Technical Track Abstracts – Wednesday, Nov 4

Track 1 AI Enablers for Rapid T&E

Agentic AI for Mission Assurance: Automating Digital Integration and Test from Requirements to Verification
Presenter: Dr. Policarpio Soberanis

We present a model-based digital testing framework that leverages large language models (LLMs) to automate the translation of structured test plans into executable mission scenarios within a simulation environment. The approach begins by parsing a formal test plan using an LLM, which interprets requirements, constraints, and test objectives to generate a corresponding mission configuration in a physics-based system simulation platform. Within this environment, a reduced-order model (ROM) of the system under test is developed using AI-assisted modeling techniques, enabling efficient yet high-fidelity representation of system behavior.

The ROM is integrated into a ROM factory pipeline, where it is continuously refined and validated using empirical data from targeted physical testing. This closed-loop calibration ensures alignment between simulated and real-world performance, increasing confidence in the model’s predictive capability. The validated ROM supports verification activities by enabling rapid execution of test scenarios and automated assessment of requirement compliance. Traceability is maintained throughout the workflow, linking system requirements to test plans, simulation artifacts, and verification results.

The framework supports a continuous integration paradigm for digital engineering, where updates to system requirements trigger re-evaluation against the current ROM, and updates to the ROM prompt re-validation against existing requirements. This bidirectional feedback loop establishes a DevOps-style pipeline for digital integration and digital testing (DI/DT), enabling continuous verification, rapid iteration, and sustained alignment between design intent and system performance. To further automate and orchestrate this workflow, an agentic AI layer is introduced to autonomously manage task sequencing, data exchange, and decision-making across the pipeline. These intelligent agents monitor requirement changes, initiate simulation updates, trigger ROM retraining with new empirical data, and execute validation campaigns without human intervention, ensuring a resilient, adaptive, and self-improving digital testing ecosystem.

Edge AI In Test Instrumentation with ATARI
Presenter: Michael M. Delaney Jr, P.E.

There is a need to accelerate Flight Test and Evaluation, compressing the time from test point complete to useful information. To meet this need in a spectrum efficient manner, flight test instrumentation systems need to incorporate High-Performance Computing (HPC) with Artificial Intelligence and Machine Learning (AI/ML) capabilities. This presentation will provide an overview of use cases for both test acceleration and spectrum efficiency, as well as an overview of the AI Test Aircraft Research and Integration (ATARI) program at Edwards AFB. This overview will include an update on current and future program plans, challenges, and early lessons learned.

Neuromorphic Aircraft Stores-Centric Expendable and returnable Networked Technology (ASCENT) – Forging operational advantage with intelligent & autonomous armament systems in the Synthetical Age
Presenter: Dr. Malcolm G. Tutty

Aircraft stores separation techniques have long relied on traditional methodologies such as wind tunnel testing, computational fluid dynamics, modelling and simulation, and flight testing as outlined in MIL-HDBK-1763. While these approaches have proven very effective, they are often costly, time-intensive, and constrained by outdated assumptions. This paper updates progress made with Neuromorphic Aircraft Stores-Centric Employable and returnable Networked Technology (ASCENT) at University of NSW to leverage neuromorphic event cameras, advanced instrumentation, and neural networks for determining aircraft stores compatibility and interoperability by integrating physics-informed neural networks and combinatorial test technique use, Neuromorphic ASCENT offers a significant reduction to certification timelines and costs while enhancing operational safety and efficiency.

