2013-2014 Assessment of the Army Research Laboratory (2015)

4


Information Sciences

INTRODUCTION

The Panel on Information Sciences at the Army Research Laboratory (ARL) is charged with reviewing ARL research in the broad areas of autonomous systems, atmospheric sciences, computational sciences (including high-performance computing), and network sciences. The panel reviewed the ARL work in autonomous systems at Adelphi, Maryland, on August 13-15, 2013; the work in high-performance computing at Aberdeen Proving Ground, Maryland, on June 23, 2014; and the work in atmospheric sciences, computational sciences, and network sciences at Adelphi, Maryland, on June 24-25, 2014.

Autonomous Systems

While there was considerable variation in both the quality and impact of the research presented, the researchers were largely aware of the progress in their fields, and that had a noticeable impact on their own work. ARL has recently recruited a number of very promising early-career scientists. Careful attention needs to be directed at ensuring that they receive appropriate mentoring and career development opportunities as they develop their individual research portfolios. The new indoor Military Operations in Urban Terrain (MOUT) facility was impressive and will go a long way toward furthering the goals of the intelligence and planning program. The tour of the ARL Sensors and Electron Devices Directorate’s (SEDD’s) Specialty Electronic Materials and Sensors Cleanroom research facility helped the review team understand the infrastructure support available to ARL researchers.

A summary assessment of research in each of four subject areas—manipulation and mobility, perception, robotic intelligence, and human–robot interaction—is presented in the following sections of this chapter. ARL has a leading program in the area of small-scale robotics. A demonstrated ability to design, fabricate, and test these devices gives it a place of distinction in this field. Similarly, research in the area of perception is being performed at a high level of quality. With a mission to develop machine

understanding of objects, actions, and interrelationships in a specified environment, this work is critical for advancing the state of autonomous systems. Ongoing research is focused on advancing unsupervised approaches to human detection and advancing sensing and perception capabilities on constrained platforms. Research in the areas of human–robot interaction and intelligence is addressing important problems of mapping, cognition, and communication, as well as issues related to trust in autonomous systems. This research is cutting edge and comparable to work at federal, university, and industrial laboratories here and abroad, and portions of the work are poised for successful transition to applied research.

Atmospheric Sciences, Computational Sciences, and Network Sciences

The Army Research Laboratory Technical Assessment Board (ARLTAB) was tasked with evaluating the quality and impact of the ARL research program. A backdrop for this assessment was the aspirational goal outlined by the ARL director: pursuing research to support the U.S. Army to fight the war after the next. This statement speaks to a focus on fundamental research that may not have near-term applications or directly support the immediate needs of the U.S. Army. The ARL research agenda in the subject areas reviewed is both broad and diverse. The ARLTAB did not focus on the appropriateness or scope of the research agenda but noted that it needs a description of the context in which the internal research program was structured.

The linkages between in-house research, the research of external partners, and collaborative efforts are key in assessing the impact of ongoing work. This aspect could be considered for future reviews. Additional desirable adjustments to the presentation format including the following:

• Clearly articulate the problem at the beginning of each presentation. Whenever appropriate, a mathematical statement of problems is desirable.
• For continuing programs, identify how the program has changed since the previous review and indicate how recommendations from that review have been incorporated into those decisions to change. This was not a uniform practice in all presentations. Including a timeline that indicates the stage of completion for a project would also be helpful.
• Include benchmarks against which research is being calibrated. Research presentations need to identify what is unique and what new knowledge has been generated.
• Wherever possible, identify quantitative metrics against which research is to be assessed, and describe how these metrics challenge existing limitations of the status quo.

These comments notwithstanding, the work that was reviewed was of good quality. There was considerable variation in both the quality and the impact of research, but the researchers were largely aware of the progress in their fields. Related work was cited, and in some instances this awareness had a positive effect on their own work.

ARL has recently recruited a number of promising early-career scientists, and it will be important to ensure that they receive appropriate mentoring and career development opportunities as they develop their individual research portfolios. The ARLTAB panel’s visits to laboratories and research facilities at the ARL were useful for the assessment process. Members of the computational sciences assessment team had the opportunity to visit the high-performance computing facilities at Aberdeen Proving Ground and to see live demonstrations of some aspects of the ongoing research. Additional visits to select facilities will be helpful to future review panels.

All elements of the research program that were assessed have continued to progress in the quality of their work, the dissemination of results in distinctive publications, and apparent overall morale. The

ARL was responsive to the earlier reviews and recommendations of the ARLTAB. This includes both recommendations related to specific projects and recommendations on how to organize their activities to enhance their research capability and improve its visibility. As an example, the Computational Sciences Division (CSD) has better integrated its research on multiscale materials modeling with experimental validation and has developed a research roadmap for its work on tactical high-performance computing (HPC) that involves short- and long-term objectives. It has also increased the number of Ph.D.’s on its research staff and the external visibility of its research by increasing the number of conference and journal publications resulting from this work. Most presenters did an excellent job of articulating connections to Army operations.

A number of the exciting new research thrusts in Battlefield Environment Division (BED) study areas appear to be in response to the ARLTAB recommendations during the 2012 review. In support of these thrusts, and faced with flat or declining financial resources, BED has creatively expanded the number of postdoctoral researchers, visiting scientists and engineers, and students. This new expertise has enabled the division to pivot rapidly to exploit new research opportunities. Significant progress has been made on research projects evaluated during the 2012 review, particularly in boundary-layer research (study of particulates and aerosols, optical turbulence, and measurements). BED has successfully acquired and adapted powerful existing software tools for the verification of its numerical prediction models and has made significant advances in its suite of multiscale models.

In network sciences, there has been significant improvement in the quality of the research and the presentations and in the morale of the team. Almost all presenters did an excellent job in articulating connections to Army operations, although some presentations lacked clear problem definitions. Some of the weak projects from earlier reviews have now been eliminated. One issue from the previous review that may not have been addressed is the implementation of a support system for promoting recognition of researchers in professional societies. This needs to continue to be a priority for ARL management.

ACCOMPLISHMENTS AND ADVANCEMENTS

Autonomous Systems

All elements of the autonomous systems research program at ARL have continued to show progress, both in the quality of work and dissemination of results in high-quality publications. The program focuses on mobility and manipulation of robotic devices and on technologies that improve the usefulness of these devices, such as intelligence, perception, and improved human–robot interaction. The ARL research program is part of a larger collaborative effort involving external partners. A better definition of the role of the internal research in the overall program goals and of continued collaboration with partners is strongly encouraged.

Manipulation and Mobility

Research in the area of manipulation and mobility is closely linked to the ARL’s Collaborative Technology Alliance (CTA) in Autonomous Systems,¹ where significant collaboration with those partners is to be found. Three areas were highlighted during the review: replicating locomotion found in biological systems to improve robot mobility, autonomous manipulation of robots, and piezomicroelectromechani-

_______________

¹ There are currently two active CTAs related to autonomous systems: Micro Autonomous Systems and Technology (MAST) and Robotics.

cal systems (piezoMEMS) technologies to develop small-scale robotic systems. New updates on the project related to the CANID robot,² mimicking the movements of a canine hound, were presented. The primary thrust of the new development was the addition of a flexible spine to the robot. This is a challenging and interesting project, and the addition of this key degree of freedom to a walking robot is a good idea, because it more closely resembles the complexity found in nature. There needs to be a concerted effort to better understand the physics of this machine. Researchers have proposed low-order models for the system, and it might be fruitful to continue this line of inquiry. Good modeling will be imperative in any numerical simulations required to explore gains that are possible and to guide the focus of new experimentation.

Work related to self-righting robots is of a very high caliber and also has direct applications in the field. This was evident from the fact that the project was conceived through interactions with soldiers who are often confronted with the task of retrieving immobilized robots in combat. The research seeks to develop solutions for a broad class of physical conditions that affect stability of mobile robotic platforms. The current focus is on examining the underlying mechanical issues of self-righting in a quasi-static environment. It was not clear how the upright, stable position would be sensed on a sloping surface, where self-righting is most likely to be needed. ARL’s consideration of dynamics in 2014 is applauded.

The piezoMEMS research and associated small robotics effort are first rate, with elements that are at the vanguard of this field. The robotic devices under development with integrated piezoelectric materials demonstrated work that is at the forefront of MEMS design, fabrication, and experimentation. Specifically, the work in motion generation at the MEMS scale is seminal. Large-amplitude motions are being created at the micron scale using integrated actuators, structures, and electronics cofabricated on silicon. Techniques and approaches to generating articulating limbs with integrated flexure hinges and actuators represent advances in the engineering of MEMS technology. This has broad implications and applications to numerous MEMS systems—for example, MEMS-based microscale sensors and instrumentations such as mass spectrometers on a chip, drug delivery systems, and chemical assay analysis, where controlling microfluids is critical. The work being performed by ARL in the piezoelectric actuation of MEMS will impact more than just the creation of bioinspired microscale robotic systems.

Perception

The research in perception is of a high caliber. The work focuses on developing techniques that allow for describing the robot’s environments from sensor data. While there has been considerable progress toward describing environment for the purpose of mobility, deriving higher-level descriptions such as subtle cues and references that distinguish different behaviors and intents, recognition of specific classes of objects and features that are directly relevant to tactical behaviors, and labeling of object, feature, and terrain classes remain a challenge.

