Sensing, Effecting & Battlefield Environment

Research concentrates on understanding and exploiting information gained through sensing and exploiting data to drive effectors. Both sensing and effecting necessitate detailed understanding of corresponding physical behaviors that generate and utilize data, as well as effective means for storage, retrieval, and manipulation of data.  Additionally, knowledge of the physical environment is necessary to understand the impacts on mission planning and decision-making including quantifying uncertainty and reducing the element of surprise.

Atmospheric Characterization, Analysis and Effects (WSMR and ALC)

This research seeks to advance the science of atmospheric boundary layer (ABL) physics, processes, and associated impacts; in highly-complex terrain and urban megacity environments. Theoretical, laboratory, and field studies are conducted to investigate near-surface ABL processes that are most impactful to Army operations.  The characterization research is greatly augmented by the Meteorological Sensor Array (MSA) that is currently being deployed adjacent to and within the boundaries of White Sands Missile Range, NM (WSMR). When fully complete, the MSA will feature a continuous 40 km x 15 km domain that encompasses a valley (the USDA’s Jornada Experimental Range) at an elevation of 1300 m MSL, and the San Andres Mountains which peak at an elevation of 2500 m MSL.  This unique facility is comprised of meteorological sensors, at unprecedented spatial and temporal scales, both for in situ and remotely-sensed data collection.  Our ABL research directly leads to advancements in the understanding of atmospheric and climatic effects on Army operations, infrastructure, weapon systems, and renewable energy production systems.    Additionally, tactical decision aids being developed characterize and depict atmospheric and climatic impacts on mission systems, operations, and personnel; thereby providing battlefield commanders with readily-actionable intelligence information. Development environments include Soldier/system-level crowdsourcing of meteorological data from mobile platforms, and technologies for mitigation of adverse environmental conditions. Additionally, analysis is conducted to identify statistical relationships between atmospheric conditions and battlefield events, and the likelihood of an event given the probability of the environmental conditions occurring.

Principal Investigator:

Dr. Robb Randall,, DSN 258-3123

Atmospheric Sensing (ALC)

This research aims to advance the state-of-the-science and understanding of atmospheric aerosols, remote sensing of atmospheric parameters, and theory and modeling of the propagation of electro-optic and acoustic signals through the atmosphere. The development of new and novel sensing techniques is a common thread of effort for each of these research opportunities. Aerosols play a number of distinct roles in the atmosphere, such as they:  are key in affecting radiative transfer and energy balance; are potential confounders of chemical and biological detectors deployed to protect the Soldier; act as condensation nuclei for clouds and precipitation; significantly contribute to the uncertainty in climate modeling. ARL has developed and employs a unique capability to optically-trap single aerosol particles out of a continuous sampling stream with relatively high efficiency. Once an aerosol particle is trapped it can be optically characterized to determine chemical composition using multiple forms of spectroscopy such as Raman, cavity ring-down, and fluorescence spectroscopy to name a few. Additionally we employ other instruments such as the Battelle Labs Resource Effective Bio-Identification System (REBS) to collect and characterize background atmospheric aerosols. Profiling atmospheric parameters such as wind speed, wind direction, air temperature, and aerosol loading through both ground- and airborne-based remote sensing platforms is key to characterizing important parameters used in prognostic modeling of the atmosphere that span the scales of Army operations. Instruments such as ultra-compact Doppler LiDAR Systems, Stimulated Raman Scattering LiDARs, and ultra-compact Doppler radars are key tools in the mission to investigate atmospheric boundary layer physics and processes. Partnering with private industry and academia, ARL is working to reduce the SWaP and cost of these instruments as we look to future operational needs. The development of novel retrieval algorithms and methods is of significant interest. The atmosphere also significantly impacts the propagation of both electro-optic and acoustic signals. With respect to acoustics, ARL’s research efforts cover a significant portion of the acoustic spectrum. ARL develops acoustic propagation models to accurately capture the effect of the atmospheric turbulence, wind speed and direction profiles, and thermal profile on the characteristics of an acoustic signal. In particular we develop means to optimize the collection and analysis of infrasonic signals. ARL is also interested in collaborations on open-path acoustic tomography, leveraging the MSA located at WSMR.

