What project AA9 buys
This project supports basic research to enable intelligent and survivable command, control, communication, computing, and intelligence (C4I) systems for the future force. As the combat force structure decreases and operates in more dispersed formations, information systems must be more robust, intelligent, interoperable, and survivable if the Army is to retain both information and maneuver dominance. This research addresses the areas of information assurance, signal processing for wireless battlefield communications, information extraction from multi-modal data human-agent naturalistic communication, and intelligent systems for C4I. Research will focus on understanding and solving inherent vulnerabilities associated with using standardized protocols and commercial technologies while addressing survivability in a unique hostile military environment that includes highly mobile nodes and infrastructure, bandwidth-constrained communications at the edge, resource-constrained sensor networks, diverse networks with dynamic topologies, high-level multi-path interference and fading, jamming and multi-access interference, levels of noise in speech signals and document images, and information warfare threats. These C4I technologies must accommodate heterogeneous security infrastructures, multi-service and multi-national interoperability, and information exchange/security mechanisms between multiple levels of security. The intelligent systems for C4I research focus on providing machine learning methods to overcome noisy, sparse, and heterogeneous data with artificial intelligence algorithms that can transfer learning from one domain to another. This foundational research will help identify highly relevant tactical events for mounted or dismounted commanders, leaders and Soldiers; improve the timeliness, quality, and effectiveness of actions; and speed the decision-making process of small teams operating in complex natural or urban terrain. Work in this project supports key Army needs and provides the theoretical underpinnings for Program Element (PE) 0602146A (Network C3I Technology), PE 0602143A (Soldier Lethality Technology), and PE 0602145A (Next Generation Combat Vehicle Technology). Work in this project is performed by the Army Research Laboratory (ARL).
Project AA9 funding, FY2025–FY2031
Prior years are actuals, the budget year is the request, and the outyears are the FYDP plan. Estimate types are colored and never summed into one figure. Projects carry the full five-year plan; the activities inside them stop at the budget year.
| Fiscal Year | Estimate Type | Amount ($M) |
|---|---|---|
| FY2025 | Actual | 43.4 |
| FY2026 | Enacted | 30.9 |
| FY2027 | Request | 32.0 |
| FY2028 | Outyear | 32.5 |
| FY2029 | Outyear | 37.0 |
| FY2030 | Outyear | 37.6 |
| FY2031 | Outyear | 37.9 |
16 accomplishments / planned programs
The R-2A exhibit — the only level of the budget that describes work that has not happened yet. Activities carry the prior, current and budget year only, no five-year plan. Coverage is partial across the corpus, so count activities, never total them.
FY2027 planned work Will explore the resilience of distributed analytics that account for dynamics in multi-domain environments and constrained network and computational resources; conduct research on novel methods for the control of distributed analytics involving multiple information modalities and dynamic environments; explore novel architectures for network experimentation integrating emulated and real-world networking/communications/computational systems; explore network models incorporating heterogeneous autonomous agents and dynamics in both the physical and electromagnetic domain to enable the study and validation of novel approaches for network signature management, deception, and resilience against…
FY2026 to FY2027 change This is not a new start effort. FY 2027 funding increase reflects the consolidation of other ongoing efforts within this project from Communications in Complex Dynamic Networks and Information Protection for Mobile Dynamic Networks to support the creation of Communications in Distributed Dynamic Networks. Funding increase reflects additional research in in-situ quantum network characterization.
FY2027 planned work Will conduct research into minimizing deep learning model complexity, mitigating model bias, incorporating algorithmic rule-based approaches, enabling sequential artificial reasoning, and increasing explainability in foundation models and generative artificial intelligence (AI); investigate methods for enabling intelligent software agents and autonomous systems with the adaptability, explainability, and artificial reasoning capabilities of agentic AI; explore natural language understanding, semantic comprehension, and sequential reasoning techniques for the interpretation of multi-modal information sources within foundation models for shared understanding during human-machine collaborative…
FY2026 to FY2027 change This is not a new start effort. FY 2027 funding increase reflects the consolidation of other ongoing efforts within this project from Advanced Computing Architectures and Algorithms and Machine Learning for Intelligent Agent and Human Decision Making to support the creation of Advanced Computing for Modeling and Learning. Funding increase reflects additional research in intelligent software agents and acceleration technologies for optimized ML model deployment.
