conducted in other laboratories of the Department of Defense (DoD). In addition, there is an impressive list of university partnerships, industry cooperative relationships, and links to related government and international research programs. The assessment covered a selection of projects from a much broader program of research.

The research on spatial audio with a bone-conduction interface provided very encouraging results for sound localization via bone conduction—bone conduction audio, combined with head-related transfer functions, produced absolute localization errors no greater than audio from standard headphones. While the results so far are limited, the direction for future research is clear and critical—to examine sound localization performance with transducers placed on locations of the skull other than the mandibular condyle. Another promising line of research in sound localization involves determining the effect of sound duration on localization accuracy. While this preliminary study was also considerably limited in scope (the listener’s head movements were restricted, the sound durations were very short, and the experimenters had not yet had an opportunity to replicate results with a paradigm that requires subjects to perform meaningful additional activities other than listening alone), it provided clear direction for further research. The Board recommends experimental designs in which head movements are measured during signal localization, and analyses that address the extent to which degree and type of head movements predict variations in responses. The Board believes that the measurement of head movements would be relatively easy and inexpensive to implement.

The Human Factors Integration Division is providing a growing amount of support to Army customers (more than 150 projects each year), influencing the design and fielding of systems to enhance soldier performance. These efforts are typically conducted under the structure of the manpower integra-

tion (MANPRINT) program that addresses the integration of manpower, training, personnel, human engineering, safety, health, and survivability. An outstanding example of this type of effort was the field evaluations recently completed by the HRED Communications and Electronics Command (CECOM) Field Element at Fort Monmouth, New Jersey, of the hand-held Global Positioning System (GPS) receiver. The poor human factors design of this system had led to, among numerous other incidents, an incident in Afghanistan in which three Special Forces soldiers were killed and 20 injured when they mistakenly called fire on their own position. The study resulted in the identification of 1 critical, 5 major, and 13 minor design problems in the operation of this device, together with associated redesign recommendations.

As the lead for the Human Centric-Network Enabled Battle Command (HC-NEBC) project, HRED is undertaking a very ambitious and challenging effort. The project utilizes the Modeling Architecture for Technology, Research and Experimentation (MATREX) environment to launch the development of an integrated collection of models, simulations, and tools that can be employed for analysis, experimentation, and technology trade-off studies. HRED heads a team consisting of the contractor, DCS, the MATREX infrastructure team, the Training and Doctrine Command (TRADOC) Analysis Center, the Simulation and Training Technology Center (STTC), and the Army’s Aviation and Missile Research, Development and Engineering Center (AMRDEC). The Board identifies the following as some of the major challenges: designing and implementing the middleware required for integrating the various components; addressing the complex conceptual framework as suggested by the behavioral terminology being employed (e.g., role, task, service, behavior, activity); addressing active as well as passive processing of information; implementing the needed feedback structures; and developing, implementing, and validating the behavioral models. Program success may require focusing, initially, on incremental system development—starting, for example, with a simple command structure and behavioral model and building on the workload scheduling algorithms being developed for IMPRINT.

A recently initiated modeling effort addresses the intelligence, surveillance, and reconnaissance (ISR) functions that support the military decision-making process. The objective is to adapt IMPRINT to answer critical organizational and human performance questions imposed by new force structures and technologies. The Board believes that the effort must go beyond IMPRINT to consider the incorporation of appropriate cognitive models. While ACT-R has an empirical base, the model is not suitable for decision-making tasks involving these types of inputs; consequently, adapting or developing an appropriate cognitive model is likely to be a major challenge. The Board suggests that it may be advisable to mitigate the challenge at the start by limiting the source of intelligence being modeled to the most tractable, such as sensor data. Since this research is forward looking, the current emphasis on the non-urban battlefield environment may not adequately reflect the future needs of the military.

HRED is engaged in a joint effort with the Night Vision and Electronic Sensors Directorate (NVESD) of the Communications and Electronics Command to provide 24-hour mobility vision for dismounted soldiers. The goal is to provide adequate situation awareness, target detection, and obstacle avoidance when traversing field and urban terrain on foot. The principal focus of the research is on performance with a sensor that is capable of providing both thermal and intensified images in various selected combinations on a single display. The program has made effective use of static experimental tasks to answer some of the preliminary questions concerning the effects of sensor offsets and image fusion algorithms. However, definitive results to support the fundamental research objectives must necessarily employ dynamic, real-world tasks in the future. In real-world situations, for example, the two sensor modes

may not reliably produce the same setoff contours, leading to situations in which the two images are not readily reconciled or fused. Consequently, one needs to address the possibility of sensory rivalry or sensor veto, where one source is given dominance over the other.