Principal Investigators:

Mr. Chatt Williamson,, (301) 394-1771

Atmospheric Modeling (WSMR and ALC)

The overall goal of this research thrust is to understand and forecast the time-dependent, three-dimensional atmospheric state to determine its impacts on Army capabilities and operations. Through physical modeling, data driven approaches, or a combination of both, we seek to represent atmospheric features with spatial and temporal scales on the order of tens of meters and a few minutes within challenging environments such as complex, urban, and forested terrain. Represented physics span the gamut from full-physics numerical weather prediction (NWP) to low-computational cost, kinematic models. The atmospheric behavior at these scales exhibits significant variability and uncertainty, due to incomplete initial or boundary conditions and numerical errors; this uncertainty must be efficiently quantified and appropriately expressed in order to accelerate decision making. The level of detail required of the models will be application-specific and is constrained by limited computational resources (especially at tactical echelons), intermittent network communications, and sparse initialization data. Methods to mitigate these factors are critical, and we are developing new verification and validation techniques to better quantify model accuracy in these high-variability settings.

Principal Investigator:

Dr. Ben MacCall,, 301-394-1463

Acoustics Sensing & Processing (ALC)

Acoustic/Infrasonic Sensor Concepts
Research will be focused on the development of unique, high dynamic range, low noise sensors for acoustics and infrasound. Sensors that measure acoustic pressures and those that resolve the acoustic particle velocity will be included. These devices must be robust and be able to survive for extended use in an outdoor environment. Acoustic anechoic chamber, speakers, array configurations, instrumentation microphones and data acquisition are resources that can be applied to this research area.

Direction of Arrival Estimation
This research will be on developing direction of arrival (DOA) estimation algorithms for non-stationary signals, noise and interference. The focus will be on processing acoustic signals collected using either traditional microphone arrays or novel vector sensing concepts.

Robust Multi-target Tracking
This effort will focus on developing multi-target detection and tracking algorithms in the presence of non-stationary interference and noise. The focus will be on processing acoustic signals generated from aerial targets in a noisy environment using one or more arrays of microphones.

Classification of Acoustic Signals
This effort will focus on developing algorithms to classify targets of military significance from acoustic signals. The research will include preprocessing the data, extracting robust features, and classification. Algorithms that offer efficient processing requirements are most favored.

Acoustic Array Signal Processing
Work here will focus on signal processing techniques to handle multi-sensor microphone arrays for infrasonic as well as acoustic applications. Array signal processing for wind noise reduction, time difference of arrival and beam forming techniques are of interest. Exploiting the benefits derived from microphone arrays including the determination of optimum numbers of microphones within an array is of interest.

Wind Noise Reduction
This effort is focused on the development of novel acoustic and infrasonic wind noise reduction technologies, e.g. investigating various porous materials and windscreen sizes both theoretically and experimentally. There is a need to minimize the size of both acoustic and infrasonic windscreens while maximizing their sound transmission capabilities and their environmental robustness. Further, this research will also address signal-processing methods to extract infrasonic signals from data corrupted by wind noise.

Ground Effect Mitigation
This research will focus on directly estimating and mitigating the adverse effects of interfering ground-reflected acoustic waves measured by three-dimensional microphone arrays. The thrust of the research is to improve estimated source elevation angles by removing the phase- and amplitude-altering effects of the porous ground.

Principal Investigator:

Kirk Alberts, , (301) 394-2121

Radar Technology (ALC)

Conduct research and development of all aspects of radar technology. Specific research areas of interest include:  high-fidelity radar signature modeling, advanced radar signal processing including synthetic aperture radar (SAR) processing, advanced state-of-the-art radar architectures for adaptable/agile waveforms, multi-channel radars, and cognitive radar techniques and approaches for the congested RF environment.

Principal Investigator:

Anders Sullivan,, (301) 394-0838

Radar Technology for Degraded Visual Environment Mitigation (ALC)

Interest is in small, affordable millimeter-wave (mmW) radar concepts, electronic beam forming, mmW circuit design and fabrication, novel antenna concepts, integrated microprocessor design and fabrication. 

Principal Investigator:

David Wikner,, (301) 394-0865

Supporting Facilities:
mmW instrumentation laboratory and planar antenna circuit fabrication lab.