FY2027 planned work Will examine novel anapole resonators for coupling to single site color centers in solids for enhancing quantum sensor sensitivity; explore large sample characterization throughput for material optimization; experimentally investigate the simulated prediction that operating in a nonlinear regime could enable steeper discriminator and enhanced signal-to-noise; investigate application of similar techniques to atoms and ions in vacuum, such as Rydberg atoms for electric field sensing and trapped ions for quantum information processing; study the effects of stray surface charges and investigate resonator material coatings to mitigate deleterious perturbations.
FY2026 to FY2027 change Funding increase reflects additional research in the area of large sample characterization materials optimization and resonator material coatings to mitigate deleterious perturbations.
FY2026 plans — current year Will investigate techniques, such as pulsed interrogation, to advance electric-field sensor sensitivity towards the quantum noise limit; investigate the previously developed coupling and resonator designs for generating quantum states that offer sensor advantage over classical states; explore integrated photonic devices for coupling optical modes to quantum spins; develop solid state quantum magnetic field sensor that operates at better than the thermal noise limit set by its physical temperature; explore the rate of entanglement generation between a trapped spin qubit and a telecom-wavelength photon.
FY2025 accomplishments Will investigate new resonator geometries for field concentration that improves efficiency in light-matter coupling; investigate trade-offs between small mode volume waveguides/resonators and perturbations to material quantum bits from nearby surfaces; explore new geometries for resonant coupling, including 2-Dimensional and 3-Dimensional designs, and characterize the relative quality factors, coupling strengths, repeatability, and scalability; analyze approaches for both vapor-phase atoms and solid-state atom-like color centers and explore these in the context of improving quantum clocks, sensors, and quantum bit processing capabilities.
FY2027 planned work Will explore the ability to leverage advanced artificial intelligence (AI) models optimized via an automated pipeline on resource constrained edge devices enabling local processing without significant reach back over Denied, Disrupted, Intermittent, and Limited environments (DDIL) networks; conduct research into adaptive methods for resiliency enabling complex distributed workflow execution across tactical DDIL networks; investigate methods to convert multimodal data into text-based information that can be processed by language models for improved situational awareness; explore novel cyber defenses that include advanced methods of deceiving or diverting adversaries who have penetrated…
FY2026 to FY2027 change Funding increase reflects additional research in the area of DDIL networks.
FY2026 plans — current year Will investigate and develop frameworks to support data ingest and dissemination across command and control information system (C2IS) infrastructure that include intelligent adaptive strategies to optimize network performance and provide accurate information recommendation based on context; investigate improved generalization, robustness, and explainability of machine learning models through the development of physics-motivated data augmentation strategies and datatype-specific layers.
FY2025 accomplishments Will investigate and understand commercial off the shelf (COTS), domain specific processors for perception algorithm inference performance with splicing and partitioning of large neural networks; study methods of real time processing to support autonomous systems; research machine learning techniques for the cyber/electromagnetic domain, robust to adversarial detection and interference; explore machine learning techniques to identify and correct atmospheric distortions to support assured targeting; investigate methods for deep reinforcement learning based on novel information criteria; conduct research on bounded, incremental learning in real-time systems; investigate transfer of machine…
FY2027 planned work Will conduct research into leveraging self-play learning to explore strategies leading to the identification and exploitation of WoO; examine neuro-symbolic information fusion strategies that build upon the various modalities, such as video and language, and connect them to knowledge representations to enhance the identification of WoO; examine multi-agent reinforcement learning (MARL) for air and ground robotic and autonomous systems (RAS) to learn to optimally identify, create, and explore WoO across domains; explore a multi-player game theoretic investigation of blue and red autonomous agents in non-stationary environment to understand the performance of MARL; extend quickest change…
FY2026 to FY2027 change Funding decrease reflects reduction in research supporting human-agent interaction.
FY2026 plans — current year Will examine merits of computational models of artificial reasoning to identify WoO in specific domains for effective autonomous agent behavior and adaptable automated decision-making; study initial promising methods of natural language understanding, multimodal information extraction, and advanced knowledge representations to enable human-agent interaction and collaboration to predict WoO during mission execution; examine computer vision model for multi-modal sensing to detect threats and adversarial intent in hostile military environments; study efficient communication methodologies for formation controls; investigate reasoning frameworks and hybrid neuro-symbolic machine learning models…
FY2027 planned work Will investigate adversarial machine learning methods to understand vulnerabilities of artificial intelligence/machine learning (AI/ML) vision-language models to protect against adversarial attacks.