The many customer-driven reactive research projects conducted each year (more than 150), such as the GPS study noted above, provide a long-term opportunity for HRED to become more proactive in providing customer support. Such an approach would entail the development of standardized approaches (e.g., common measures, systematic variation of parameters) to the design and conduct of studies, leading to the compilation of the data from multiple studies in a form that would permit meta analyses (such as multiple regression) across studies at a later time. Such an approach would also actively seek opportunities to incorporate additional, low-cost, potentially useful measures in these studies, even though not required to meet the original objectives of a study. The goal would be to develop resources such as additional, more specific models, rules, and guidelines that can be applied immediately in response to customer needs, reducing the need for lengthy field studies and analyses. It may also be worthwhile to increase designer access to certain human factors modeling tools as another means of supporting Army customers. A user-friendly Web-based interface that allows designers to use rudimentary human factors modeling tools during the operational concept exploration phase could result in human factors considerations being included earlier in the system design cycle. This proactive approach to customer-driven research would be facilitated by concerted activities that integrate the plans, approaches, and results of various HRED projects.

An axiomatic principle of human factors research and design is that the development of equipment and systems is most efficient when human factors principles and guidelines are considered early during the determination of system requirements and the specification of system functions and performance parameters. The Board realizes that the ARL mission does not normally include the actual design of equipments and systems. HRED research projects do produce data that readily can be transformed into a relevant set of human factors needs and design recommendations and, in turn, into recommended designs or design modifications. It is the view of the Board that the ultimate impact of human factors research may be considerably greater if HRED actually took the next step, when appropriate, and generated model design specifications and design exemplars as products of its efforts. For example, both the GPS and helmet studies referenced earlier produced sufficient design-relevant data to support such efforts; the research results plus application of the proposed utility analysis could be the basis for a model ARL helmet. This is not an approach that would be totally new to HRED. In a project reviewed earlier on Army fixed-bridge construction, HRED developed short-term solutions to problems identified during its innovative use of ergonomics analysis tools to document the biomechanical stress and physical hazards imposed when lifting bridge components using existing manual methods. HRED developed prototype hand tools and portable hoist and cart systems from the results of its studies for further testing and refinement in the field.

A battlefield testbed at Fort Dix, New Jersey, was the setting for a field study to determine the impact of a digitized command interface on situation awareness and workload at the platoon level. This study demonstrated the challenges of fielding computer support systems and conducting research to assess their impact: there were frequent problems with communications networks; equipment usability problems led to distractions and frustrations; soldiers had not developed proficiency with tasks because they had received only classroom training; and there was an inability to freeze action in the field to enable the completion of data collection instruments required by the experimental design. As a consequence, the research results were not definitive and were difficult to interpret. While the Board appreciates the challenges of conducting field studies that rely on coordination with and support of others, it believes that in planning complex field studies such as this, adopting its previous suggestions for the use of more

intensive peer reviews during the planning of projects would serve to anticipate and help avoid problems of the types encountered.

In general, HRED programs demonstrated excellent anticipation of future needs, research properly cast in terms of fundamental requirements, and well-constructed experimental designs with appropriate data analyses. On the other hand, some programs would benefit from more formalized research plans with better-articulated goals and objectives. As the Board has suggested previously, research planning probably would benefit from more intensive internal (and possibly external) peer reviews of research objectives and of the methods proposed for the conduct and analysis of research projects. Such a practice would leverage the considerable expertise of senior scientists and, hopefully, lead to the development of better theoretical constructs, the formulation of more appropriate research hypotheses, and the construction of more powerful experimental designs and statistical analyses to better support the resulting assertions and conclusions. This in-house review program would complement the desirable practice of peer review through journal publications and presentations, where feasible.

As modeling development efforts move forward, there will be an increasing need to develop and apply effective processes for validating model output and archiving validation results, particularly with tasks that have a less structured workflow, such as intelligence, communications, and maintenance. The Board envisions a verification and validation process designed to assess the underlying assumptions of the model and to identify the potential advantages and limitations of a particular model. The process would encompass the collection and use of “ground-truth” data where appropriate. In this regard, the Board recognizes that, for many model applications, there is no ground truth; moreover, there may be no funds to collect and archive ground-truth data when they are available. Unlike a statistical modeling situation in which there are training and testing data sets, the modeling conducted by HRED often requires going beyond any observed reality. Consequently, an analytic validation approach may be employed, rather than empirically-based ground-truth comparisons.