The current research plan is focused on three critical areas: perception on constrained platforms, robotic intelligence, and human–robot interaction. As for sensing and perception on constrained platforms, the scale and size of the platforms being explored in the autonomous systems enterprise pose technical challenges in sensor design. Sensors have to deliver the requisite accuracy and precision for surveillance and navigation, but they also have to reconcile with power limitations on smaller platforms. Other problems being addressed in the perception area include detection of humans in still images and strengthening object and material recognition capabilities, including an ability to recognize actions and

_______________

² CANID is a quadruped designed to test hypotheses regarding dynamic bounding using an actuated compliant spine mechanism.

imminent actions. The latter is based on scene parsing and action grammar and represents an interesting approach. The problem of developing real-time human detection algorithms is important, because existing approaches are based on supervised learning techniques that are computationally cumbersome. In operational environments that may be diverse and exhibit large variations, computational efficiency is an important consideration. The unsupervised learning approach used in this work is more efficient but may still yield false positives, and additional work is required to overcome this drawback. The basic approach for each of the research tasks is fundamentally sound, breaking new ground. Results of this research are publishable in archival literature, and the work is on a par with research being done at universities and other laboratories. The researchers are aware of related work being done elsewhere and recognize deficiencies in their individual approaches. Each presented a good plan for the future activities.

Robotic Intelligence

The primary accomplishments in robotic intelligence are advances in mapping capability, control for communications, and cognition. Much of this work is being published in top journal and conference venues (including the International Conference on Robotics and Automation, the Institute of Electrical and Electronics Engineers Proceedings, and the International Journal of Robotics Research), which attests to the overall quality of the research. Former students funded by the CTAs have been recruited to ARL and are important contributors to the research effort. Some of the 6.1 (basic) research projects are also making their way to 6.2 (applied) research, an important step before transitioning this work to the field.

In long-duration, three-dimensional (3D) mapping and navigation, the principal focus is on the development of a laser-based approach to 3D mapping that combines features from three existing mapping techniques. Demonstrating the effectiveness of the approach by deploying it on physical robots is an important accomplishment. Another thrust of intelligence research is on developing robust methods for control of mobility and communications. The primary focus of this effort is the development of a centralized, optimal solution of mobile node positions, subject to point-to-point communications constraints. While promising results were presented, additional details are necessary, particularly for issues such as scalability and whether such issues impose limitations on the proposed approach. Research in intelligence is also examining new cognitive architectures for robotic control. This work is focused on developing a new cognitive architecture that combines long-term memory, working memory, and perception. The mapping problem is driving this development, but it is difficult to understand how the cognitive approach improves mapping performance.

Human–Robot Interaction

In the area of human–robot interaction (HRI), research at ARL is looking at design issues for safe operation of autonomous reconnaissance systems in complex environments. The emphasis of this effort is human factors experiments to investigate interaction with, and control of, multiple autonomous systems. The design of interfaces is an important aspect of this investigation. Studies are focused on graphical user interfaces (GUIs), multimodal interfaces (including voice), and telepresence with stereovision and haptic interfaces. The experimentation conducted at Fort Benning has yielded an important basis for making design decisions. For example, experiments have demonstrated voice commands to be suitable for discrete actions but less so for controlling continuous processes. Similarly, the research has demonstrated how audio cues in 3D improve situation awareness in telepresence tasks.

The RoboLeader project continues to be an important component of the ARL HRI program. The research draws on a large body of experimental work related to evaluating the effectiveness of autono-

mous and semiautonomous control of teams of robots. The RoboLeader intelligent system was evaluated in this context and shown to provide benefits from the standpoint of both task performance and workload management. Testing with humans showed individual differences in performance: People with good spatial ability and significant videogame experience had better situation awareness of the mission environment. The results have implications for personnel selection and training and for user interface design.

Another research thrust in the HRI arena is bringing greater understanding of automation actions to the human in the loop. The focus of this effort is the use of visual display screen overlays to communicate robot perceptions and intentions to a human operator during an automated navigation task. The experimental approach is sound and based on prior studies of shared mental models and automation transparency. Results of this work support the use of such visual aids to reduce teleoperation occurrences, teleoperation times, and subjective workload.

The HRI group’s publications reflect a broad understanding of the science and research conducted elsewhere—related work is cited and contributions are placed in context. The group has also edited a book, Human-Robot Interactions in Future Military Operation,³ which includes input from external sources, including academia. It also has a demonstrated record of successfully transitioning its work to the U.S. Army Tank Automotive Research Development and Engineering Center (TARDEC) robotics programs.

Atmospheric Sciences

The U.S. Army faces many challenges to its future dominance on the battlefield, including environmental impacts on operations, which are chiefly concentrated in the atmospheric boundary layer. The research portfolio for the BED includes a range of unique atmospheric science problems of vital importance to the Army that are not addressed elsewhere in the Department of Defense (DoD) or the civilian scientific community. The division’s projects are concentrated in three critical areas: atmospheric sensing, including aerosol sampling and properties, and optical imaging and processing; atmospheric dynamics, including turbulence physics and fine-scale observations; and atmospheric modeling applications, including mesoscale and microscale models, data assimilation, and verification/validation. The BED has made impressive progress in each of these research areas.

Model of Atmospheric Boundary Layer Environment

The Atmospheric Boundary Layer Environment (ABLE) research model has been under development for about 3 years and is currently undergoing rigorous testing as a direct numerical solution (DNS) to the Navier-Stokes equations of fluid dynamics at grid resolutions fine enough to resolve the full range of turbulence down to the dissipative range, known as the Kolmogorov microscale. Once completed, the ABLE modeling system will be a primary research tool for understanding, characterizing, and predicting the evolution of the turbulent boundary layer in time and space. Innovative numerical methods, such as the Lattice Boltzmann method, are being introduced to adapt ABLE for highly complex terrain and urban conditions and to support various physics options. A special feature of ABLE is that it provides a modeling framework extensible to other BED atmospheric modeling systems, such as BED’s Weather Running Estimate–Nowcast (WRE-N) model, to maximize parallelization efficiencies and to be adaptable to a variety of novel computing platforms. In this context, it is more useful than an existing well-

_______________

³ Michael Barnes and Florian Jentsch, eds., 2010, Human-Robot Interactions in Future Military Operations, Burlington, Vt.: Ashgate Publishing Company.

tested DNS/large-eddy simulation (LES) system. Direct assimilation of lidar wind observations within ABLE is projected as a possibility for the future. This innovative work is of high quality, comparable to that in the universities and other major laboratories. The research team is well qualified and adequately supported.

Advanced Artillery Meteorology for Army and Marines

The advanced artillery meteorology system represents a mature BED project in model development and adaptation that harvests and exploits research emerging from within ARL and from the wider research community in mesoscale modeling, data assimilation, and observation sciences. Once tested and validated in this data-assimilating modeling system, improved capabilities are transitioned to operational status at the Air Force Weather Agency (AFWA) to support Army applications around the world. The project has played a vital role in improving the accuracy of both traditional artillery and rocket-assisted munitions, while decreasing costs and manpower in the field. The current project represents the ongoing evolution of Profiler, the first Army artillery meteorology system to use data from a numerical prediction model. ARL has assisted the Army and Marines in development and evaluation of the follow-on laptop version, the Computer Meteorological Data-Profiler. Ongoing work is focused on gathering and analyzing meaningful performance metrics for the currently fielded laptop system.

Aerosol Sampling

The aerosol sampling project is motivated by the need to understand and mitigate the adverse effects of combustion products and other aerosol constituents on soldier health in Army theaters of operation. The atmospheric science community has a large effort in correlating aerosol composition with health effects. Ongoing work represents the first step toward achieving a correlation between these two areas of interest while specifically focusing on Army-related sites. Although the project team did not present information beyond the identification of aerosol composition as a function of particle size and day (time), it is clear that careful thought regarding the application of the data in a larger context (e.g., source apportionment and health effects monitoring) has already occurred. This research adds a significant component to the larger atmospheric science community’s efforts to correlate aerosol composition and health effects.

Applied Anomaly Detection Tool

The anomaly detection project is an excellent example of how the understanding and characterization of the atmospheric environment can be successfully linked to the social-cognitive aspects of information science to develop practical and effective methods to improve soldier safety in hostile environments. Begun with both ARL mission and customer-driven funding, and utilizing expert insights from in-theater expertise of veteran soldiers, the project originally provided a training simulator in which soldiers learned how to recognize potential anomalies in color, texture, and other parameters of the terrain that could reveal hidden sites of roadside improvised explosive devices (IEDs) under various atmospheric conditions. More recently, the project has been extended to provide automated real-time decision support for soldiers in the field, giving them real-time warning to avoid imminent threats. The concept is now being extended to give warning of threats aloft using a range of environmental information as inputs. The work is both important and of the highest quality.

Effects of Atmospheric Conditions on Bioaerosol Viability and Optical Properties

This project focuses on understanding the fate of various bioaerosols as they are exposed to conditions in the atmosphere. A combination of laboratory and field-based experiments is used to obtain basic information to characterize the effects of gases, sunlight, and humidity on the viability of the bioaerosol constituents. The work results from a solid, collaborative effort between BED’s Atmospheric Sensing Branch, the Johns Hopkins University, Texas A&M University, and Sandia National Laboratories. The area of bioaerosols is particularly important to the broader atmospheric science community, and few studies exist to examine the heterogeneous interactions of these particles and effects on particle viability. Useful results have been obtained, including distinct changes in fluorescence level for bioaerosols exposed to different concentrations of ozone and water vapor. Variations in the fluorescence signals have been correlated with changes in the viability of the biological particle simulants.

