10 Jul 2026

Microsoft Azure Artificial Intelligence / Machine Learning Hackathon for Development of Retrieval-Augmented Generation Large Language Model

Abstract: The US Army Corps of Engineers (USACE) Civil Works (CW) research and development (R&D) mission is to address challenging environmental sustainability problems through innovative science and engineering, which helps to ensure a safer, more prosperous, and more resilient nation. To achieve this, the US Army Engineer Research and Development Center (ERDC) plans, executes, leads, and directs many R&D programs in coordination with USACE Headquarters, Districts, and Divisions through its multiple strategic focus areas, which include infrastructure, water modeling, crisis preparedness, ecosystem, sediment management, data, artificial intelligence, and robotics. In this process, much information is generated, including internal progress reviews, financial reports, scopes of work, work package planning, and success stories.

10 Jun 2026

Multisource Knowledge Graph Architecture for Air-Gapped AI Systems: Design Patterns and a Geospatial Reference Implementation

Purpose: This technical note presents a reference architecture for constructing multisource knowledge graphs in air-gapped, domain-specific AI systems. The architecture addresses recurring problems in military, intelligence, and secure-enterprise environments: integrating heterogeneous authoritative data sources while preserving source schema fidelity, enabling deterministic semantic resolution, and operating without external network dependencies. While this is a geospatial implementation, the architectural principles in the core design are transferable. The implementation is organized around four separable layers—schema registry, domain ontology, reasoning patterns, and relationship vocabulary—and adopts a canonical-with-aliasing integration strategy that supports cross-schema reasoning without forcing premature schema con-vergence. These patterns are validated through a geospatial intelligence implementation supporting US Army operations and demonstrate how abstract design principles translate into an operationally relevant system. This knowledge graph is designed to serve as the semantic substrate for router-based AI systems. A companion technical note (Drouillard and Lewis 2026) describes the geospatial AI (GeoAI) agent stack, which is a router-based orchestration architecture that coordinates multiple retrieval backends and reasoning tools. Within that architecture, the knowledge graph functions as a specialized retrieval backend that runs alongside document retrieval and vector search, serving queries that require structured entity-relationship reasoning, provenance tracking, or deterministic semantic resolution. The router directs spatial relationship queries (e.g., “which roads cross this river”), multihop dependency queries (e.g., “what infrastructure depends on this power station”), and schema-resolution queries (e.g., “find all transportation features in Multinational Geospatial Co-production Program [MGCP] format”) to the knowledge graph while routing conceptual or analytical questions to document retrieval. This technical note focuses exclusively on the knowledge graph architecture; the broader orchestration patterns and routing logic are detailed in the companion paper. While the examples presented are geospatial, the architectural principles, validation strategies, and design tradeoffs documented here have broader applicability where deterministic semantic integration is required under air-gapped constraints. The geospatial instantiation serves as a concrete demonstration of abstract patterns that may inform future knowledge graph efforts in other US Army Engineer Research and Development Center (ERDC) research domains.

13 May 2026

Geospatial AI (GeoAI) Agent Stack: Router-Based Orchestration and Design Rationale

Abstract: This report summarizes the current state of a router-based, multiagent Geospatial AI (GeoAI) system designed to reliably execute geospatial workflows while retaining the flexibility of large language model (LLM) reasoning. The architecture is intentionally both language-model agnostic and orchestration-framework agnostic to support organizational controls and mandates, and it is designed to operate in air-gapped environments. It uses a domain router to scope tools before the model is invoked; a microrouter to decide whether the system should execute tools, retrieve knowledge, or produce a direct response; and a bounded cycle of execution and validation that supports multistep tool use. The design emphasizes determinism after the model makes decisions, strict boundaries around what the model can “see,” and modularity that keeps core business logic largely independent from orchestration and tool-protocol frameworks. The remainder of this report describes the architecture as implemented today, explains the design rationale, and outlines anticipated future work.

29 Apr 2026

Deep Learning Approaches for Buried Object Detection in Infrared Imagery

Abstract: Artificial intelligence and machine learning techniques are increasingly utilized to detect buried objects in thermal infrared imagery. This task relies heavily on the quality and diversity of the training dataset, requiring datasets that capture variability present in real-world environments. Synthetic imagery offers a means to expose algorithms to a greater range of conditions than is often available in real-world data alone. This study evaluates the performance of three open-source object detection models—Faster Region-Based Convolutional Neural Network (R-CNN), You Only Look Once (YOLOv8), and Single Shot Multibox Detector—trained using real-world, synthetic, and hybrid datasets. Real-world imagery was collected from a single field site, while synthetic data were generated using the Virtual Environmental Simulation for Physics-Based Analysis software suite. Model performance was evaluated using Intersection over Union and confidence scores. Models trained exclusively on synthetic datasets with limited scene diversity, when tested on real-world imagery from the same location, produce high false-positive and false-negative rates. Detection performance im-proved significantly for Faster R-CNN and YOLOv8 when trained using a hybrid dataset combining real-world and synthetic data. Analysis of red-green-blue histograms revealed differences in pixel intensity distributions between real and synthetic imagery, indicating areas for improving synthetic data generation.

