Artificial Intelligence and Machine Learning Technologies

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will conduct experiments with game theory and multi-agent reinforcement learning models and algorithms, integrating them with an available simulation framework to create courses of action (COAs) at the theater echelons; focus on refining and enhancing capabilities that support command and control, fires, and sustainment to ensure algorithm effectiveness in generating COAs.

FY2025 accomplishments Will design and develop game theory and multi-agent reinforcement learning and other foundational AI models and algorithms to integrate with an available simulation framework to create courses of action (COAs) at the theater echelons. Investigate and determine scenario criteria need for the algorithm to function, design and develop learning strategies and utility functions, and integrate the AI system into an available simulation suite to enable model training.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will conduct prototyping and experimentation with the integration of generative artificial intelligence capabilities to explore the effectiveness of achieving decision dominance by improving staff workflows in maintaining running estimates and improving situational awareness.

FY2025 accomplishments Develop AI-enabled common operating picture that surfaces ML/AI insights from the Sustainment, Intelligence, Fires, Protection, Movement and Maneuver, and Information Advantage warfighting functions.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will conduct experiments with the distributed AI framework that investigates the interoperability between generative AI models at the edge, cloud hosted models, and software based at the enterprise to inform emerging requirements in the command and control space.

FY2025 accomplishments Will design and develop a distributed AI framework, algorithm(s), abstraction layer, and human-distributed AI interface developed around All-Domain CONOPs. Will investigate the advances in algorithms, autonomy, and artificial intelligence and several key research areas to accelerate the capabilities and impact of Distributed AI capabilities for the US Army.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will conduct experiments to refine algorithms which will inform infrastructure and platform requirements for future artificial intelligence capabilities in support of mission command in command posts across all echelons.

FY2025 accomplishments Design and develop advanced algorithms for use by wider force and Operational Data Science Teams (ODSTs) to build and support emerging artificial intelligence enabled mission command information applications for the command post. Validates emerging lower echelon analytic platform tactical data fabric.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will develop and validate appropriate application(s) or system(s) leveraging emerging language-based AI technologies for mission command of operational forces.

FY2026 to FY2027 change Funding decrease reflects realignment to Program Element (PE) 0602146A (Network C3I Technology) / Project AN9 (UNT - Every Receiver is a Sensor Technology).

FY2026 plans — current year Will apply and experiment with AI agents for logistics and sustainment, medical, and operations in automated real-time strategy war games between synthetic agents at Corps and above echelons to explore possible improving workflow efficiencies in the staff.

FY2025 accomplishments Design and develop AI agents to employ METT-TC information available to Commanders to generate courses of action for threat formations as well as conduct AI-war gaming in support of Intelligence Preparation of the Operational Environment and the Military Decision Making Process. This effort will conduct experiments of automated real-time strategy war gaming between synthetic agents representing friendly and adversary forces at Corps and above echelons.

FY2026 to FY2027 change Funding decrease reflects completion of this effort.

FY2026 plans — current year Will validate operational prototype designs to support Army efforts to develop and deploy trusted AI/ML through fusion of intelligence data from multiple military intelligence systems; provide infrastructure components capable of implementing AI/ML algorithms across multiple domains.

FY2025 accomplishments Will continue to mature data frameworks and data pipelines for fusion of intelligence data from multiple military intelligence systems. Will continue to develop and conduct experiments with infrastructure components that can implement machine learning algorithms across multiple AI domains.

FY2026 to FY2027 change Funding decrease reflects completion of this effort.

FY2026 plans — current year Will develop and design a system of applications that utilize AI technologies to identify targets of interest and develop algorithms that use multiple data sources to predict representation for novel object classes from a small number of novel class samples; investigate the fusion of visual, language, signal, and event-based information and semantic relationships to learn new objects and relationships and validate knowledge transfer from base classes to novel classes to reduce the time it takes to train AI algorithms.

FY2025 accomplishments Will develop and mature a system of applications that utilize AI technologies to identify targets of interest and develop algorithms that use multiple data sources to predict representation for novel object classes from a small number of novel class samples. Will investigate the fusion of visual, language, signal, and event-based information and semantic relationships to learn new objects and relationships and validate knowledge transfer from base classes to novel classes to reduce the time it takes to train AI algorithms.

FY2026 to FY2027 change Funding decrease reflects realignment to Program Element (PE) 0602146A (Network C3I Technology) / Project AN9 (UNT - Every Receiver is a Sensor Technology).

FY2026 plans — current year Will design and develop an application for the purpose of investigating network science algorithms that apply natural language and low shot learning technologies; enable exploitation of social media platforms and publicly available information for increased battlefield awareness.

FY2025 accomplishments Will design, develop, and mature an application for the purpose of investigating network science algorithms that apply natural language and low shot learning technologies for the purposes exploiting social media platforms and publicly available information for increased battlefield awareness. Will experiment internally to determine which technical approaches are most effective at achieving the desired effect.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will improve algorithms for disambiguation of detected targets across multiple sensors of various modalities and resolutions; develop efficient techniques to adapt detection algorithms to mutable targets in novel environments.