Equipment Available:
RF test equipment (signal generators, power measurement equipment, sources & amplifiers), RF circuit fabrication equipment (milling machine for RF circuit board fabrication, circuit board assembly equipment), CAD design and electromagnetic simulation software (AUTOCAD, HFSS), and data analysis software (MATLAB).

Technology for Multi-mission Radar (ALC)

ARL is developing a technology base that could be leveraged for the development of future radars that are more adaptive, scalable, and low-cost.

Principal Investigators:

Abigail Hedden,, (301) 394-0877
Tony Martone,, (301) 394-2531

Supporting Facilities:
Anechoic chamber, circuit fabrication lab, and RF devices and circuits characterization lab.

Equipment Available:
RF test equipment (signal generators, power measurement equipment, sources & amplifiers), RF circuit fabrication equipment (milling machine for RF circuit board fabrication, circuit board assembly equipment), Electromagnetic simulation software (HFSS, Sonnet, Axiem, Momentum), CAD software (AWR, Keysignt, Cadence), and Device measurement equipment (RF probe station with vector network analyzer)

Image, Video Processing & Analysis (ALC)

Sensor Fusion and Biometrics
Develop new mathematical tools for multimodal fusion of homogenous/heterogeneous sensor data for display as well as identification. Provide algorithms to identify individuals from their biometrics data.

Video and Imaging Understanding
This effort will focus on developing machine learning algorithms that analyze and label images and videos for detection and identification of faces and people’s activities. Research in computer vision, machine learning, and statistical techniques for developing new methods for object classification and identification. Emphasis on developing new methods to analyze full motion video for tracking individuals or a group of people. Machine learning techniques will be investigated to recognize people’s activities in an unconstrained environment. The research includes semantic parsing, image understanding, activity recognition, big-data analytics, and automated forms of (deductive, inductive, or analogical) reasoning on this data.

Anomaly Determination
This effort will focus on research and development of a robust and flexible framework, context-aware feature extraction and anomaly determination algorithms in order to determine anomalous activities and analyze patterns of life. Focus will be on exploiting and processing disparate structured and unstructured data including sensor data, video data and social media data.

Supporting Facility:
Anomaly Determination Testbed (ALC)
Networked testbed with software and analysis tools to collect and process multi-modal sensor data and social media data for anomaly determination algorithm development.

Applied Anomaly Detection
Visualization tools are being developed to aid the Warfighter in identifying potential areas and objects of interest as well as other battlefield hazards. The project is being transitioned from a training tool environment to a tactically-available decision support application. Part of that transition is the research and development of anomaly detection algorithms to detect and possibly identify anomalous regions, hazard indicators, and potential hazards. In addition, the research will explore information saliency in various environments.

Anomaly Detection Algorithms
Research and develop multi-modal sensor and fusion algorithms for anomaly detection. Bayesian and non-Bayesian methods will be explored.

Spectral-Spatiotemporal Sensor Data Analysis for Target Detection and Identification
This effort will focus on the research and development of advanced autonomous decision methods for target detection and identification, using multivariate sensor data, longitudinal data analysis techniques, and sensor phenomenology. Focus will be on exploiting and processing correlated and uncorrelated spectral-spatiotemporal imageries of a particular site continuously recorded over a period of an entire year in the longwave infrared (LWIR) region of the spectrum.

Algorithms for Daytime and Nighttime Target Detection and Identification
Focus are threefold: (1) conduct a rigorous spectral-spatiotemporal, longitudinal data analysis of a LWIR hyperspectral dataset, aimed at statistically quantifying the evolution of spectral patterns of multiple materials in a target site as a function of short-term and long-term time periods, and through various weather conditions naturally occurring in four meteorological seasons; (2) determine sources of performance degradation using state of the art algorithms to process the dataset; and (3) leverage results from (1) and (2) to introduce new algorithmic concepts for daytime/nighttime target detection and identification. Parametric, nonparametric, and semiparametric methods will be explored.