The experimental techniques used in this work are sound. Researchers have conducted experiments using both clean laboratory air as well as actual ambient air samples. The use of actual ambient air likely resulted in additional chemistry that may not have been well represented in the laboratory experiments.

Data Assimilation Advances Supporting Nowcast Modeling

This effort focuses on developing strategies for integrating atmospheric data into existing and future ARL atmospheric models. Specific emphasis is given to assimilating Meteorological Data Collection and Reporting System (MDCRS) and Tropospheric Airborne Meteorological Data Reporting (TAMDAR) aircraft observations over the continental United States (CONUS) in WRE-N as a means of determining their practical value to short-term forecasts (Nowcasts). A key for Army applications is the ability to improve performance in data-poor regions. Though simple and easily adapted to new domains, scales, and environments, the observation nudging technique used in the WRE-N model is dated. The significant computational constraints under which the WRE-N operates preclude use of the more sophisticated approaches at this time, such as three-dimensional variational (3DVAR) analysis, increment analysis updating, and 4DVAR or variants of Kalman filtering.

Ongoing numerical experimentation is focused on determining the best radius of influence and weighting strength for assimilating the temperature and moisture data; testing indicates that the optimal choices for nudging weight and radius of influence may be case-dependent. The quality of the research is high, given the constraints mentioned earlier, and the researchers are clearly knowledgeable about the field of data assimilation.

Detection of Ultrathin Clouds

This research seeks to understand and differentiate the bulk properties of particles with sizes comparable to optical wavelengths and to quantify their impacts on the Army’s electro-optical systems. Transparent cirrus clouds fit this category, as do super-thin water clouds nearer the surface and various ensembles of aerosols. Given such diversity of causation, this is a long-term effort aimed at identifying contributors to the scattered signal in active remote sensing (e.g., lidars) or passive sensing (e.g., optical receivers). In this initial stage it is an inverse scattering problem, where a range of physically plausible solutions are matched against existing observations of known or expected particle type and distribution. The model of agglomerated spheres is unique, and it produces results that can explain the measurements without resorting to case-dependent parameters. It is computationally intense, but once done for a single particle and at various orientations and compositions, it is relatively simple to obtain the result for the bulk properties that lidar or optical instruments would sense.

The work is a fundamentally important part of the diagnosis of what is where in the atmosphere and seeks to provide an explanation of why. As for the question of why, the polarimetric signatures used in the work are convincing and capture the gist of the physical reasons. The quality of this effort is on a par with that of similar work by others dealing with, for example, classification and quantification of precipitation particles in clouds and inferred from polarimetric radar signatures. The challenge remains verification, because there are no compatible in situ sensors to use as standards. The investigators are on the right path as evidenced by the quantitative agreement of modeled polarimetric properties with the observed ones.

Human Vulnerability Forecasting

The human vulnerability forecasting project is a new focus to couple multiple sources of natural environmental data with data on human activity in order to identify potential correlations between human vulnerability occurrences (responses) and specific environmental variables (predictors). This project is in the data-gathering and initial analysis stages, and the team is currently evaluating two different computational approaches. The first approach developed a prototype forecasting software system to ingest environmental data automatically while simultaneously ingesting human activity data in the form of dates and activity type(s). The second approach makes use of the statistical learning tool Random Forest to derive a predictive model. The ability to model human behaviors based on specific environmental conditions has tremendous upside potential by providing tactical ground units with a new source of predictive intelligence for safe and effective execution of highly dangerous operations. The fact that there have been no publications or conference/workshop presentations from this project is to be expected, given the project’s newness. It is also noteworthy that the project team includes an active-duty Army senior officer, whose operational insights provide a valuable user’s perspective to the project.

Real-Time Atmospheric Imaging and Processing

The achievement of collimating beams from several lasers for the purpose of increasing power density while reducing instrument size is significant. In particular, the use of beam tails from the hexagonal sources arrangement to collimate the beams has paid off, as demonstrated in action. Considering that these are the first attempts, 70 percent efficiency is impressive, and it is realistic to expect greater than 90 percent efficiency in the very near future. The development of an agile electronically steerable laser beam would be among the first of its kind.

The use of signal processing schemes has been integrated into this capability. The lucky region fusion algorithm for mitigating the effects of turbulence along the propagation path has progressed since the last review in 2012. This technological enhancement, made possible by real-time computations on field-programmable gate arrays, increases the overall capability in optics at the ARL.

Meteorological Sensor Array at White Sands Missile Range

This development project is designed to establish, in a phased deployment approach and in response to previous ARLTAB recommendations, multiple regular arrays of instrumented meteorological towers, initially in and near the White Sands Missile Range (WSMR) complex. Especially important and unique is the addition of above-ground observing assets (e.g., triple lidars and unmanned air systems) that have the potential to provide considerable detail aloft to augment the ground-based towers. In addition, the new sensor array designs are expected to maintain some flexibility to allow adaptation as modeling needs

for verification data continue to evolve. The phased approach is sound and thoughtfully considered, and a clear context was provided for creating the network. The principal objective is to provide ground-truth data for the verification and validation of weather forecast models at the scale (1 km horizontal spacing) and with the geometry (Cartesian grid) of a native model grid, as is typically used by WRE-N. Additionally, the sensor array will provide a persistent framework for studying the atmospheric boundary layer in semiarid, complex terrain. Installation of this network has the potential to highlight ARL in the international research community in terms of observations, atmospheric boundary layer research, and numerical modeling and validation of the boundary layer.

Multiscale Turbulence in the Atmospheric Boundary Layer

The objective of this study is to learn how the anisotropy of atmospheric turbulence (that is, different turbulence along the altitude in the atmosphere) affects the transport of momentum, heat, and moisture in the boundary layer. The project focuses on the larger scales of turbulence, which are well known to be highly anisotropic but have nonsteady, nonuniform characteristics that affect fluxes through the boundary layer and near the surface. Rigorous data analysis and theoretical analyses are the preferred means of study. LES and DNS provide additional numerical means of investigating turbulence anisotropy, but anisotropic processes and evolution have remained poorly understood. As a result, anisotropic effects are only partly accounted for in most numerical models. ARL has proposed a new approach for investigating turbulence anisotropy that combines multiresolution decomposition and anisotropy of the Reynolds stress tensor analysis and tensor invariants. This research approach divides and analyzes the turbulence motion at different scales. The study allows investigation of the decay of larger and presumably more anisotropic eddies into smaller scale eddies, which are more nearly isotropic.

The work is of very good quality and has the potential to be integrated into BED’s fine-scale numerical modeling in WRE-N and ABLE, notably for better determination of the characteristics of the outer length scale in the surface layer. Recent findings have been published in Boundary-Layer Meteorology, attesting to its quality.

Real-Time Detection of Nerve Agent Stimulants Using Multiwavelength Photoacoustics

This project seeks to develop a novel form of laser photoacoustic spectroscopy to rapidly detect the presence and concentration of complex gases. Nerve agent simulants (e.g., dimethyl methyl phosphonate) were utilized as test cases. Pursuing the goal of a practical field-deployable system, researchers have modified a traditional photoacoustic system that analyzed an enclosed air sample in steady state to detect trace gaseous species. The modifications included the development of a flow-through, nonresonant photoacoustic cell for continuous sampling. Perhaps the most innovative modification is the system’s use of three separate lasers at different wavelengths, which extends the available analytic range into the mid-IR region. Currently, the system is able to detect the simulants at concentrations down to 5-15 ppm, but new adaptations and experiments are expected to lower the detection threshold by an order of magnitude. It is theoretically possible to add more lasers to broaden the analytic range, improve the accuracy of the response spectra, and further lower the threshold of detection. ARL researchers hope to be able to detect nerve agent simulants and other gaseous compounds in concentrations down to parts per billion, which would dramatically enhance soldier safety on the battlefield. The research is both innovative and of high impact.

Nowcast Model Assessment

Objective verification of numerical prediction models at the scales of interest to the Army is especially difficult. In response to prior ARLTAB recommendations, BED is now making use of the Model Evaluation Tool (MET) software developed by the National Center for Atmospheric Research (NCAR), which is widely used by the meteorological research community for verification of numerical weather predictions. In addition BED has adopted the Geographical Information System (GIS), which is well adapted for the unique and challenging model verification problems at horizontal scales finer than 1 km. Taken together, MET and GIS provide BED with expanded and flexible verification options for quantifying predictive skill and assessing product value. These acquisitions have enabled BED to rapidly advance its Nowcast model assessment capability to approach the state of the science in the operational and research communities for fine-scale atmospheric modeling.

Nowcast Modeling

The vision for the Nowcast modeling project is to give a soldier or team of soldiers the ability to enter the theater of operations with a portable computer containing a suite of numerical weather prediction models and automated tactical decision tools suitable for small-scale (tens of km) operations on the battlefield. The software system would be complete, with all the observational data and gridded model input fields the user needs to run the software on the laptop/tablet in order to generate meteorological fields and automated decision assistance products within 30 minutes.

