Platform Intelligence focuses upon fundamental research that enables effective teaming of Soldiers and robots to conduct maneuver and military missions. ARL’s activities are centered upon enhancing the autonomous capabilities of unmanned systems. Knowledge gained in this area is expected to impact a wide array of vehicle systems, including the ground, air, and maritime domains, ranging from micro- to macro-scales. All of the research topics outlined in Platform Intelligence, except for “Atmospheric Characterization, Analysis, and Effects”, supports the Human-Agent Teaming and Artificial Intelligence/ Machine Learning ERA.
Meta-Cognition, Self-Reflection and Proprioception (APG)
We develop novel techniques that enable autonomous systems to track performance on assigned tasks, recognize failures, and adapt or adjust performance. We develop adaptable robotic behavior algorithms that enable autonomous robots to participate in small human/robot team missions with minimal human oversight utilizing techniques from the fields of statistical learning theory, artificial intelligence, and perception. Two topics of particular interest are recognizing and understanding team behaviors from visual and other sensory information; and adapting autonomous performance to dynamically changing conditions.
MaryAnne Fields, Ph.D., email@example.com, (410) 278-6675
APPLE: Adaptive Perception Processes for Learning from Experience (APG)
Develop a framework to integrate the perception, cognition, mission planning and experiential knowledge to achieve adaptive learning from experience. Find effective high-level object descriptors which can be used for online learning. Develop techniques to determine if detected objects are from previously learned classes or represent new classes that need to be dynamically added to the learned object categories. Complementary expertise is especially sought in unsupervised learning and feature discovery, scene understanding, knowledge representation, object recognition and statistical relational learning.
Intelligent Vehicle Technology Experimentation (APG)
ARL conducts formal assessments of integrated research capabilities of autonomous systems technologies. It applies the fundamentals of experimental design and applied statistics to enable assessment events that provide quantitative performance metrics and data. This data is used to track incremental progress of funded research, provide useful feedback to the researchers, and communicate to stakeholders that integrated performance was achieved. The experimental design and the data collection at assessment events are cooperative efforts led by the ARL assessment team with help from the researchers who provide iterative feedback on the experimental design, data capture and analysis tools. Experiments at Aberdeen Proving Ground are primarily conducted on Spesutie Island. These facilities include a 75’ x 450’ Unmanned Aerial System (UAS) runway with controlled airspace, an indoor 20’ x 20’ confinement cage with test stand, and an indoor 50’ x 30’ x 22’ Vicon motion capture area.
Human-Robot Interaction (APG and Orlando)
The goal of the Human-Robot Interaction (HRI) program is to maximize the effectiveness of integrating autonomous, intelligent systems technology into the Soldier team and operations through the development of state-of-the-art (or beyond) Soldier-system interactions. We seek to identify tools, techniques, and measures that can be used to improve and assess performance for human-systems teams. We are interested in research results that may be applicable across environments, operations, and platforms. Our HRI program focuses on Soldier-system teaming, Soldier-system bi-directional communications, and Soldier-system performance. Specific areas of interest include: naturalistic communications including multimodal and natural language interfaces; manned-unmanned teaming; ; human-systems team processes & performance; human interaction with microsystems; HRI metrics development; trust; transparency; situation awareness; and strategies for workload management.
A. William Evans, Ph.D., firstname.lastname@example.org, (410) 278-5982
Computational Intelligence (ALC)
Research seeks computationally feasible techniques and theories to ad¬dress the synergistic integration of components of intelligent behavior, including perception, reasoning, planning/execution, and decision making. These behaviors can be individual or collaborative. This includes: (1) semantic perception and scene understanding investigating innovative concepts for recognizing and under¬standing the world in which the system’s intelligent agents operate; (2) robot planning and behaviors in complex environments; (3) reasoning under uncertainty;
Stuart Young, email@example.com, (301) 394-5618
Reasoning and Decision Making Under Uncertainty (ALC)
Research in developing a method for analogical and deductive reasoning and decision making by automated systems such as intelligence analysis systems or battlefield robots. In order to deal with the inherent uncertainty that comes with military operations, this approach needs to be able to handle approximate and partial information. Our method includes both a rich common-sense knowledge base for deductive reasoning and a distributional semantic vector space trained on large volumes of text corpora for analogical reasoning at query-time. This combination gives the system the ability to answer questions approximately when full information is not available. Our system can, for example, reason that because a tank is similar to a truck, it probably contains a steering mechanism and a transmission.
Doug Summers-Stay, firstname.lastname@example.org, (301) 394 -1990
Atmospheric Characterization, Analysis and Effects [WSMR and ALC]
This research area seeks to fully-define atmospheric boundary layer physics, processes and climate variability, such as wind flow and turbulence generation in highly-complex terrain or in the urban street canyons of megacities. Theoretical and field studies are conducted to expand the knowledge-base about the precise workings of the near-surface atmospheric layer. The characterization research will be greatly augmented by the implementation of the Meteorological Sensor Array (MSA). The MSA is a facility now under development and fielding; initially at two sites within and near White Sands Missile Range, NM; but potentially at an East Coast site as well. This unique facility will be comprised of a highly-dense deployment of meteorological sensors, active and passive, both for in situ and remotely-sensed data collection. An understanding of the atmospheric characterization leads into research of those atmospheric and climatic effects on Army operations, infrastructure, and weapon systems to include renewable energy production systems. It is vitally important that Army Commanders have a clear understanding of the atmospheric/climatic impact on their mission. An important associated component of this research then, is the development of decision support applications which characterize and depict atmospheric and climatic impacts on systems, operations, and personnel; thereby providing battlefield commanders with readily-actionable intelligence information. Development environments will include Soldier/system-level crowd-sourcing of meteorological data from mobile platforms, and technologies for mitigation of adverse environmental conditions. Additionally, analyses will be 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.
Dr. Robb Randall, email@example.com, (575) 678-3123
Artificial Intelligence for Autonomous Systems (APG and ALC)
Artificial intelligence (AI) and neural networks are playing a key role in Army autonomous systems, both virtual and robotic. Despite rapid application and development of AI for autonomous systems, there are many basic unresolved issues. Components, such as convolutional neural networks (CNNs), recursive neural networks (RNNs), kernel based learners, and auto-encoders, are empirically state of the art, yet we lack fundamental understanding of their performance limits, how much reliance on supervised training is needed, or how robust and resilient these components can be. These AI elements will be coupled with reinforcement learning and other techniques for incorporation into systems such as autonomous robotic navigation and surveillance, and human-machine dialog. The invention and study of these emerging systems is critical for Army applications. In addition to the Human Agent Teaming and Artificial Intelligence and Machine Learning ERA, this work is also closely linked to the Distributed and Cooperative Engagement in Contested Environments ERA
Brian M. Sadler, Ph.D., Brian.firstname.lastname@example.org, (301) 394-1239