Neuromorphic ASCENT employs event-driven processing inspired by human neural functions, enabling real-time decision-making and adaptability across diverse platforms. Neuromorphic event cameras focus on significant changes in the environment, providing precise imagery and reducing computational overhead. These cameras, combined with instrumented measurement units and PINNs, allow for tailored ejection forces and trajectories, minimizing structural stress on aircraft and ensuring safe autonomous separation of stores. Unlike traditional stores, which rely on high ejection velocities and onboard cameras, neuromorphic stores autonomously “know” their position relative to the aircraft and safely navigate away at lower velocities. This capability eliminates the need for platform-mounted cameras and enhances operational flexibility. The paper explores operational implications of Neuromorphic ASCENT, including its role in enabling returnable aircraft stores for refuelling, rearming, reprogramming, and secure analysis. This shift supports reuse of stores, reducing costs and promoting sustainability. Additionally, Neuromorphic ASCENT enhances the F2T2EA kill chain process by introducing enhanced continuous “Assess” and “Abort” criteria using a joint fires armament integrated mission environment needed for complex, adaptive kill webs in the Synthetical Age. Forward-facing cameras on stores capture real-time data to refine abort/assess decisions, enabling dynamic mission adjustments and safe return of stores to other platforms. Neuromorphic ASCENT also addresses critical challenges in current aircraft stores compatibility and interoperability processes. By leveraging advanced CFD models and simulation tools, the technology ensures aerodynamic, structural, electrical, and functional compatibility across legacy and next-generation systems. Combinatorial test techniques further optimize certification processes, reducing sample sizes while improving fault detection rates. This approach supports multi-platform and multi-weapon integration, enabling accurate predictions across diverse flight conditions and scenarios.

The paper highlights the broader implications of Neuromorphic ASCENT for modern warfare. Its neuromorphic architecture facilitates real-time coordination of distributed assets, enhancing mission success rates and operational safety. The technology’s adaptability and reusability align with the evolving needs of military operations, offering scalable and sustainable solutions. By overcoming limitations in traditional testing methods and operational trade-offs, Neuromorphic ASCENT paves the way for smarter, safer, more efficient military systems and represents a transformative leap in aircraft stores technology, combining brain-inspired computing with advanced neural networks to redefine compatibility, interoperability, and operational efficiency. This innovation not only reduces costs and development timelines but also lays the foundation for a more adaptable and sustainable future in military technology. By integrating cutting-edge neuromorphic capabilities, Neuromorphic ASCENT is poised to revolutionize the science and engineering underpinning aircraft stores compatibility, ensuring mission success in the face of increasing complexities and the character of modern warfare.

Track 2 – Test Automation

Test Automation Framework
Presenter: Larry O. Harris

The Test Automation Framework (TAF) is a government-owned automation framework designed to improve the speed, consistency, and reusability of software test and verification activities across diverse development and operational environments. Built as an orchestration layer over open-source automation capabilities, TAF supports a broad range of testing use cases and helps reduce duplicative effort by enabling programs to share common automation methods, tools, and skillsets. The framework is designed for flexible deployment across multiple platforms and environments, allowing organizations to integrate automated testing into the software lifecycle more efficiently. TAF has demonstrated substantial reductions in test execution time across multiple implementation efforts, supporting faster issue identification and improved testing efficiency.

Accelerated Testing Through the NAVWAR Test Automation Framework
Presenter: Bill Huggins

To maintain operational advantage, modern defense systems require rapid, repeatable, and scalable Test & Evaluation (T&E) pipelines. Under NAVWAR 5.0 and NIWC Pacific, our team led the operational rollout of the NAVWAR Test and Evaluation Framework v2.0, pairing architectural modernization with an intensive workforce upskilling initiative. This presentation details the methodology, technical implementation, and cultural transformation that delivered an 81% to 99% reduction in test execution durations across key baseline configurations.

Upgrading Existing Test Frameworks through AI-Assisted Development
Presenter: Richard J. Harris

This test automation framework utilizes a sophisticated AI-assisted workflow that significantly accelerates both the creation of new test cases and the analysis of test failures while maintaining strict adherence to a custom Java, Selenium, and Cucumber-based architecture.