The Nowcast modeling project is one of several observational- and modeling-based projects that constitute a comprehensive model development, adaptation, testing, and verification program. The six-person project team has also engaged with other mesoscale modeling groups around the country, including the National Oceanic and Atmospheric Administration (NOAA) Earth Systems Research Laboratory High-Resolution Rapid Refresh Model group, whose civilian users at the Federal Aviation Administration have requirements like those of the Army for small-scale, highly accurate meteorological information packaged into a useable format for the end user. The project team produced six ARL reports and two conference/workshop papers over the last 3 years. However, this work could be submitted to a suitable, high-value, peer-reviewed journal for vetting by the rest of the mesoscale modeling community.

Quantified Weather Impacts and Friendly vs. Threat Advantage

The goal of this project is to develop a quantitative, highly granular weather effects display system to enable a more intuitive, flexible, and informed interpretation of weather impacts upon operations and personnel. Such a system would replace a legacy decision assistance system, Integrated Weather Effects Decision Aid (IWEDA), which has deficiencies related to arbitrary boundaries and inflexibility to adapt to mission priorities. The proposed approach will allow meteorological parameters affecting the mission impacts to be weighted as desired by the user, with transitions between severity levels depicted as linear, quadratic, exponential, or other, as appropriate. In addition to providing a smoother impact continuum, this project introduces the important innovation of allowing a direct comparison of which side, friendly or threat, has the advantage or disadvantage vis-à-vis forecasted adverse weather conditions.

Computational Sciences

The structure of the Computational Sciences Division (CSD) includes both a substantial facilities component that serves the high-performance computing (HPC) and networking infrastructure needs of ARL, the Army, the DoD, and a growing research component focused on interdisciplinary computational science. The CSD team is making concerted efforts to move forward new Army-relevant directions around high-throughput distributed processing and the applications of modern computer science. This is a commendable approach, and CSD could consider the development of similar approaches to advance its research projects in traditional HPC and physics-based modeling. Projects were presented along five themes: programming models for distributed processing on heterogeneous architectures, tactical HPC, multiscale materials modeling around dislocation dynamics and defects, modeling mobile networks, and emerging areas.

Programming Models for Distributed Processing on Heterogeneous Architectures

This research focuses on developing programming models and associated runtime systems to allow near-real-time distributed processing on heterogeneous architectures. Establishing near-real-time situation awareness through the use of contemporary programming models and networked heterogeneous computing architectures could lead to significant and game-changing battlefield advantages. The use of portable application programming interfaces (APIs)—extremely important because HPC architectures are constantly changing—coupled with stable APIs is consistent with state-of-the art HPC thinking. ARL’s focus on tactical ad hoc HPC architectures is novel; tactical and ad hoc features are not fully accounted for in classical HPC architectures, which are well-defined, static interconnect topologies with well-behaved bandwidth characteristics. Moreover, CSD’s investigation and use of different types of heterogeneous computing architectures (e.g., general-purpose central processing units (CPUs), general-purpose computing on graphics processing units, Xeon Phi, and low-power ARM processors⁴) along with programming models and run-time systems that leverage existing and well-funded industry standards (e.g., open computing language) ensure that the research applies to existing and emerging HPC architectures and leverages the best that the industrial and academic research has to provide.

Overall, this work is likely to be an important part of ARL’s portfolio going forward, and its continued development and enhancement can be leveraged across different projects.

Tactical HPC (Cloudlets)

The objective of this research is to develop a comprehensive infrastructure for deploying and using advanced computational capabilities on the battlefield. The research has primarily focused on addressing some of the fundamental questions regarding the optimal placement of mobile computing resources and on how to identify the routes that these resources should follow in order to maximize the throughput of the computations and data transfer rates. In particular, the project on tactical cloudlet seeding considers the optimal placement of a limited number of mobile computing resources to increase the soldier’s capability at the tactical edge. The problem formulation is sound. Key strengths include their plans to explore a variety of algorithmic approaches that offer a range of trade-offs between computation costs and quality of solution. CSD has made significant progress in better defining its concept of tactical HPC

_______________

⁴ ARM processors are a family of 32-bit microprocessors developed by Advanced RISC Machines, Ltd., that power a wide variety of electronic devices, including mobile phones, tablets, and multimedia players.

(cloudlets), an area that holds the promise of providing significant operational capabilities in the field. The work has the potential of making a significant impact.

Multiscale Materials Modeling

CSD has made significant progress in work related to multiscale modeling and simulation techniques, and has applied these methods to modeling materials dislocations in microstructured crystals and multiscale computation of electron transport. The projects on computational materials science with two-scale hierarchical models and the simulation of dislocation dynamics are generally sound. In particular the problem of enhancing the scalability of scheme to model dislocation dynamics of complex microstructures appears to hold the potential for serving multiple applications, including those involving gallium nitride (GaN) materials, other electronic materials, and metallic-ceramic composites. In addition to the numerical methods and algorithms, the research has developed a software framework that employs multiple principles that have been developed in the broader computer science community but have not previously been applied for large-scale scientific problems. These include dynamic scheduling to achieve load balance (and potentially fault tolerance) and memorization to avoid repeated computation. The overall scientific work is of very high caliber; it reflects a broad understanding of the basic science and combines disparate techniques. It reflects an appropriate mix of theory, computation, and experimentation.

Modeling of Mobile Networks

CSD has made significant advances in modeling mobile networks by combining simulation and emulation with experiments. This work has resulted in the development of a parallel version of the NS-3 simulator software, and the EMANCE software has been used to emulate Army radio models that have been transitioned and used by the U.S. Army’s Communications-Electronics Research, Development and Engineering Center (CERDEC) and other DoD organizations. Related to that is the work on real-time radio frequency (RF) realistic propagation modeling and simulation that is used by CERDEC and the Naval Research Laboratory for network modeling and jamming applications. The scientific quality of the research presented is comparable to that of leading federal, industry, and university laboratories. ARL researchers evinced a clear understanding of the underlying science that supports the research activity. The quality of the research team in both depth and breadth of knowledge is strong, but the number of research personnel assigned to each project was less than that allocated to similar research projects in other federal and industrial research organizations. Overall, the project is coming to closure and is being transitioned from a research to a development project.

Emerging Areas: Software-Defined Networks and Quantum Computing and Networks

CSD has initiated projects to explore research issues in two emerging areas: cybersecurity in the context of software-defined networks (SDN) and the field of quantum computing and quantum networks.

Access to a secure network is mandatory if the future warfighter is to use data to make effective offensive and defensive battlefield decisions. CSD’s effort in this area examines automatic (cognitive intelligence) detection of various network attacks and how such detection can be used in the context of SDNs to protect the network through automatic reconfiguration. The results, when they become available, will be implemented on ARL’s current Cisco SDN routers. This is an important area of network research, and early research results are promising.

The research on quantum computing has the potential to be significant. The investigators translate a classical Boolean satisfiability (often abbreviated SAT) problem into a spin chain whose ground state can be interpreted to be an instance. The idea of using quantum algorithms to apply to SAT, one of the canonical classically exponential NP-complete (NP refers to nondeterministic polynomial time) problems, would not only give another example of a powerful quantum algorithm but also be generally useful in that SAT solvers have been applied to a wide range of concrete problems. The second step of this work to extend SAT to first-order predicate calculus will yield not only a generally useful algorithm for application but also a general approach to programming quantum computers in a natural way. This work represents some original and exciting thinking.

Network Sciences

Research presentations in network sciences were organized in the three broad areas of communications, sensing, and operational management of networked robotic systems. There was good science and engineering in many areas of the work. Since the previous review in 2012 there has been significant improvement in the quality of the research and of the presentations and in the apparent morale of the team. The lightweight intrusion detection work is promising. The issue of burstiness in intrusion detection evinced an interdisciplinary content. Strong work is being performed in tactical communications, robotics, language technologies, and E-field measures in standoff power sensors. Almost all presenters did an excellent job of articulating connections to Army operations. In some instances, however, presentations lacked clear problem definitions and suffered from poor articulation of the problem. Some weak projects have, appropriately, been eliminated since the 2012 review.

The areas in which ARL is leading the research community need to be publicized to raise the visibility of the organization. As an example, in the cybersecurity area ARL is one of the few research institutions that have access to real data. This is a major strength and would undoubtedly attract very favorable attention. The impressive early-career scientists appear to be receiving appropriate mentoring. Some work (e.g., temporal logic for intelligent systems) was in a speculative stage and such interesting work would provide good candidates for future presentations.

Multiagent Adversarial Games

This emerging project seeks to develop a simulation platform to quantify different tactical strategies. The goal is to gain insights into best strategies and to provide training technologies. The team has applied different perspectives that include behavioral game theory, dynamic behavior and strategies, simulations based on cognitive architectures, individual and small group (scalable) conditions, and formalizations. This study is led by ARL’s Human Research and Engineering Directorate.

Overall, the formulation of the problem was clear, the technical quality of the work was good, and the work is directionally new and relevant to the Army.

Data-Driven Analysis of Collaboration Structure and Evolution

The research goal is to use topological (e.g., homology) properties of social networks (i.e., network representations of relationships) from data sets to gain insights into coalition formation and decision making in groups. Data sets used include the Digital Bibliography and Library Project (DBLP) computer science (CS) bibliography, and certain group e-mails. This project has been ongoing for some years; the main contribution of the work lies in strong collapsing, an algorithmic advance for computing coverage in sensor networks. The research team is well qualified and aware of related work in the field.