Phenomenology of LWIR Hyperspectral Signatures
Focus on working with LWIR hyperspectral imageries collected on approximately an hourly basis of a fixed area over a long period of time (currently a year). Effort will specifically focus on identifying, characterizing, and understanding the phenomenology of the spectral signatures of the materials in the scene (both man-made and natural) and how they are affected over time and various atmospheric and weather conditions.

Text and Video Analytics
Research seeks to develop means of enabling automated analysis of text, image, and video data sources, and developing techniques that allow natural language queries regarding image and video data. Approaches go beyond mere object recognition to the ability to summarize, interpret, and answer questions about images. It requires integrating information about the images with prior world knowledge and associated language in a robust and effective way. The research includes semantic parsing, image understanding, activity recognition, big-data analytics, and automated forms of (deductive, inductive, or analogical) reasoning on this data.

Principal Investigator:

Raghu Rao,, (301) 394-0860

Sensing, Signal Processing & Fusion (ALC)

Distributed Signal Processing over Sensor Networks for Detection and Estimation
Advances in micro-electromechanical systems technology have led to the development of low-power, low-cost, intelligent sensors equipped with multiple functionalities and can be networked through wireless links. These wireless sensor networks have revolutionized the sensing technology for a broad spectrum of applications including monitoring and surveillance, disaster management, estimation and tracking. However, most advancements in the field are restricted to industrial and civilian applications where the environmental tempo and resource constraints are much relaxed compared to the typical military application. Also, quick deployment requirements and lack of troubleshooting capability demands low-power, durable, fault-tolerant, resilient, and robust sensor networks for most military applications. Development of such wireless sensor networks require low-energy, scalable, distributed signal processing and data aggregation techniques to deal with the large number of measurements, and require efficient protocols to deal with communication limitations. The goal of this research is to develop state of the art and emerging distributed signal processing and machine learning methods that deal with some of the above-mentioned design challenges that arise in distributed learning, estimation, detection, synchronization, and localization in wireless sensor networks. More specifically, we focus on developing robust algorithms that are fault tolerant and resilient to external attacks and would facilitate distributed learning and signal processing for military relevant scenarios.  Furthermore, we aim to develop accelerated decentralized algorithms for clustering and learning low-dimensional representations of target objects for detection and classification at the tactical edge.

Multi-modal Sensor Data and Information Processing
Research and develop semantically-aware machine learning and signal processing algorithms to discover, extract, correlate, combine and/or fuse text, photo, sound, image and video data/information to facilitate automatic and intelligent understanding of the context (location, sensors, network topology, etc.) in which data/information was generated and stored, so that the machine can learn to draw logical conclusions and establish relationships from sets of heterogeneous sensors.

Exploitation of compressive sensing techniques to learn new approaches for signal representation and analysis will be considered. Distributed processing of sensor data under a variety of conditions, such as sensors either co-located or not co-located, to produce robust detections and classifications. In addition, feature extraction of other types of information sources such as electronic signals and social media will also be investigated for enhanced situational awareness. New tracking algorithms using distributed sensors will also be explored.

Research and develop robust and efficient multi-modal fusion algorithms for SWaPT (size, weight, power, time) constrained unattended ground sensors.  Resiliency to communication failures, data corruption, missing sensors and/or modalities will be considered.

Uncertain Probabilistic Reasoning over Conflicting Information
This research effort will focus on uncertain probabilistic characterizations and reasoning frameworks to enable uncertainty-aware artificial intelligence and machine learning (AI&ML) systems. . While AI&ML systems enable remarkable prediction capabilities, their performance is known to degrade drastically under austere training environments. Furthermore, data for both training and testing can come from sources of unknown reliability.  It is crucial to retain system robustness despite sparse and conflicting data. Research to determine the reliability of sources and their corresponding information in face of these conflicts is needed, which in turn, enables innovative methods to fuse information from both honest and possibly malicious sources. Furthermore, research is needed to understand how training data sparsity leads to uncertainty in low-level reasoning ML methods such as neural networks and in higher-level reasoning AI methods.  Current research is investigating subjective logic as an uncertain probabilistic reasoning framework to enable reasoning over multiple propositions from multiple sources using various reasoning constructs, e.g., evidential neural networks, subjective Bayesian networks, probabilistic logics, etc.

Principle Investigator:

Nino Srour,, (301) 394-2623