Test case generation is performed by an autonomous SDET agent integrated with Continue in VS Code, connecting to a Gemini API endpoint. The agent receives requirement tags, contextual documentation, and live page information, then generates complete, framework-compliant Gherkin scenarios. These scenarios are built exclusively using a standardized library of universal “super steps” for clicking, data entry, verification, database operations, and navigation. This ensures consistency, maintainability, and eliminates the use of raw Selenium code.

Upon execution, any test failures initiate an intelligent self-healing loop. The agent receives the precise failed step, full stack trace, and a captured DOM snapshot. It performs root cause analysis — identifying issues such as fragile locators, timing problems with dynamic content, missing test data, or actual application defects — and produces a corrected Gherkin scenario. Complementing this, an AI-powered defect analysis service provides structured insights including failure categorization, confidence scoring, requirement traceability assessment, and recommended remediation steps.

Future planned enhancements include the implementation of a human-in-the-loop test generation and recording interface. This capability will allow testers to interactively record actions in the browser, with AI automatically translating the recorded interactions into clean, framework-compliant Gherkin scenarios in real time. Human reviewers will retain the ability to refine, annotate, or approve the generated steps, creating a powerful hybrid authoring experience that combines the speed of AI generation with the contextual knowledge of human testers.

This AI-augmented approach dramatically reduces manual test authoring effort, improves defect diagnosis speed and accuracy, and establishes a scalable foundation for maintaining high-quality automated test coverage.

Track 3 – Instrumentation

Cat A Ground test to Cat A Mishap: Lessons Learned from a Unanticipated Failure
Presenter: Dr. Donald “Bucket” Costello

On 16 November 2020, an MQ-8C Firescout was preparing for a form/fit/function ground test of a new battery. The ground test was considered low risk and therefore did not require extensive risk mitigation steps. The test points were simple: start the air vehicle, do a systems check, and shut down the air vehicle. On the morning of the test event, the MQ-8C was chained to the ground (as it was a maintenance ground turn), and a maintenance laptop was connected to interface with the air vehicle systems. A command and control communication link was not established with a ground station. The test team started the air system, and everything functioned as expected until test personnel initiated shutdown. Within 30 seconds, the air vehicle went to full power, the rotor disk shifted from parallel to the ground to full forward and full right, the shifted disk cut the only connection to the maintenance laptop, and the system ran for approximately 6 minutes until the transmission came apart causing over $2,000,000 of damage. This paper summarizes the facts surrounding the mishap and offers conclusions and recommendations on how to avoid similar mishaps in the future.

Addressing challenges in T&E using phased arrays
Presenter: Satya Ponnaluri

In this presentation, we will discuss the various challenges encountered during the use of digital phased array antennas for Test and Evaluation. We will also discuss techniques that address some of the challenges.

Mission Information Distribution and Processing System (MIDAPS)
Presenter: Mark Wigent
No Abstract

Track 4 – Digital Engineering

From Digital Evidence to Fielded Capability: A Physical Evidence Insertion Framework for Faster, More Affordable T&E
Presenter: Greg Brown

Delivering new and updated capabilities at the speed of operational need requires T&E to produce decision-quality evidence earlier, expose integration risks before they reach high-cost test events, and make effective use of prior evidence as systems evolve. Digital twins, HIL-enabled System Integration Laboratories, surrogate platforms, and integrated developmental and operational testing can support this objective. However, their results differ in purpose, configuration, applicability, and degree of physical, system, user, and mission realism. T&E leaders must determine what has already been learned, what remains uncertain, and where additional testing will provide the greatest decision value.