Tactical Communications

A number of projects were organized around the common theme of tactical communications in generally harsh environments. The range of research in this context spanned low to very high frequency (VHF) and optical communications, as well as communications-enabled measurements that are used to control robots in new tactical environments. This exemplary body of work puts ARL at the cutting edge of tactical communications research. Among the many positive attributes of this work, these communications techniques are developed and investigated in the context of specific, interesting Army applications. The researchers were well aware of related work in the field, and they have established good external collaborations (such as with the University of Michigan in the area of small antennas) when appropriate. There is also excellent synergy within the extended ARL group, as evidenced by joint publications. The research group enjoys good external visibility, as indicated by members serving in an editorial capacity for special issues related to the subject matter (for example, the special issue on robotic communications and collaborations in complex environments, published in 2013).

Temporal Logic for Intelligent Robots

This is exemplary, early-stage work that promises to deliver considerable impact. The goal is to develop a planning system for intelligent responses by robots to new events. The work uses linear temporal logic to interface between high- and low-level planners. The decomposition and interfacing between high- and low-level planning is a novel approach. The work is synergistically linked to other ARL research on mapping for robots, and it is well integrated within a much larger project, Adaptive Collaborative Sensing, with the Naval Research Laboratory (NRL) and the Air Force Research Laboratory (AFRL), within which its niche is well defined. The work also benefits from connections to external university work, for instance, in the use of a Cornell University toolkit.

Operational Management of Networked Robot Operation

This research explores how cognitive and network information can be combined to influence decision making. This is done in the context of the robotic operation manager’s role, where the task involved a decision maker deciding whether to follow a path or not, given five cues about the environment that might include an improvised explosive device (IED). A rudimentary experiment with a 2 (time pressure/no time pressure) by 2 (bandwidth reduction/no bandwidth reduction) design was established, and 16 participants performed the experimental protocol.

Dynamics of Trust and Information Sharing

This research thrust seeks to enhance and improve the distributed decision making capability within the Army, particularly in an environment where competitive advantage is obtained through effective communications in a network of distributed and informed sources. The research examines and builds the scientific foundations for such decision making. Researchers have looked at key questions related to trust, including how trust influences the behavior of social, information, and communication networks. This work is part of the Network Science Collaborative Technology Alliance (CTA) and is performed in collaboration with researchers in the ARL Social Cognitive Network Academic Research Center (SCNARC). Researchers have developed metrics by which to gauge trust, models that describe trust propagation in a network, and trust management protocols. According to ARL, its work in this area has received significant external attention and accolades in the form of best paper citations.

The poster presentation on this topic lacked a clear articulation of the problem, the mathematical basis for the proposed solution, and comparisons with competing methods that demonstrate how the approach yielded superior results. There are some good ideas behind this research, and these need to be highlighted more effectively. How this research thrust fits in to the broader context of network science research also needs sharper definition.

OPPORTUNITIES AND CHALLENGES

Many of the presenters did not describe well how their research projects fit within the larger research context. Without such a roadmap, there is very little indication of the connectivity of the research projects, either within the subareas or across the enterprise. The ARL research effort is part of a larger collaborative effort involving external partners. A better definition of the role of the internal research in addressing the overall program goals and continued collaboration with partners are strongly encouraged. There is an expectation that the ARL strategic plan with several campaigns and crosscutting research initiatives will provide a better framework in which the research can be contextualized. In particular, it may shed light on the process by which research projects are selected. The presentations provide only the individual researchers’ motivations in problem selection; the strategic dimension is often missing.

ARL would do well to include other cross-divisional and cross-directorate projects that leverage the multidisciplinary composition of the laboratory workforce. These are vital to building a strong external presence for ARL, and such collaboration needs to be a priority for the laboratory management. A combination of top-down efforts complemented by invited ideas from researchers has been successfully implemented in other large research organizations. However, the support for such initiatives will require a broadening of the discipline base. Certain disciplines, notably the mathematical sciences and the statistical sciences, cry out for infusion of new talent. Similarly, social scientists, especially those trained in mathematical and quantitative analysis, would bring considerable value to social-cognitive sciences research and to the design and interpretation of experiments. Game theory has grown to be dominant in many areas of interest to the Army, such as understanding and shaping social relationships and behavior, and even networking. ARL could draw from the economics, mathematics, and computer science communities to strengthen its base in game theory.

In some areas of research, particularly in the computational sciences, many of the projects have a considerable software development component, which is currently being done by the researchers working on the project. This may not be the best use of their time; it is important for ARL to build capacity for software development by hiring programmers who can work in close contact with scientists on systems and algorithms.

ARL can do more to enhance its external visibility. This could begin with encouraging greater exposure for the research staff and their work to the outside world. The limitations placed on researchers’ travel to conferences is damaging to the mission of the ARL. Inability to network and meet with colleagues not only compromises the growth of research staff but also denies ARL the opportunity to brand itself in the areas of its strength. It also contributes to an insular trend in the research program. If its researchers cannot travel to conferences, greater attention could be paid to organizing onsite workshops involving a larger external research community, and possibilities of onsite short courses. ARL’s hosting of seminar series at the laboratory is a good first step in this direction.

ARL needs to leverage its scientific leadership in certain areas, such as tactical communications, machine translation, and tactical robotics, by offering incentives for academic and industrial researchers to collaborate. Such incentives could be funneled through ARL’s open campus initiative and could possibly involve the use of ARL’s laboratory facilities. ARL needs to consider designating senior members

of staff to act as official liaisons with individual university departments. Their task would be to build relationships with faculty and students, which begins by facilitating a two-way flow of information and gaining familiarity with ongoing research and students who are contemplating joining the workforce.

Autonomous Systems

It was not clear how the individual research projects in each area of the ARL autonomous systems enterprise—manipulation and mobility, perception, robotic intelligence, and human–robot interaction—fit within the larger research effort. Without such a roadmap, there is very little indication of the connectivity of the research projects either within the subareas or across the enterprise.

The internal research efforts at ARL are being performed against the backdrop of a research roadmap that includes contributions from partners and contractors. It would be useful to clarify this context in order to better understand any gaps that might exist in the research approach. Other specific opportunities to improve the overall research enterprise include the following:

• Require researchers to clearly articulate the existing technical challenges in their research and how and why the proposed tools and methods are likely to resolve those challenges.
• As ARL continues to build its research staff, give some attention to bringing in mid-career and senior personnel to mentor the outstanding early-career scientists who have been recruited.
• Look for additional ways to increase interaction of its researchers with leaders in industry and academia, given that limitations on travel have restricted this important function.
• Focus on developing a mature framework to guide the conception, design, development, and testing of small, unmanned autonomous systems, including definitions of pertinent parameters and their domain (values).
• Encourage a systems integration approach across its research enterprise to engender interconnectivity between ARL’s science and technology (S&T) campaign plans.

At a fundamental level, ARL can take additional steps to enhance the quality and impact of its research efforts. There is a trinity in research and development: analysis, computation or simulation, and testing. Analysis is essential, and there is room for improvement on this front, before one proceeds with numerical simulation or building and testing artifacts. Results from analytical modeling can guide the subsequent steps in development and identify possible missteps; this analytical component needs to be integrated into the approach to research. For example, it is not enough to build a robot and begin to generate a gait similar to that of an animal—one must understand the physics behind the energy converted in a machine when using such a method of locomotion.

Manipulation and Mobility

In the area of manipulation and mobility, there is need for a more coherent approach to vehicle design and development. It would appear that while many excellent issues are being addressed, the overall approach is somewhat ad hoc. For instance, there is an absence of nondimensional scaling in platform design. Characterization of the fundamental physics of flying vehicles—length:diameter ratio, drag polar, coefficients of lift and drag, and power requirements—must be part of the basic design philosophy. Similarly, metrics for system performance evaluation in generalized terms, such as actuation efficiency, propulsive efficiency, hover power loading, power:weight ratio for the actuator, endurance, and specific energy of the fuel source, would add focus to coupled systems and vehicles to include physics-based

performance attributes. For robot systems, the specific resistance or cost of transport for any locomotive machine, natural or man-made, is a measure of a machine’s locomotive efficiency. It would be beneficial for ARL to encourage this traditional thinking as part of the research mindset.

There is an opportunity to perform simulations of robots and vehicles based on analytical models of the physical systems operating in different environments and to include uncertainties in these models. The models can further be coupled to real control systems, leading to hardware-in-the-loop control design. The ARL may need to consider procuring development hardware, such as D-Space, for robotic controller development. Such systems would accelerate results and allow the integration of complex, nonlinear controllers, based on traditional sensors and sensor fusion, states of the machine, learned behaviors, and complex logic of state machines. Once developed, successful controllers could be programmed, at which point the developmental kinks will have been sorted out, to perform laboratory and field tests of the new controllers.

Integration of systems components is essential to the robotics research area. The system is much more than the sum of its components—nonlinear interactions, sometimes stochastic in nature, can have significant impact on overall operation. In this context, an integrated approach (systems engineering) is a fundamental (6.1 level) domain of research. Such an approach will also allow researchers to best trade concept options against the desired output or value functions. There is an opportunity for the ARL autonomous systems enterprise to assume the leadership in advancing a highly quantitative and scientific approach to systems engineering as it relates to integration of systems components into the robotics research area.