FY2025 accomplishments Develop and experiment with the means to perform multi-scale detections on static and mobile targets, where initial detections from a wide-angle sensor are further discriminated using a detector that processes images with more pixels of the target provided by a separate pan, tilt, zoom (PTZ) sensor. Develop a cross-platform fusion model that uses the appearance of targets - to include 3D information to determine whether newly detected targets are the same as those previously reported to the common operating picture (COP). Develop and experiment with the means to pre-process imagery from sensors - using machine learning or computer vision - to optimize camera parameters so that high-quality…

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will develop improvements to algorithms used for teaming and information sharing between air and ground sensors with a focus on environments with low and intermittent network connectivity; develop adaptable autonomous behaviors for different mission areas, environments, or targets; improve autonomy algorithms to improve collaborative behaviors between air and ground vehicles.

FY2025 accomplishments Develop and mature 3D stereo data self-registration techniques to improve robustness of perception in rough terrain by correcting for pose estimation error. Integrate multi-scale processing techniques (e.g., variable resolution and frame rates) to improve robustness of perception at higher traversal speeds. Develop a module that optionally activates and leverages data from a LiDAR sensor when the threat of detection is minimal. Develop and demonstrate autonomous operation without using or dependency on a global prior cost map. Develop terrain awareness for autonomous UAS's - using pre-loaded or referenced elevation data - so that UAS's avoid hazardous terrain features and can self-identify…

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will develop efficient techniques, interfaces, information display, and user feedback to allow a single user to appropriately task many air and ground sensors; develop improved system feedback mechanisms to the operator; develop the ability for multiple users to split and hand off control of sensors as mission dictates; improve the user experience to allow full control and tasking ability of air and ground sensors throughout a mission, to include robust fault identification and user feedback for recovery.

FY2025 accomplishments Mature the User Interface/User Experience (UI/UX) to develop an updated messaging solution that supports interoperability to the dismounted, mounted and fires community as an improved Android Tactical Assault Kit (ATAK) plug-in across multiple WfF. The UI/UX would define critical command and control messages for the air and ground robots to ensure the protocol specification includes the automatic acknowledgement and retransmission of these messages that communicate to the Tactical Operations Center. Develop algorithms to reside on the robots and verify whether missions received from ATAK are valid (e.g., whether on area designated for reconnaissance is feasible based on platform range or…

FY2026 to FY2027 change Funding decrease reflects transfer of effort to Program Element (PE) 0602146A (Network C3I Technology) / Project AN9 (UNT - Every Receiver is a Sensor Technology) to streamline and optimize the Science & Technology (S&T) portfolio.

FY2026 plans — current year Will design and develop novel algorithms to detect malicious data actions and deep fakes in computer vision applications and develop tools for the self-testing and evaluation of AI/ML models for robustness against adversarial activities.

FY2026 to FY2027 change Funding decrease reflects termination of this initiative and the strategic reallocation of resources to support evolving priorities.

FY2026 plans — current year Will capitalize on component level analysis by developing models to predict part replacements based on training cycles and major maintenance inspections; utilize the model development pipeline, predictive capabilities will be expanded across the force to identify when equipment will be down and when it will be fixed based on the maintenance, operations, and personnel landscape within a unit.

FY2025 accomplishments Designs and develops models for serialized component level analysis that are enhanced with non-serialized component information based off fault write-ups associated with a particular sub-component. Matures the model development and deployment pipeline to provide the ability to train, retrain, or update the component model and redeploy to the flight line in mission relevant time for predictive analytics. Predictive maintenance modeling will be expanded to proper maintenance management to allow for battalion maintenance officers to properly manage their unit's maintenance program and forecast upcoming scheduled and unscheduled maintenance.

FY2026 to FY2027 change Funding decrease reflects the strategic reallocation of resources to support evolving priorities and objectives.

FY2026 plans — current year Will develop the data pipeline from accessions through initial military training to assess Soldier performance; validate models to identify future Soldiers and increase the quantity and quality of Soldiers entering the accessions process; validate models which can predict Soldier graduation rates through initial military training; model individual and collective unit mission essential tasks to analyze and predict mission readiness.

FY2025 accomplishments Will investigate the scalability of the application to enterprise-level requirements. This will include, but not limited to, identifying various datasets of interest that are relevant to various skill sets, education, training, and expertise of candidates, investigating and analyses of these datasets by using natural language processing, large language models and other means. This project will design and develop algorithms to identify complementary team members and recommend individual substitutions, along with the retention of individuals to improve and maintain team performance.

FY2026 to FY2027 change Funding decrease reflects transfer of effort to Program Element (PE) 0602146A (Network C3I Technology) / Project AN9 (UNT - Every Receiver is a Sensor Technology) to streamline and optimize the Science & Technology (S&T) portfolio.

FY2026 plans — current year Will investigate maintenance data to include the full spectrum of information necessary to integrate multiple programs of record including data streams for operations, personnel, and maintenance; validates different predictive modeling techniques which increase decision making capabilities for the warfighter.