This presentation and associated paper propose the Physical Evidence Insertion Framework, a conceptual decision-support framework that integrates a decision often distributed across T&E planning, systems engineering, VV&A, and test-resource allocation. PEIF is organized around an evaluation question established by the responsible program, developmental-test, or operational-test authority. Using Model-Based Test Engineering, it relates that question to available evidence, relevant uncertainties, candidate physical observations, and the decision being supported. PEIF distinguishes what may need to be physically represented from the integration context in which it should be exercised.Within PEIF, digital engineering models and digital twins identify sensitivities and conditions warranting physical observation. LRU-level HIL examines processor execution, software behavior, and implemented interfaces. Subsystem HIL and HIL-enabled System Integration Laboratories evaluate interactions among hardware, software, buses, networks, platform simulations, and mission environments. Surrogate-platform and developmental testing address integrated performance and platform behavior. Operational testing retains responsibility for evaluating operational effectiveness, suitability, survivability, and mission performance under representative conditions. PEIF does not treat results from these environments as interchangeable. It preserves each result’s purpose, configuration, uncertainty, provenance, and applicability.

Standardized, machine-interpretable interfaces allow test intent, predictions, observations, configurations, and applicability information to move consistently among MBTE models, digital twins, HIL systems, System Integration Laboratories, and analysis tools. This foundation supports reusable test definitions, automated regression testing, and future AI-assisted test selection, anomaly detection, discrepancy analysis, and configuration-impact assessment. AI-generated results remain traceable to source evidence and subject to human review and independent T&E judgment.

The paper demonstrates PEIF through a notional mission-system use case and proposes measures and a staged validation strategy. PEIF is presented as a testable concept, not a proven method. Its intended contribution is to find implementation and integration problems earlier, reduce repeated test-development and evidence-reconstruction effort, identify which prior results remain applicable after system changes, and focus expensive platform and operational test resources on questions requiring their realism. These mechanisms provide a testable approach for accelerating fielding and capability updates while controlling T&E cost and preserving the rigor and independence of operational evaluation.

The Digital Engineering Environment Maturity Model (DEE-MM) A Dual-Axis Framework for Aerospace and Defense
Instructor: Dr. Armond E. Sinclair / Dr. Policarpio Soberanis

Existing maturity models for digital engineering, including the INCOSE Model-Based Capabilities Matrix and the DAF Digital Maturity Assessment, evaluate what organizations do with model-based practices. None evaluate the capability and maturity of the digital engineering environment itself as an engineered system. This paper argues that the environment, not the practice, is the binding constraint on digital engineering effectiveness, and introduces the Digital Engineering Environment Maturity Model (DEE-MM) to address that gap. The DEE-MM defines a five-level, six-domain maturity framework and introduces a dual-axis classification scheme that independently characterizes environments by deployment age (greenfield through legacy modernizing) and functional maturity (ad hoc through optimizing), enabling differentiated improvement strategies for organizationally distinct starting points.A hypothetical comparative case study illustrates how the dual-axis classification produces materially different diagnostic results and investment recommendations for two programs at the same nominal maturity level. The framework is grounded in DoDI 5000.97, CMMI structural conventions, and a cross-industry literature review.

From Digital Silos to Digital Threads: Enabling Interoperable Digital Engineering for Test & Evaluation
Presenter: Aaron Ashley

Digital engineering promises to accelerate the development, integration, test, & delivery of increasingly complex defense systems. Yet many organizations continue to operate across disconnected engineering environments characterized by siloed tools, fragmented data, inconsistent interfaces, and document-centric processes. Models, requirements, simulations, test plans, software, and analysis products may exist across multiple organizations and security boundaries without a reliable mechanism to discover, connect, or establish authority for those artifacts. The result is a fundamental interoperability and knowledge-management problem: Engineers & decision-makers often spend significant time locating information, determining which data are authoritative, manually transferring information between tools, and recreating work that may already exist elsewhere. These challenges weaken end-to-end traceability, increase technical and schedule risk, and limit the ability of Test & Evaluation (T&E) organizations to make timely, data-driven decisions.

This abstract presents an enterprise digital engineering approach for overcoming these barriers by connecting tools, models, data, workflows, and users through an integrated and persistent digital thread. The approach combines Government-Off-the-Shelf (GOTS) & Commercial-Off-the-Shelf (COTS) capabilities within a secure, cloud-native environment while allowing organizations to retain existing tools and authoritative data sources. Rather than requiring a single monolithic toolset, integration and orchestration services coordinate information across the engineering lifecycle, enabling data updates, traceability, discovery, and reuse while preserving appropriate ownership and access controls.