Other general approaches for consideration related to this area of effort may be summarized as follows:

• Notionally establish a family of small robots (ground and air) of varying size (between 1.0 g and 100 kg) and define and reach both the baseline and performance goals for each robotic class. These specifications could be based on a limited selection of potential Army scenarios or vignettes.
• Establish a directed robotic mobility propulsion effort to unify and direct activities required to produce very high-power and high-energy-density systems.
• Establish an integrated design and optimization methodology (considering key parameters like energy, power conversion efficiency, and locomotive efficiency) for the design of these highly integrated robotic systems. The interdependence of all the subsystems will quickly become clear to the researchers as they try to categorize future robotic systems.
• Consider the establishment of a robotic mobility systems integration laboratory. This laboratory would allow for the integration of complete physics-based air and ground mechanics models of selective robots with candidate control systems in a simulated real-world operating environment.

Perception

Although the mission statement pertaining to perception research is rather broad, the actual ongoing focus is restricted to a rather narrow set of problems. It was not clear if this focus was driven by gaps or deficiencies identified by the Army or how the work fits with contributions from partners. As an example, perception needs to support more than obstacle avoidance; it needs to support a richer semantic understanding of terrain that would be useful for people as well as autonomous vehicles. Furthermore, there is a need to address scalability issues with regard to sensor capabilities, such as varying fidelity

and power requirements with size. Sensors provide measurements of data pertinent to the operational environment. For autonomous behavior, however, it is necessary to process these physical measurements to glean information. In the area of unsupervised human detection, researchers need to explore a hybrid supervised/unsupervised algorithm that would not only be computationally efficient but could also curtail the number of false positives that the current approach seems to yield.

Sensors are also linked to communications. The processing and communications power available to a single platform determines what is transmitted—measurements, processed data, or commands. The work on parsing and action grammars could benefit from transmitting unknown constructed images over a communication network to a node with greater processing capability and reference data. In this context, focus needs to be directed at combining scene parsing, scene surface layout analysis, and 3D reconstruction to advance the state of the art in overall scene understanding.

It may be useful to place bounds on the problem dictated by mission requirements. This would help identify quantitative metrics against which progress in research tasks can be measured. Ongoing research in perception is aimed at enabling cooperative interaction between robots and humans at multiple levels. Accomplishing this within a mission context, accepted military doctrine, and social norms of the society in which the soldier-robot teams operate is a major technical challenge.

Robotic Intelligence

Overall, the research programs in intelligence reflect a broad understanding of the underlying science and research conducted elsewhere. However, researchers need to more clearly state the scientific problems they are addressing and the metrics they are using for evaluation. The work also needs to be properly placed within the current state of the art. Presenters need to better articulate the primary contributions of the research and how the presented approaches achieve those contributions. The recently hired Ph.D. researchers need to better clarify how their new work at ARL is going beyond their dissertation contributions as students.

More broadly, the challenges of robot intelligence specific to the needs of the Army need to be clarified. While the three areas related to mapping, control for communication, and cognition are important, they do not provide a perspective for the ARL vision for robotic intelligence for Army applications.

In the work on long-duration 3D mapping and navigation, there is a need for justification of why the approach performs better for this problem of long-duration mapping, including identifying metrics against which progress can be gauged. The scalability issues with the approach were correctly identified, and it will be interesting to see how the proposed strategy of forgetting parts of the map helps address this problem. In the area of robust control of mobility and communications, it is important to calibrate the performance of the approach against related methodologies in the fields of mobile ad hoc networking and optimization. There could be some important linkages of this work to the network sciences CTA, and that could present opportunities for leveraging.

The design of new cognitive architectures for control can play an important role in robotics. However, it is difficult to understand the benefit of this cognitive approach to the mapping problem. Many working solutions for mapping exist that do not make use of cognitive solutions, and it is not clear how such an approach improves mapping performance. The approach could instead be motivated by a different application domain requiring more high-level cognition, such as a scenario that entails going to the back of a building and watching a door for persons of interest.

Human–Robot Interaction

In the area of HRI interface design, the breadth of experiments is clearly commendable. The experiments, however, are conducted as separate efforts and employ different tasks. This limits the ability to draw meaningful insights from the results. For example, an Android touchscreen interface was compared to an Xbox controller, and separately a speech interface was compared to manual navigation through a GUI. An opportunity exists to place these experiments and results within a larger context and to interpret results across experiments to provide more general design guidance.

In HRI research efforts related to understanding automated system actions, the initial experiments provide encouraging results for simple task scenarios involving few factors at a time. There is a need for follow-on experiments that validate the use of visual aids when performing more complex tasks, particularly in a multitasking environment, and on dismounted soldier interfaces.

It is important to conduct more basic research that takes advantage of ARL’s unique access to soldiers. As all of the services move toward the inclusion of more robot systems, it is necessary to conduct the basic research that will allow these systems to be effective and efficient members of the team. It was heartening to note the existing collaborations with researchers in cognitive architectures and perception. This model needs to be replicated across other projects at the 6.1 level, where HRI can help guide the development of capabilities that are still very difficult to achieve without a human in the loop (e.g., perception work). Even with increased system capabilities, both in terms of intelligence and perception, there will still be a need for a soldier to interact with the robot systems. The nature of the HRI will change, from remote operation to a supervisory role and, eventually, to interaction with the robot as a team member. HRI, however, will remain the key to the effective deployment of robots in the Army and other services.

The HRI group would benefit greatly from wider exposure. ARL could consider sponsoring a workshop that would discuss HRI from the soldier’s perspective. In addition to inviting academics and people from the other service laboratories, sponsors of HRI research from other agencies such as the Office of Naval Research, National Science Foundation, Army Research Office, and Defense Advanced Research Projects Agency could bring valuable insight to this effort.

On a more general level, much of the work presented was mature; new opportunities for providing input and feedback on projects that were in their inception stage could be beneficial to the overall research enterprise.

Atmospheric Sciences

Future Army needs for environmental information and decision aids, particularly in aerosol and optics research—which is vital for understanding, collecting data within, and predicting the atmospheric boundary layer—would be significantly compromised unless adequate resources needed to build on the excellent and innovative studies currently under way are provided. The ARL team has developed a laser-based tool for capturing, holding, and analyzing an individual aerosol or dust particle. The ability to trap particles, expose individual particles to atmospheric oxidants, and monitor changes in particle morphology, chemical composition, and optical properties is highly significant and at the cutting edge of developments. This work alone has the ability to propel BED’s Atmospheric Sensing Branch to the forefront of the atmospheric chemistry community. However, ARL may not be able to exploit fully this remarkable technology because of major cuts to BED’s funding for aerosol and optics studies now planned for 2015 and 2016.

The lack of references to atmospheric science in the Laboratory’s 20-year vision document is noteworthy. There is significant interdependence among the various elements of the atmospheric science research portfolio. For example, future advances in boundary layer turbulence and predictive modeling development are linked to aerosol studies, and these ultimately have an impact on imaging and optics. Eliminating aerosol research, which is exceptionally novel and of high quality, could significantly impact the other areas. BED needs to strengthen its efforts to communicate the interdependency of the major branches of its research and their individual and collective value for improving the Army’s ability to understand and exploit environmental information to its advantage.

Most appropriately, DoD is directing significant resources in the newly emergent area of big data, with a special focus on data analytics. The atmospheric science community has been working with big data sets and extracting information for predictive purposes for many decades. It is important that the atmospheric sciences team has a seat at the table along with the computational and information sciences areas, with each sharing its history and capabilities so as to maximize synergy and rate of progress.

The transition of prediction technologies to operations is unusual in that the Air Force is responsible for providing operational weather products and services for Army operations. Matured predictive technologies developed in BED are passed to the Army Research, Development, and Engineering Centers in preparation for their implementation into the Air Force operations suite. Therefore, any evaluation of BED’s effectiveness in this arena must take into account the collaborative relationship between Army and the Air Force and its effect on the technology transition efforts. Research in this arena is further complicated by the need to come up with immediate technical solutions to pressing operational problems of Army users.

There needs to be an emphasis on enhanced communications between division leaders at the laboratory, especially with a focus on possible synergistic activities and avoiding duplication of effort. As an example, it may not be advisable to spend considerable resources studying and developing ways of mitigating human heat stress in BED while the advanced materials group in another division is working on a new self-cooling suit that would achieve the same goal.

Atmospheric Boundary Layer Environment Model Development

A number of challenging issues need to eventually be resolved for ABLE to provide useful LES Nowcasts in the complex environments necessary for actual Army applications. For example, the existing subgrid-scale (SGS) closure scheme would likely be acceptable for grids as fine as several tens of meters, but a more sophisticated SGS may be necessary for the types of complex applications envisioned. Additionally, it is certain that the cyclic lateral boundary conditions commonly used in conventional LES systems cannot be adopted for real-case predictions in highly complex terrain. It would be useful to review the literature on realistic noncyclic lateral boundary problems for possible adaptation into future work. Extension to or coupling LES with deep convection models capable of modeling moist turbulence is also a possible topic for future long-term consideration.

Advanced Artillery Meteorology for Army and Marines

The difficulty in acquiring suitable firing-range data, in addition to the necessary atmospheric data, is a concern. This type of information is not easy to extract unless the artillery trajectory and wind profile data are automatically collected and stored for later analysis. If coordinated with a carefully planned set of weapons test-firing and special upper-air meteorological data, the new meteorological sensor array under development at WSMR may help provide the type of data sets needed to perform this type of

definitive validation. Fuller potential of this work may be realized by transitioning to the newer BED models under development, specifically WRE-N and ABLE.

Aerosol Sampling

A significant challenge in the aerosol sampling work is to obtain real-time measurements that can be correlated to immediate effects on the warfighter. The project team has given thought to developing field-deployable systems to obtain real-time aerosol composition information that would be deployed for this purpose. Research also needs to be directed at obtaining a better understanding of heterogeneous (gas-particle) interactions to help quantify uncertainty in the reported chemical concentrations.