Even so, HRED should recognize the value of and be alert to opportunities for establishing some type of ground-truth database (consisting of previously collected data for which correct outcomes are known and previously published data that have been incorporated into the model) to assess the value of model improvements and enhancements. Such a database would need to be both comprehensive, in terms of sampling key system dynamics, and ecologically sound in order to help prevent spurious model validation. In this context, a formal process for assessing how the models have been used to guide further developments could be helpful. Such a tracking system could document successes as well as help prevent model improvements in one dimension of the system at the expense of predictive capability in other important dimensions. The bottom line is that the user will then know the extent to which the model produces the true result when that result is known, thus providing an appropriate level of confidence in its application when truth is not known. One approach to using a ground-truth database consists of a train-the-model phase and a test-the-model phase. In the training phase, one set of data is employed to adjust the parameters of the model so as to obtain the correct outcomes. In the testing phase, another comparable set of data is employed to cross-validate the results to see if the correct outcomes are obtained on data other than those on which the model was trained. Positive results provide confidence that the model is both valid and robust.

The Board would like to see the coherence of the sensors and perception research program enhanced by a formalized research plan with specific goals and objectives, relating basic and applied research efforts, and showing a more active connection with the broader vision of the research community. For example, the military is fond of the idea that auditory and tactile channels are underutilized. While it is plausible that these modalities could be used for alerts, the Board worries about plans that suggest that information can be handed off to another sense when vision is overloaded. There are “central executive” limitations that create bottlenecks between the senses as well as bottlenecks within a sensory modality. For example, one cannot read this report and listen to speech at the same time without sacrificing speed or comprehension or both. The Board recommends, in addition, that those working in the area of sensor fusion review papers on binocular vision and cue combination that have been presented at meetings of the Vision Sciences Society.¹ These papers are not specifically about sensor fusion, but much in them is relevant to this research. They also identify a number of researchers worthy of being contacted on these issues.

Feedforward control (or a forward model) is a concept derived from control theory to define a process in which a system takes action in advance of error caused by a disturbance or demand on that system. This proactive manner of control is superior to compensatory forms of control that wait until an error state exists before corrective action is taken. Feedforward control requires that some aspects of the input disturbance be measurable, the effect of the disturbance on system output be known, and the feedforward control mechanism have the capability to counteract the effects of the disturbance. Taking this concept from simple, tightly controlled systems (like advanced temperature control) and applying it to the complex world of command and control is a sizeable challenge, but one that HRED is undertaking. Specifically, feedforward control was studied to understand its importance in command predictions, learning, and performance in a game-based exercise. While the Board applauds HRED for undertaking this line of research with this limited first step, further efforts will require a more clearly defined and operationalized construct for feedforward control in the command and control environment. For example, feedforward control in command operations must be clearly distinguished from what may otherwise be considered simply a problem of information presentation or training.

As stated earlier, the design and conduct of experimental studies at HRED are showing continuing improvement, leading to projects that are producing definitive, relevant, and useful research results. The next step in enhancing the utility of experimental research will be to employ deeper analyses of behavioral measures. For example, many of the studies presented during this assessment employed factorial designs that were analyzed by means of analysis of variance. While these analyses are very useful in determining the main effects and interactions among variables, they have limitations when attempting to apply the results to designs or decisions. Designing the studies so that regression analysis is also possible may provide results in a more usable form that would reveal graded effects of varying design parameters. Other avenues to deeper analysis include process tracing techniques to better understand underlying strategies, defining key aspects of situation awareness necessary for mission success, analyzing and displaying variance to better understand measurement reliability and causes of performance outliers, and further assessing experimental results by comparing them with other studies completed under comparable conditions. Deeper analyses, for example, could have been applied to the well-executed study of the effects of latency and levels of autonomy on the control effectiveness and situation awareness in human-robot interaction. With four latency conditions applied in two different modes, four levels of autonomy, two measures of situation awareness, and two dependent measures of performance, there are limitations to analyzing and presenting the results solely in a series of bar

charts. It would be helpful to present these results in a more integrated and meaningful way, and this also would help clarify the limitations of experimental findings and conclusions. Using the experimental data to estimate parameters of a process model, such as one developed with IMPRINT, could effectively integrate and extend the experimental results.