The environment further incorporates customizable user portals, automated DevSecOps pipelines, modeling & simulation resources, collaborative tools, test planning and management capabilities, and AI-assisted knowledge discovery. AI assistants can help users locate and summarize information across the digital thread, analyze engineering and test data, generate initial technical content, and identify relationships among previously disconnected artifacts while operating within established security and authorization boundaries.

For the T&E community, this approach provides a pathway from isolated digital tools toward an interconnected engineering ecosystem in which requirements, models, simulations, test plans, test assets, results, and analysis remain linked throughout the system lifecycle. The resulting digital continuity can reduce manual handoffs and rework, improve configuration and authority awareness, accelerate test planning and analysis, and enable earlier identification of technical risk.The broader lesson is that successful digital engineering transformation depends not simply on introducing new tools, but on connecting existing people, processes, data, and technologies into an interoperable digital ecosystem. Establishing that connective digital fabric enables organizations to move from merely digitizing engineering activities towards achieving true digital engineering at enterprise scale.

PeopleTec is the lead integrator for Fusion Forge. Fusion Forge is built to deliver the DoW’s requirement for a scalable, secure, and fully integrated Digital Engineering Environment, capable of supporting thousands of concurrent users & programs across acquisition and T&E communities. In alignment with DoDI 5000.97 and TRMC’s directive to build an “acquisition-wide digital engineering infrastructure,” Fusion Forge stands as TRMC’s enterprise-grade solution. Key Takeaway: Digital engineering is not achieved simply by digitizing engineering. The value comes from connecting the digital artifacts.

Technical Track Abstracts – Friday, Nov 6

Track 5 – M&S

Implementing Model Validation Levels (MVLs) to Accelerate Aircraft Certification w/ Ansys
Presenters: Matt Adams, Henrique Correa

This work demonstrates the process of taking a model through the Model Validation Level (MVL) framework, developed by Northrop Grumman, for the purpose of assessing what use cases the model is accurate enough for, like replacing particular test points of operational data. The emphasis of this work is on the MVL process and not a particular model or set of data. For demonstration purposes, we used a UAV electric propulsion subsystem model comprising of a battery, electronic speed controller, brushless motor, and fixed-pitch propeller and synthetic data to demonstrate the process. We define the model, its sensitivity and UQ, and propagate the uncertainty through subsytems to the system level. We then show the level of accuracy to referent data given the model analysis and UQ to validate where the model can be used. We show how iterative analysis as more data becomes available, such as test or operational data, can reduce uncertainty and increase the use cases to which the model can be applied to.The MVL process has nine levels with approved uses ranging from initial architecture development, to model-based design, risk reduction, virtual test planning, targeted pre-production testing support, targeted production testing support, flight test support (certification by analysis), and finally, “As-Built” models matching exact tail numbers approved for sustainment of low-rate initial production (LRIP) operational aircraft.

Beyond the Digital Twin: Turning Reality into Test Evidence
Presenters: Yaniv Minkov, Rami Rokach , Einav Kiperman, Yinon Atzmon

As autonomous and AI-enabled systems become increasingly dependent on the environments in which they operate, Test & Evaluation must address a growing challenge: how can digital testing efficiently produce meaningful evidence when the simulated environment does not adequately represent operational reality? Traditional simulation environments are often generic, manually constructed, or difficult to update as real-world conditions change. This presentation explores an approach in which recently captured real-world data is rapidly transformed into simulation-ready digital twins that become reusable test assets. As a practical example, we suggest an automated real-to-simulation pipeline that converts low-altitude & high-resolution imagery into geo-specific semantic 3D environments. Reconstruction is combined with semantic understanding to identify simulation-relevant elements such as terrain, roads, structures and vegetation, producing environments that can be consumed and functionally utilized by multiple simulation engines.