Effects of Atmospheric Conditions on Bioaerosol Viability and Optical Properties

Additional work is needed to examine the mechanisms of interactions with ozone (O₃) and with hydroxy (OH) radicals. This may involve laboratory studies where different mixtures of chemicals can be controlled. The project team has already developed a plan to conduct field-based experiments that would allow them to effectively assess the mechanisms of inactivity as a function of atmospheric chemical properties.

Data Assimilation Advances Supporting Nowcast Modeling

Further testing over a range of weather conditions and over various types of terrain is needed to determine whether a single optimal set of nudging weights and radii of influence is indeed acceptable. Data starvation experiments would be revealing, especially if conducted over a data-rich environment, such as the CONUS, where the rest of the data can be used for thorough evaluation of impacts.

It would be useful for the research team to perform experiments identical to those now being conducted with observation nudging but instead using one or more of the sophisticated data assimilation techniques now available. This is important for showing how differences in quality relate to computational complexity/domain size and location/run time; demonstrating the required trajectory of computing to meet future needs; and identifying ways of improving upon observation nudging. Additionally, as part of its data assimilation program, BED needs to consider undertaking the design of observing system experiments and observing system simulation experiments that would help answer such questions. In addition to the emphasis on model assessment via statistical approaches for measuring skill, there is a need to assess the value of forecasts. Fine-scale forecasting requires confrontation of the value/skill paradox, and ARL needs to collaborate with other groups that are dealing with this same challenge, such as the use of object-oriented verifications. The NCAR MET verification software, already imported and used by BED, provides a ready path for such collaborations.

Real-Time Atmospheric Imaging and Processing

It might be possible to adapt the technology to quantify turbulence along the path. The cumulative propagation effects (i.e., the distribution of power projected perpendicularly to the propagation) are related to the turbulence along the path. It is necessary to explore the feasibility of retrieving turbulence properties from such projections. Other methods for de-blurring images have been developed by the image processing community, and it would be helpful to present a comparison of the lucky fusion algorithm with these other techniques.

Development of a Meteorological Sensor Array at White Sands Missile Range

The addition of one or more upwelling infrared (IR) sensors would be quite valuable for capturing the diurnal structure of the atmospheric boundary layer and would help to increase the value of the network for broader ARL applications. Additionally, the network would be ideally suited for the use of mobile Doppler radars as an adjunct to the planned lidars and other above-ground sensors, especially during intensive observing periods. Maintaining the equipment and recording, archiving, and quality-checking the data streaming from an observing network of this magnitude will be a daunting task, despite BED’s long history of atmospheric monitoring at WSMR. ARL could beneficially investigate direct collaborations with operators of large existing mesonets, especially with regard to data quality control, management, and availability of replacement components needed to operate and sustain an extensive network on this scale. It may be useful to consider embedding future installations of the network within other existing networks to leverage their infrastructure and share costs. An enormous range of scientific and technical expertise as well as monetary resources will be required to operate and maintain the network.

Multiscale Turbulence in the Atmospheric Boundary Layer

More attention needs to be given to identify the specific weaknesses in existing turbulence parameterizations within BED Nowcast models, whether it be those used for all turbulence scales, as in a mesoscale model, or those used for the subgrid-scale closure in LES.

Real-Time Detection of Nerve Agent Stimulants Using Multiwavelength Photoacoustics

Additional investigations to examine the effects of atmospheric chemistry, in particular the effects of O₃ and NO_x on the compounds of interest, are under consideration. Additional work is needed that specifically focuses on the potential interferences of OH radical chemistry with the sensing of the chemical agents of interest.

Nowcast Model Assessment

The problem of meaningful model evaluation at the scale of 1 km or less is particularly challenging, and BED is encouraged to assess existing model verification techniques for any gaps or weaknesses for its Nowcast model applications. There are relatively few examples of prediction verification for turbulence, boundary layer depth, and the range of variability in the boundary layer for wind, temperature, and moisture. Developing a comprehensive approach to identify previously ignored or underassessed boundary layer quantities of interest for Army operations, and developing appropriate verification techniques that leverage and extend the existing software would be valuable contributions.

Quantified Weather Impacts and Friendly vs. Threat Advantage

The project represents a welcome upgrade, but the team faces significant challenging questions: What is the best way to depict transitions between impact classes? How difficult will it be to incorporate the analogous threat-side thresholds? Are the measures of impact the same or different for friendly and threat forces, and on what basis? As noted with the advanced artillery meteorology project, the appropriate type of system verification and validation data will not be available unless rigorously defined field testing or an automated data collection scheme can be developed and executed by the team. There may be some

benefit for team members to engage with researchers in the Next Generation Air Transportation System (NextGen) program who are developing automated decision assistance tools for air traffic managers to assess the adverse impacts of weather.

Computational Sciences

The overall research portfolio represents a good mix of projects covering ongoing high-impact areas of computational science research as well as emerging areas in which both the traditional HPC infrastructure (large center-based HPC resources) and nontraditional high-performance computing platforms (mobile/distributed heterogeneous computational nodes at the battlefield) can be leveraged to provide to the Army unique and novel capabilities.

A crosscutting area of opportunity exists around interoperable high-throughput run-time systems. Such run-time systems already play an integral role in the research related to programming models, distributed heterogeneous architectures, tactical HPC, and characterization of materials properties. This area needs to be considered a core research area so as to bring about a significant enhancement of capabilities and development of enabling technology for various computational and big data problems. There is a need to define the basic research problems, articulate strategies for sound software development, determine milestones and measures of success, and develop plans for field testing and pathways for deployments. There needs to be a concomitant focus on building close collaborations with outside entities working in this area to ensure that their work informs and is being informed by related research efforts.

The project on tactical HPC (cloudlets) is designed to provide capabilities and address issues that the commercial and academic research is just starting to consider. However, the number of research personnel assigned to work on this project is considerably smaller than what is required to achieve its objectives and make meaningful research and operational (via realistic proof-of-concept implementations) contributions. In addition, the research in this project could leverage some of the ideas that are being developed in the context of off-loading some computing power-intensive operations from cell phones to the cloud. Though this academic and commercial research is rather simple and deals with nonadversarial settings, it can still inform some of the design and algorithmic decisions. ARL research will benefit from getting data on soldier usage models through early-stage testing while development options remain flexible. It is also important to consider uncertainty in input parameters. Consequently, in addition to traditional linear programming, CSD needs to also explore techniques for optimization under uncertainty, including chance-constrained optimization and recourse optimization.

In the multiscale materials modeling area, it will be important to analyze fundamental limits to scalability and opportunities around data assimilation before investing in significant implementation efforts. Additionally, having professional software developers involved in the project will likely lead to a product that can be utilized more broadly in Army applications.

To provide meaningful cybersecurity capabilities, the research needs to quickly transition to detecting more sophisticated attacks such as distributed denial of service, Internet protocol (IP) spoofing, and misaligned packets.

CSD has initiated promising new projects related to software-defined networks, quantum computing, and quantum networks. These areas represent cutting-edge research. CSD is applying its computational expertise on addressing computational and methodological bottlenecks associated with analyzing very large data sets. The general area of data-intensive computing (big data) represents an application area in which CSD needs to focus in order to allow Army warfighters to obtain complete situation awareness and to respond in real time.

A key challenge is the understaffing in the CSD. More than 50 percent of the team lead positions are vacant, and many of the researchers are early-career. The branch chiefs are working as team leads,

but they shoulder many disparate responsibilities in the various projects. As a result, many projects lack a sufficient number of senior, experienced researchers to lead, guide, and mentor the young researchers.

Network Sciences

In general, the quality of the presentations was quite good. It was clear that the presenters were working from a template to provide context about their projects. In some cases, there was no clear problem definition. A more effective standard format would summarize what problem is being solved, why it could be important, what is the state of the art, and what is being addressed in a given project. For ongoing projects it would be helpful to know what progress has been made since the previous review.

Some of the work presented was insular, however, and could benefit from more connections to outside work, including that being done in academia, industry, and other government laboratories. This was particularly true in the areas of social media and social networks, but also in other fields as well. As an example, in the area of cybersecurity, there is a plan to build a supervisory control and data acquisition (SCADA) test range, which may duplicate outside efforts (e.g., the SCADA test-bed at the Idaho National Laboratory). Such insularity lowers the effectiveness of the Army’s investment in this research.

Greater interaction within ARL would be beneficial in some cases as well. As an example, research in the area of social media and social networks could benefit from interactions with the cybersecurity area. Similarly, even a very strong area of research like the E-field and RF measurement effort could benefit from greater interaction with the signal processing community in ARL. More generally, there is a good bit of effort on sensor information exploitation and fusion that could benefit from an integrated approach.

There is a need to diversify the talent pool in ARL to bring a broader range of ideas to some problems by adding strengths in areas such as the social sciences and mathematical sciences. There were instances in which broad interactions are already occurring. One such area is language technologies, which is very well connected to the larger community. Similarly, the work on temporal logic for intelligent systems has strong connections to NRL and AFRL.

The work on multiagent adversarial games is in the early stages, and so there are shortcomings. For instance, the context is not very dynamic, and the use of real-time feedback, which must be critical in tactical situations, is not overtly modeled. There is a wealth of relevant literature that could be consulted related to dynamic environments, adversarial intent, the impact of context, and individual versus small group differences. The simulations and modeling conducted had to rely on limited data and much speculation. Consequently, this research needs to have the best hypotheses tested with empirical data to see if it will have any real value.