More effective and complete analyses of work domains during the preparation and design of experiments will enhance the future interpretation and generalizability of study results. For example, in a study that attempted to determine the effects of number and type of robotic assets employed on the target detection and acquisition performance of human controllers, findings indicated that performance was best with one unmanned aerial vehicle (UAV). However, the results could be explained simply by the nature of the task—the target detection and acquisition task selected for this experiment could have been most effectively performed by the UAV alone, without the need for any other assets. That is, resource allocation by the controller may have been task dependent. An effective approach to defining the work domain, structuring the experimental approach, and selecting appropriate tasks for experiments is to employ a soldier-centered tool such as IMPRINT during the planning of the study. The Board is pleased to see this approach being utilized with some of the studies that are currently in the planning stage, particularly in the research being initiated to address adaptive automation in human-robotic interaction. This is another example of the potential utility of the model>test>model approach discussed earlier. A more complete analysis of the work domain also would benefit the study of the manning requirements for UAV systems.

HRED maintains an understanding and appreciation of Army needs through extensive contacts with soldiers through its various field elements, through feedback from soldier participants in its studies, and through analyses of databases and exploratory studies related to soldier issues and performance. As a consequence, the projects reviewed were determined to have direct relevance and the potential for significant contributions to Army needs.

The Board is concerned, however, that HRED is not looking ahead far enough during the formulation of its research strategies, objectives, and plans. Many of the research efforts continue to be conducted within the “battlefield” context, even while prominent military strategists and analysts are projecting very different scenarios for the future—scenarios for urban and asymmetric warfare, for example. A broader range of scenarios would enhance the generalizability of findings to current and projected missions and would enable HRED to better integrate the findings from its varied studies. One member of the Soldier Systems Panel recently discussed this issue with the Deputy Director of National Intelligence for Analysis. The Deputy Director made the point that the extension of current trends (for example, increased globalization, communications flow, opportunities for terrorism) will continue to blur the line between personal security and national security, which in turn will further blur the line between law enforcement and military operations, and between activities involving people and those involving territory. As a consequence, research that focuses on “blue forces” against “red forces” arrayed in the countryside may not be particularly useful in addressing problems of the future.

Among other efforts, HRED has supported the Army transition to Future Combat Systems (FCS) for several years now by conducting critical soldier workload analyses. Currently, these efforts are directed toward using modeling tools to identify potential high-workload and multimodal interface task combinations for soldiers operating military ground vehicles (e.g., to identify crew-size requirements) and robotic assets, and they are being conducted in collaboration with the program of research on robotics. The combination of Unified Modeling Language (UML) to define system function allocations and IMPRINT

to define crew function allocations is very promising, as is the application of modeling to assess the value of auditory and haptic interface alternatives. Some of the potential limitations of this approach are the ability to recognize when tasks are synergistic or share a common mode so that they each demand fewer resources than when performed alone; the ability to handle ill-defined tasks; the capability of considering the consequences of future tasks in scheduling; and the accommodation of the full range of factors influencing workload—such as vehicle motion and robotic monitoring and control.

Future Combat Systems and associated future networked systems-of-systems will intensify the need to make critical design decisions relative to the roles of automation and supervisory control. Thus, HRED should be at the forefront of anticipating and conducting research in this area. The case study of fratricide involving the Patriot missile system, along with statistical data on incidents of fratricide, provided an excellent historical perspective on these problems and their importance to the Army. Researchable issues that should be aggressively addressed by HRED include those involving appropriate organizational structures, configuration management and placement, training requirements and system understanding (including the degree of knowledge of operational systems functions and error recovery), calibration of trust, issues involving situation awareness, and support requirements for decision making. The Board does understand that HRED is addressing some of these issues in its ongoing programs of research on adaptive automation and robotics.

Because of the importance to the Army of having trained personnel available in its critical specialties, the macro-ergonomic approach taken by HRED to reducing attrition during highly demanding military training is a significant contribution to the Army’s needs. The study demonstrated HRED recognition of a significant military problem and also demonstrated the challenge of addressing complex problems to accommodate immediate demands while also working toward a systemic solution. The development of a proactive computer-based assessment and decision tool, employable across a range of specialties, to analyze trainee characteristics and prescribe programmatic interventions and changes, is likely to be a valuable contribution to the solution of attrition problems in the Army. The complexity of the interactions between the many psychological and social factors, however, may require a team with a broader range of expertise than is currently being utilized.