Rapid environment generation changes the scale of digital T&E. Rather than investing heavily in a small number of handcrafted virtual test sites, evaluators can create and maintain a broader portfolio of geographically and operationally diverse environments. This enables autonomous systems to be assessed across a larger variety of terrain, road networks, built environments and environmental configurations taken from real-world, reducing the risk of overfitting evaluation to a limited set of digital proving grounds. The value of rapidly generated digital twins is not only that they reproduce the real world, but that they allow the real world itself to become a scalable, repeatable and measurable test space. The resulting simulation model can also serve as a controlled ground trough for evaluation. Geometry, object classes, locations, environmental state and scenario parameters can be explicitly known and recorded, providing a form of digital ground truth against which perception, navigation, decision-making and system behavior can be measured. Test conditions can then be varied systematically while retaining repeatability, traceability and configuration control.

This approach also creates a foundation for stronger correlation between digital and live T&E. A test scenario can be defined and executed in a digital twin of an actual test site, and then performed under similar or, where practical, identical conditions in the corresponding physical environment. Results from the simulated and live executions can be compared to quantify model fidelity, identify simulation-to-reality gaps and improve confidence in digitally generated evidence.

The longer-term vision is an integrated V&E workflow connecting real-world environments, digital twins, system models, scenarios and test evidence to help forge more capable, robust and trustworthy physical AI systems. Rapidly generated digital twins can expand the test space, focus live-test resources on the most informative cases, and enable comparable evaluation across the physical and digital domains.

A Model-Based Systems Engineering Framework for the Test and Evaluation of Human-Autonomy Teams
Presenter: Dr. Sandy Patel and Amanda Lee

Advances in artificial intelligence (AI), autonomy, and decision-support technologies are transforming the role of machines from passive tools into active mission partners. Future military operations are expected to increasingly rely on Human Autonomous Teams (HATs) in which human operators, AI-enabled systems, autonomous agents, sensors, and command-and-control (C2) processes collaborate to achieve mission objectives. While significant investment has been made in the development of autonomous capabilities, the Department of War (DoW) continues to face challenges in defining, measuring, and evaluating the operational effectiveness of integrated HATs. Traditional test and evaluation (T&E) approaches were largely developed for deterministic systems and typcially assesses human and machine performance independently, making them insufficient for evaluating the emergent behaviors and dynamic interactions that characterize HATs.This paper proposes a Model-Based Systems Engineering (MBSE) framework for the T&E of HATs. The framework applies digital engineering principles to establish traceability between mission objectives, operational concepts, team behaviors, system requirements, test scenarios, measures of performance (MOPs), and measures of effectiveness (MOEs). By representing the relationships among human decision-makers, autonomous agents, mission tasks, operational environments, and supporting systems within a common model-based architecture, the framework enables T&E practitioners to systematically design and execute evaluations that assess team-level outcomes rather than individual component performance.

The proposed approach further extends MBSE concepts into a broader Model-Based Test and Evaluation (MBT&E) methodology. In this paradigm, digital models become the authoritative source for generating test objectives, developing mission scenarios, identifying data collection requirements, assessing coverage, and tracing evaluation results back to operational requirements. Rather than relying solely on document-centric processes, MBT&E provides a mechanism for maintaining consistency across requirements development, system design, test planning, modeling and simulation, live testing, and operational assessment.

As autonomous systems become increasingly integrated into operational missions, the ability to evaluate team performance will become as important as evaluating platform performance. This work proposes a practical framework for leveraging MBSE and MBT&E principles to support the rigorous, repeatable, and operationally relevant assessment of HATs, enabling the T&E community to better evaluate future AI-enabled systems and mission concepts.