In the project on data-driven analysis of collaboration structure and evolution, the results and methods are interesting, but it is not obvious whether the project is achieving its stated goal of understanding networks in unfriendly environments and detecting inefficiencies in tactical communications. The project could more directly benefit from the use of data sets that better align with these objectives.

In the project on operational management of networked robot operation, it was not clear from the presentation whether this was a between-subjects or within-subjects design. Nevertheless, the number of participants was too low to draw any meaningful conclusions. Other shortcomings of the study included only 5 cues out of 16 available (degrading the task condition), and fitting too many parameters for the ACT-R model.⁵ Because ACT-R had to fit too many parameters, perhaps a different approach to human cognitive modeling could be pursued. If more participants were included, then signal detection theory

_______________

⁵ ACT-R (short for “Adaptive Control of Thought—Rational”) is a cognitive architecture: a theory for simulating and understanding human cognition. Researchers working on ACT-R strive to understand how people organize knowledge and produce intelligent behavior.

could be applied. Similar to the current approach of the robot operator giving advice, research related to a judge-advisor system might be particularly powerful.

In the broad areas of social media and social networks, industry (e.g., Google and Facebook) is investing significant resources on research. ARL will be better served by focusing on niche areas that are important to the Army and needs to yield on the more general problems that are being heavily researched elsewhere. There is some concern at ARL’s Computational and Information Sciences Directorate (CISD) that leadership and responsibility for the area of social networking have been delegated to the Network Sciences CTA, and that ARL has not developed a pool of in-house talent to perform research that complements and leverages the work being done by the CTA. Besides needing additional social networking experts with strong analytical skills, ARL security protocols appear to limit the ability of the researchers to access the appropriate software needed for this work.

Research related to trust management and quality of information needs a better articulation of the conceptual framework and a clearer statement of the problem that is being solved. This is a promising area of research for ARL and could benefit from inclusion of social networking and social media experts with strong analytical skills. The work as presented did not effectively establish a case for why a trust management framework was necessary; the impact was lost in the jargon of trust management and value of information. Better use of Army-pertinent examples can forcefully establish the importance of the work and help to identify challenges that need to be overcome in ongoing research.

A related crosscutting thrust could be automated extraction and analysis of implicit canonical features in data. This could be useful to ground the value of information research. The image analysis group’s approach to value of information is automated extraction of features followed by statistical analysis to identify and quantify anomalies. There is also ongoing work on use of automatically extracted features in low-resource language research.

Another potentially productive area of related research is to add a social network layer to the cybersecurity tool set that models organizational and command structure and to exploit natural language analysis capabilities. ARL would benefit by including research on a mathematical foundation for system cybersecurity in its science of cybersecurity research.

The project on a fuzzy logic approach for value of information focuses on how intelligence analysts use information in making recommendations to operational commanders. The particular emphasis of the work was on examining the way information was valued by the intelligence analysts. A fuzzy logic approach was used to model the value of information. While this approach can be effective by appropriately tuned quality function deployment (QFD), there is a need to compare the effectiveness against that of competing methods. In particular, switching from fuzzy logic to a Bayesian or other probabilistic or information integration modeling system would be useful. A complementary effort at ARL using Bayesian statistics for information integration was also presented, and this represents a natural pathway for possible collaboration.

The focus on cloud computing, which is primarily covered in the CSD, is an important thrust. Given the tremendous size of the problem and the embedded opportunities, however, an enlarged and integrated effort with network sciences is needed. In the external networking community, the design and control of data centers has been recognized as a research area of some significance; ARL needs to realize the scale of the challenges and opportunities that this offers.

There is considerable interest in research initiatives that integrate (1) materials science for new sources of renewable energy (e.g., solar cells) and (2) energy storage devices with energy management in networking (such as ad hoc routing of packets via nodes with adequate energy reserves) to provide assurance that tasks are completed. ARL is well positioned to conceive a broad, multidisciplinary, cross-

organizational initiative in this area. This research would also have relevance to the issue of energy efficiency in data centers, which is a big problem in itself.

OVERALL TECHNICAL QUALITY OF THE WORK

In the area of autonomous systems, the piezoMEMS research and associated small robotics effort is first rate, with elements that are at the vanguard of this field. Specifically, the work in motion generation at the MEMS scale is seminal. Large-amplitude motions are being created at the micron scale using integrated actuators, structures, and electronics cofabricated on silicon.

In the area of atmospheric sciences, the development of a laser-based tool for capturing, holding, and analyzing an individual aerosol or dust particle adds a unique ability to monitor changes in particle morphology, chemical composition, and optical properties under changing atmospheric conditions. This work positions the ARL at the forefront of the atmospheric chemistry community.

In the area of computational sciences, the work related to multiscale modeling of materials, simulation/emulation of mobile networks, portable programing models and run-time systems for heterogeneous architectures, and tactical high-performance computations represents the state of the art.

In the area of network sciences, the work related to trust and quality of information has received broad recognition in the technical community. Similarly, the work on cybersecurity is of the highest quality and benefits from ARL access to real data.

Autonomous Systems

Many of the ARL internal research projects in the autonomous systems enterprise are of very high quality and have benefited from engagement with other research institutions, including partners in the CTAs. For each of the key areas—perception, intelligence and planning, human–robot interaction, and manipulation and mobility—the overall technical quality of the work is high and is being recognized as such by virtue of publication in archival journals and proceedings of recognized conferences and symposia. Also, the Research@ARL monograph series on autonomous systems⁶ is commendable. For most of the work reviewed, the scientific quality is comparable to that being conducted at other federal research laboratories and at universities here and abroad. The research staff is very well qualified to undertake the research projects and is broadly aware of the state of the art in the field and ongoing research at other institutions. A number of research scientists have been newly recruited, and they promise to contribute to an exciting future for the ARL. Mentorship for these early-career scientists will be of paramount importance for their long-term success at ARL. The laboratory facilities and the infrastructure are state-of-the-art and supportive of the ongoing research activities.

Atmospheric Sciences, Computational Sciences, and Network Sciences

In general the quality of the work is high, and the research is having an impact. The trend in the overall quality of research is positive. There has been a dramatic improvement in quality since the 2010 review, although there are still pockets of research that need to be brought up to expected standards.

The work is on a par with that at other government laboratories. Specifically, research in areas of multiscale modeling of materials, simulation/emulation of mobile networks, portable programming

_______________

⁶ Army Research Laboratory, 2013, Research@ARL: Autonomous Systems, July, http://www.arl.army.mil/www/pages/172/docs/Research@ARL_Autonomous_Systems_July_2013.pdf.

models and run-time systems for heterogeneous architectures, and tactical HPC are state-of-the-art. This represents a significant and continued improvement and is a clear indication that CSD’s leadership and its research staff are making CSD a leading organization in computational science.

Seminal work is being conducted in deploying laser optics for the study of aerosols. Capabilities developed and demonstrated in this work were at the cutting edge of research in this domain. Work related to multiscale modeling of materials, simulation/emulation of mobile networks, portable programing models and run-time systems for heterogeneous architectures, and tactical HPCs represents the state of the art in the field. In the area of network sciences, the work related to lightweight intrusion detection in the context of cybersecurity is important. This work is novel and has potential for future development. The work on burstiness in intrusion detection represented a novel combination of statistics and computer science, and it has a strong basis in science and a good potential payoff. The research related to tactical communications, the work in robotics, the work in language technologies, and research on E-field measures in the standoff power sensor project was also strong.

The research being conducted in several projects by the Atmospheric Sensing Branch is of an exceptionally high quality, addresses cutting-edge problems, and can be applied to meet both internal Army needs as well as the needs of the larger atmospheric sciences community. Of particular note is the quality of the ABLE modeling research that is contributing to the fundamental understanding of turbulence anisotropy. Similar quality of work was evident in the project related to data acquisition at the Nowcast model scale, and data assimilation in the Nowcast model. The work related to data assimilation capabilities for very fine scale domains promises to be at the forefront of science in this field.

In some areas, the work leads that of academic institutions. As an example, in the area of network defense, ARL has a natural advantage in having access to a facility that provides actual data. Similarly, ARL research generates papers, patents, and other products that, in an academic setting, would be considered essential to establishing the impact of the work. Broader comparison against work in academic institutions is more difficult, because the ARL work and its objectives differ somewhat in nature and style, with academia focusing on graduate education. Overall, both the number and quality of journal publications seem to be increasing over previous years. Also, the Research@ARL monograph series on network sciences,⁷ and imaging and image processing⁸ are commendable. Conference presentations have declined for reasons discussed above. Although the work is of high quality, this is a fragile situation. Careful cultivation of the research culture and appropriate mentorship of the research staff are essential to place this progress on a firm foundation. Some of the presentations and research thrusts could be improved by better articulating and defining the methods that will be used to assess their results and the approaches against which they will be compared.

_______________

⁷ Army Research Laboratory, 2013, Research@ARL: Network Sciences, March, http://www.arl.army.mil/www/pages/172/docs/ResearchARL_March2013.pdf.

⁸ Army Research Laboratory, 2014, Research@ARL: Imaging & Image Processing, January, http://www.arl.army.mil/www/pages/172/docs/Research@ARL_4_Imaging_Jan2014.pdf.