FY2026 to FY2027 change Funding increase reflects additional research in the area of adversarial AI.
FY2026 plans — current year Will research artificial intelligence/machine learning (AI/ML) multi-agent frameworks capable of hierarchical learning and multi-modal scene understanding to support autonomous maneuver of unmanned aerial and ground vehicles in complex environments; investigate network dissection techniques to understand the characteristics and differences of features learned by neural networks trained on real and synthetic datasets.
FY2025 accomplishments Will investigate self-supervised, multimodal perception models on size, weight, and power (SWaP)constrained, mobile platforms combined with natural language supervision to address the austere operating conditions, including the data scarcity problem, in rapidly learning, critical battlespace representations in tactical environments; investigate a combined synthetic rendering and perception model that enhances the realism of scene synthesis, while creating large scale, unseen novel view images with high fidelity to significantly enhance perception performance at the edge.
FY2026 to FY2027 change Funding decrease reflects realignment to Communications in Distributed Dynamic Networks within this project.
FY2026 plans — current year Will explore robustness for resource-adaptive analytics that allocate resources while accounting for dynamics in multi-domain environments and constrained network and compute resources; explore novel methods for network understanding and network control in multilayer, dynamic networks; develop and characterize a variable-fidelity network modeling framework incorporating highly dynamic and heterogeneous autonomous agents/nodes to enable the exploration of intelligent protocols that enhance resilience and applicability (in terms of capability set, size, weight, power, mobility, etc.); develop reinforcement-learning-based approaches to enhance the performance of extremely heterogeneous…
FY2025 accomplishments Will investigate novel decentralized strategies leveraging learning-based approaches for the control of extremely heterogeneous networks; explore directional networking capabilities within extremely heterogeneous networks through opportunistic beamforming to increase network performance and enhance stealth; explore resource-adaptive analytics techniques in multi-domain environments to account for dynamic environments with constrained network and computing resources; explore novel methods for resilient, dynamic, multilayer network analytics in complex network environments; investigate machine learning-based techniques for efficient and distributed placement and adaptation of complex…
FY2025 accomplishments Will explore eye movement tracking in augmented reality (AR) display for controlling autonomy assets for human-agent teaming; investigate rule-based algorithms and data-driven machine learning methods for knowledge network construction and information extraction approaches applied to natural language interpretation to enable effective automated text generation for information management tasks; conduct fundamental research into computational models of artificial reasoning to enable automated decision making that considers the impact of uncertainty and associated risks, multiple criteria, and mission context.
FY2026 to FY2027 change Funding decrease reflects realignment to Communications in Distributed Dynamic Networks within this project.
FY2026 plans — current year Will explore extensions of the classical shadow formalism for quantum state characterization that include prior information about the experimental system under investigation; study the requirements and feasibility of basic light handling operations required for quantum networking using integrated photonic platforms; investigate methods to mitigate realistic environmental noise and decoherence to improve the fidelity of quantum entanglement distribution over deployed fiber links; research algorithms and methodologies to identify and create Cyber Windows of Opportunity to create advantages in tactical operations; investigate the use of advanced machine learning algorithms to improve the…
FY2025 accomplishments Will analyze the accuracy and resource requirements of competing approaches for quantum state characterization in networks, including shadow tomography, full-state tomography, and machine learning-based techniques; study various approaches and platforms for performing basic quantum networking tasks, such as quantum frequency conversion and low-loss optical switching; study entanglement distribution over long fiber links, extending to remote physical sites, to assess realistic environmental noise and decoherence impacts; research basic algorithms and methodologies to encapsulate the technical foundations of an autonomous, intelligent cyber-defense agent for traditional networks and…
FY2026 to FY2027 change Funding decrease reflects realignment to Advanced Computing for Modeling and Learning within this project.