Track 6 – Test Planning

Mission Need to Decision Advantage: AI-Enabled Mission Engineering, Digital Engineering, and Test & Evaluation
Presenter: Dr. Suzanne Beers

Department of Defense acquisition programs are challenged by increasing system-of-systems complexity, fragmented engineering and test data, evolving threats, and the need for faster, evidence-based decisions. Traditional digital engineering and test processes often struggle to provide leadership with timely, mission-relevant insight across the capability development lifecycle.

This presentation introduces iRIS Generation 2 (Gen2), an AI-enabled mission engineering, digital engineering, test and evaluation, (ME-DE-T&E) and decision-support tool developed to accelerate capability delivery and improve decision quality. Building upon more than a decade of operational use of the legacy iRIS event-based requirements management and verification platform, Gen2 extends digital engineering foundations with mission-focused analysis, mission-based risk assessment, predictive analytics, and adaptive decision-support workflows.

The presentation describes how iRIS Gen2 integrates Mission Engineering’s mission context, mission threads, mission engineering threads, operational and technical requirements, test evidence, threat information, and resource data within a traceable digital thread. Guarded AI-assisted workflows help users rapidly develop mission-focused analyses, identify capability gaps, evaluate alternatives, and generate data-driven decision products. The tool supports the building of Developmental, Operational, and Integrated Evaluation Frameworks while aligning with current acquisition transformation initiatives and mission engineering guidance.

TRMC, DTE&A, and DOT&E’s IGNITE ’26 motivated a reimagined AI-enabled tech stack and a transition from system-centric digital engineering to mission-centric decision support. Attendees will see how mission decomposition, mission-based (including cyber) risk assessment, predictive resource readiness analytics, and real-time dashboards can be combined to inform acquisition, engineering, and test decisions. The approach enables stakeholders ranging from Program Acquisition Executives to Test Directors to evaluate technical, operational, threat, resource, and programmatic impacts through a common data environment.

The presentation will introduce current Gen2 capabilities and ongoing development efforts focused on enterprise-level mission engineering, risk quantification, design-of-experiments optimization, and multi-level role-based decision-maker support. The results demonstrate a practical pathway to delivering authoritative, mission-relevant knowledge at the speed of need, thus accelerating warfighter capability delivery.

Forging Human Readiness into Operational Advantage: Applying HRLs in Test and Evaluation and Design of Experiments
Presenter: Judi See

Operational advantage is not fully achieved once a system reaches technical maturity. It is achieved when humans can safely and effectively operate, maintain, and fight with the system under the pressures of the real mission. Technology Readiness Levels (TRL) have served the Test and Evaluation (T&E) community well as a common metric for technical maturity, but they do not capture whether a system is actually ready for its human operators. The ANSI/HFES 400-2021 standard closes this gap by instantiating the Human Readiness Level (HRL) scale. This nine-level framework was deliberately structured to mirror the TRL scale so that human systems integration (HSI) progress can be evaluated, tracked, and communicated alongside technical development. This presentation offers a practical framework to apply HRLs during test planning and design of experiments to ensure that human readiness is evaluated early and often, in parallel with technical readiness.

The HRL scale’s three phases (basic research, technology demonstration, and production and deployment) and nine levels provide a comprehensive planning and decision-making framework, from initial concept through retirement and dismantlement. T&E decisions about test resources and design of experiments include determining the appropriate level of prototype fidelity, when users representative of the target population become mandatory (no later than HRL 5), and what evidence satisfies exit criteria before a program can advance to the next readiness level. In a world where the test community is increasingly faced with evaluating autonomous, AI-enabled, and human-machine teaming systems, such decisions are no longer secondary to technical performance testing. They are central to determining whether the system will actually deliver advantage in the field rather than create new operational risks.