The HRED program Cognitive Foundations of Performance in Military Environments is addressing issues that are important and central to soldier performance. For example, one line of research in this program addresses the effects of combat stress on the processing of information presented visually on digital displays. Typically this information is presented on small displays in very close or overlapping spatial proximity; the ability to filter out distracting information and attend to important information is critical to effective performance in many common operational tasks. Researchers have taken a first step in understanding the effects of stress in this work domain, but have much work yet to do to obtain definitive results that can be applied to Army operations. Some important factors to consider in designing further experiments include incorporating control conditions to resolve potentially ambiguous results; obtaining continuous psychophysiological measures during task performance; obtaining multiple dependent measures, including physical responses; and employing tasks that incorporate elements that relate to actual operations, such as more representative test periods (e.g., length of task, time of day). The program is also addressing multitasking, an increasingly important aspect of Army operations, and has completed the first in a series of studies—factors affecting multitasking performance. Additional studies will address predictors, strategies, limitations, and teamwork in multitasking. There is an opportunity in this research to employ some of the deeper behavioral measures, particularly process tracing, control issues, variability and reliability analyses, discussed earlier.

As the Army moves to the more extensive use of semiautonomous vehicles and considers various platforms from which human operators control them, the effects of interactions with these environments

on soldier performance become critical. HRED has developed and is employing a Crew Integration and Automation Test Bed Advanced Technology Demonstrator (CAT-ATD) to study the factors that are likely to degrade performance, as well as possible measures to mitigate degradation effects. When a moving vehicle is being controlled from another moving vehicle, for example, factors other than controller performance come into play, such as the likelihood that soldier controllers become incapacitated from a combination of motion sickness and control or display disparities. Research reported during this review determined that in such an environment, incapacitation is a significant problem and that future tactics should consider ensuring that the operator be able to disassociate the motion cues on the robotic display from those of the operator’s platform. Future simulator studies will examine the specific factors that are likely to mitigate the influence of the motion environment.

In previous reviews, the Board has emphasized the need for greater HRED interaction with the external research community and has encouraged researchers to share the results of HRED projects and programs through increased publication in the scientific literature and presentations at important professional meetings. HRED has continued to move in this direction, with contributions to the broader community now at a new high and likely to increase. Some examples of contributions, collaborations, and outreach efforts are provided in this section of the report.

A study of audio cues to assist soldier visual search of robotic system displays was launched from a review of previous research conducted elsewhere that suggested the potential utility of audio spatial cues in guiding the search of map displays. The study was conducted with the collaboration of researchers at Rensselaer Polytechnic Institute, who will perform a follow-up study as part of a cooperative research and development agreement. A journal article based on the research is being prepared. The HRED research program on multitasking performance extends from fiscal year 2003 through fiscal year 2007. In addition to the dissemination of results to the U.S. Army by means of ARL technical reports, results will be disseminated to the DoD Human Factors Engineering Technical Advisory Group, Harford County Emergency Operations Center (a study participant), Air Force Research Laboratory, NASA-Johnson Space Center, and the National Space Biomedical Research Institute.

The research program on the modeling of visual search tasks has involved the participation of researchers beyond HRED and is designed to make contributions to the broader modeling community, particularly those involved with modeling efforts that have come to be known as ACT-R. In the conduct of a project on memory for unattended information during visual search, data were obtained for the research from Harvard Medical School, researchers at Rice University contributed to the design of the search task, and a postdoctoral student (now a permanent employee at HRED) participated in the conduct of the research. While the Army is the primary recipient and beneficiary of the results of this research, the findings will help to inform the further development of ACT-R. An article is in preparation for publication in the Journal of Experimental Psychology.

Undertaken to better understand and enhance the interactions of multiple soldiers controlling multiple semiautonomous unmanned ground vehicles (UGVs) during military operations in urban terrain, the HRED research program is being conducted in conjunction with the University of Central Florida. At this time, one book chapter, five conference proceedings papers, and two conference presentations have been published, are in publication, or have been accepted.

Crosscutting issues on modeling and robotics should be relevant to this directorate. As discussed previously, the development and application of soldier-centered design tools continue to be vital to many programs at HRED. The HRED researchers are cognizant of the need for verification and validation, the need to relate their programs to those of others, and the extent of collaboration required to meet their objectives. The recently initiated Army Technology Objective at HRED on human-robotic interaction and the amount of research now underway in this area suggest the potential importance of pursuing crosscutting issues with other researchers addressing the application of robotics in the Army.

This page intentionally left blank.