FY2026 plans — current year Will leverage models optimized based on characteristics and environment to realize optimized analytic tasks for heterogeneous devices; investigate online learning of the dynamic interactions between devices, applications, and data as it relates to analytics in resource constrained, tactical environments; explore development of software tools/emulator allowing more rapid assessment of potential field programmable neural array (FPNA) chip design modifications leading to potentially more advanced on-chip applications; investigate techniques to optimize Large Language Models (LLM) for use on resource-constrained devices; investigate techniques to optimize multi-modal AI models to efficiently…
FY2025 accomplishments Will study field programmable neural array (FPNA) to understand performance and computational efficiency on small neural networks; conduct research on analog neurons and use for complex, symbolic processing and inferencing; investigate strategies to characterize and predict analytic performance in resource-constrained, heterogeneous operational regimes; explore methods to identify poor analytic performance due to dynamic or complex information and resolve analytic accuracy with decentralized and distributed model learning; investigate methods to autodetect referenced model architecture, key features, framework, and its attributes in order to prioritize specific model optimizations and…
FY2026 to FY2027 change Funding decrease reflects realignment to Advanced Computing for Modeling and Learning within this project.
FY2026 plans — current year Will investigate computer vision algorithms to enable machines and systems to detect partially occluded objects, detect and track target objects, understand threat environment by multi-modal sensing in scenes, and enable object detection algorithms in High Dynamic Range (HDR) environment; develop methods and techniques that leverage shared representations to transfer knowledge learned during exploration to other agents; develop algorithmic techniques that learn role assignments in multi-agent adversarial teams based on limited observations.
FY2025 accomplishments Will investigate and conduct research on methods grounded in information theory and/or game theoretic approaches for collaborating multi-agent systems to share information in constrained environments; investigate machine learning (ML) methods for computer vision to enable autonomous systems to detect objects in high dynamic range (HDR) environments; conduct experiments with small teams of multi-agent systems to assess ability to autonomously adapt group behaviors based on partially observed reinforcement learning signals; investigate methods and techniques that allow multi-agent systems to adapt role assignments based on high level, human defined strategies; conduct research on algorithms…
FY2025 accomplishments Will conduct research into microelectronic design processes and techniques that renders device purpose unclear to frustrate reverse engineering while preserving efficiency and function; explore diamond heterostructure and transistor acceptor layer material properties; identify charge traps, impurities, and interface atomic bonding characteristics in order to improve the efficiency of radio frequency (RF) diamond transistors; examine high electron mobility transistor switch with a ferroelectric nitride to understand improved energy efficiency savings; explore Ultra-Wide Bandgap (UWBG) device designs that maximize lifetime under high energy alpha and beta radiation.
FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.
FY2026 plans — current year Will characterize climate relationships between teleconnection patterns (causal connections or correlations between meteorological or other environmental phenomena which occur a long distance apart) and the energy state at the surface, specifically surface sensible and latent energy flux (Bowen Ratio).
FY2025 accomplishments Will analyze Distributed Virtual Proving Ground (DVPG) meteorological array databases to understand the evapotranspiration cycle and flash drought onset; investigate and understand boundary layer process impacts on climatology in complex environments.
FY2025 accomplishments Will conduct research on architectures and representations for joint object detection, localization, and classification from multiple sensor modalities; research techniques for on-demand generation of synthetic data and model tuning for adapting to changing environments; investigate methods to manage information flow and communicate in a timely, effective, and adaptive manner across domain and echelon; investigate information dynamics and behaviors to develop tactics, techniques, and procedures toward resiliency against adversarial campaign; investigate novel, artificial reasoning techniques for robust, automated decision making; investigate fundamental techniques for natural language…
FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.
FY2026 plans — current year Will validate research on human-guided machine learning approaches using large language models to generate courses of actions at different scales; validate how human-guided machine learning-based course of action generation effects human situational awareness and trust.
FY2025 accomplishments Will conduct research on initial human-guided machine learning approaches using large language models to generate courses of actions at different scales; investigate how human-guided machine learning-based course of action generation influences trust amongst human users with different roles.
FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.
FY2026 plans — current year Will explore fundamental issues in characterizing and communicating the uncertainty within unstructured data, task execution, information sources, and machine learning models that decrease the accuracy and robustness to dynamic environments of autonomous agents and intelligent systems; explore computational models of uncertainty to increase transparency, interpretability, and explainability in AI systems.