25 Apr 2025

Analysis Tools and Techniques for Evaluating Quality in Synthetic Data Generated by the Virtual Autonomous Navigation Environment

Abstract: The capability to produce high-quality labeled synthetic image data is an important tool for building and maintaining machine learning datasets. However, ensuring computer-generated data is of high quality is very challenging. This report describes an effort to evaluate and improve synthetic image data generated by the Virtual Autonomous Navigation Environment’s Environment and Sensor Engine (VANE::ESE), as well as documenting a set of tools developed to process, analyze, and train models from, image datasets generated by VANE::ESE. Additionally, the results of several experiments are presented, including an investigation into using explainable AI techniques, and direct comparisons of various models trained on multiple synthetic datasets.

5 Nov 2024

Innovations of Cellular Automata

Purpose: In the past several years, there has been a rather substantial uptick in the amount of research within the realm of cellular automata due to its ability to produce complex, self-organizing behavior from simplistic rulesets. The capability to produce this behavior is essential to understanding artificial life and intelligence. This uptick has resulted in numerous novel directions for experimentation within this computational playground. This work summarizes a few of the most impactful directions that have resulted from this research.

9 Sep 2024

Leveraging Artificial Intelligence and Machine Learning (AI/ML) for Levee Culvert Inspections in USACE Flood Control Systems (FCS)

Abstract: Levee inspections are essential in preventing flooding within populated regions. Risk assessments of structures are performed to identify potential failure modes to maintain the safety and health of the structure. The data collection and defect coding parts of the inspection process can be labor-intensive and time-consuming. The integration of machine learning (ML) and artificial intelligence (AI) techniques may increase accuracy of assessments and reduce time and cost. To develop a foundation for a fully autonomous inspection process, this research investigates methods to gather information for levees, structures, and culverts as well as methods to identify indicators of future failures using AI and ML techniques. Robotic plat-form and instrumentation options that can be used in the data collection process are also explored, and a platform-agnostic solution is proposed.

18 Apr 2024

Artificial Intelligence (AI)–Enabled Wargaming Agent Training

Abstract: Fiscal Year 2021 (FY21) work from the Engineer Research and Development Center Institute for Systems Engineering Research lever-aged deep reinforcement learning to develop intelligent systems (red team agents) capable of exhibiting credible behavior within a military course of action wargaming maritime framework infrastructure. Building from the FY21 research, this research effort sought to explore options to improve upon the wargaming framework infrastructure and to investigate opportunities to improve artificial intelligence (AI) agent behavior. Wargaming framework infrastructure enhancements included updates related to supporting agent training, leveraging high-performance computing resources, and developing infrastructure to support AI versus AI agent training and gameplay. After evaluating agent training across different algorithm options, Deep Q-Network–trained agents performed better compared to those trained with Advantage Actor Critic or Proximal Policy Optimization algorithms. Experimentation in varying scenarios revealed acceptable performance from agents trained in the original baseline scenario. By training a blue agent against a previously trained red agent, researchers successfully demonstrated the AI versus AI training and gameplay capability. Observing results from agent gameplay revealed the emergence of behavior indicative of two principles of war, which were economy of force and mass.

17 Apr 2024

Enabling Understanding of Artificial Intelligence (AI) Agent Wargaming Decisions through Visualizations

Abstract: The process to develop options for military planning course of action (COA) development and analysis relies on human subject matter expertise. Analyzing COAs requires examining several factors and understanding complex interactions and dependencies associated with actions, reactions, proposed counteractions, and multiple reasonable outcomes. In Fiscal Year 2021, the Institute for Systems Engineering Research team completed efforts resulting in a wargaming maritime framework capable of training an artificial intelligence (AI) agent with deep reinforcement learning (DRL) techniques within a maritime scenario where the AI agent credibly competes against blue agents in gameplay. However, a limitation of using DRL for agent training relates to the transparency of how the AI agent makes decisions. If leaders were to rely on AI agents for COA development or analysis, they would want to understand those decisions. In or-der to support increased understanding, researchers engaged with stakeholders to determine visualization requirements and developed initial prototypes for stakeholder feedback in order to support increased understanding of AI-generated decisions and recommendations. This report describes the prototype visualizations developed to support the use case of a mission planner and an AI agent trainer. The prototypes include training results charts, heat map visualizations of agent paths, weight matrix visualizations, and ablation testing graphs.

16 Feb 2024

A Comprehensive Review on Wood Chip Moisture Content Assessment and Prediction

Abstract: Wood chips are the primary sources of raw materials for numerous industries, including pelleting mills, biorefineries, pulp-and-paper industries, and biomass-based power generation facilities. Unfortunately, when wood chips are utilized as a renewable and environmentally friendly resource, industries are constantly challenged by the consistency of the wood chip qualities (e.g., moisture/ash contents, size distributions) - a historically recognized problem on a global scale. Among other wood chip quality attributes, the moisture content is considered the most pressing one as it directly impacts the energy content, storage stability, and handling properties of the raw and finished products. Therefore, accurate wood chip moisture content prediction can help optimize the drying process and reduce energy consumption. In this review, a survey was conducted on various techniques and models employed for predicting wood chip moisture content. The advantages and limitations of these approaches, as well as their potential applications and future directions were also discussed. This review aims to provide a comprehensive overview of the current state-of-the-art in wood chip moisture content prediction and to highlight the challenges and opportunities for further research and development in this field.