A key focus of this talk is the model used in the standard to provide a straightforward, quantifiable indicator of program risk when TRLs and HRLs become misaligned. Program risk escalates sharply if human readiness lags technical readiness, especially beyond TRL 6, when designs are effectively frozen. Expensive and time-consuming rework may be necessary to improve the system’s human readiness and mitigate human-performance failures after fielding. Embedding HRL assessments into T&E strategy documents and test plans provides an early warning system that complements existing digital engineering and modeling approaches, allowing human factors risk to be surfaced and resourced alongside cyber, software, and hardware risk rather than discovered after fielding.

This presentation directly supports the cross-disciplinary systems-thinking approach highlighted in ITEA’s 2026 FORGE event. Namely, the HRL scale plugs seamlessly into existing systems engineering, HSI, and T&E processes to augment rather than replace, them. Test organizations can integrate HRL milestones into test and evaluation master plans (TEMP) and test schedules, select appropriate metrics and fidelity levels for human-in-the-loop test events, and use HRL/TRL alignment to form a shared vocabulary across stakeholders representing different disciplines (systems engineers, testers, program managers, and human factors experts). The benefit is faster, better-informed go/no-go decisions. Examples drawn from the HRL standard’s acquisition pathway mappings will illustrate application across traditional and adaptive frameworks, giving T&E practitioners a concrete tool for forging human readiness into fielded operational advantage.

Modernizing GEOINT Conformance Testing: JITC’s Data-Driven Approach to Continuous Test and Evaluation
Presenter: Russell Balderston

The Joint Interoperability Test Command (JITC) is advancing the future of test and evaluation (T&E) through the Joint BIIF Profile (JBP) Conformance Analysis Tool (JCAT), a next-generation capability for validating National Imagery Transmission Format (NITF) standards and Geospatial Intelligence (GEOINT) data products. As sensor technologies, imagery profiles, and interoperability requirements evolve, traditional conformance testing approaches can become manual, late in the lifecycle, and difficult to update at operational tempo.

This presentation examines JCAT as a practical T&E modernization case study. JCAT moves conformance rules from hard-coded application logic to editable, version-controlled data, enabling test organizations to update profiles, implement evolving standards, and integrate automated checks into development and test environments without rebuilding the underlying tool. Its application programming interface and command-line interfaces support repeatable, transparent, and automatable assessment workflows that can help shift conformance testing from a late-stage gate to a continuous T&E activity.

The session will also address the JITC-led delivery approach enabled through the Test Resources Management Center Quick Reaction Test Capability program. Artificial intelligence is discussed as an enabling engineering capability—not the focus of the presentation—within a rigorous, specification-driven process governed by domain-expert oversight. Attendees will gain transferable insights into data-driven test architectures, rapid incorporation of changing standards, user-centered test-tool modernization, and risk-managed approaches for accelerating delivery of mission-critical T&E capabilities.

Track 7 – Testing AI

Agentic AI in Cyber T&E: Closing the Machine-Tempo Survivability Gap
Presenter: Captain Sean Authelet, USMC

The Test and Evaluation (T&E) community operates under a foundational trust with the warfighter to deliver honest, mission-relevant assessments of system survivability against realistic adversary capabilities. That trust is increasingly at risk. State and non-state adversaries are already employing agentic Artificial Intelligence (AI) as a force-multiplying cyberweapon—collapsing decision-to-action timelines below human cognitive tempo and enabling adaptive, large-scale exploitation that current T&E practices are not designed to replicate [1]–[3].Agentic AI represents a qualitative shift in the cyber threat environment. Unlike traditional automation or decision-support tools, agentic systems autonomously perceive system state, reason over complex attack paths, and execute multi-step operations without continuous human control. As a result, cyber survivability assessments that rely primarily on static analysis or manually paced Red Team operations risk under-testing systems intended for contested operational environments. This paper proposes a framework for integrating agentic AI into the cyber T&E lifecycle to close this emerging capability gap. Two Concepts of Employment (CONEMP) are presented. The first applies agentic AI to Adversarial Cybersecurity Developmental Test and Evaluation (ACD), enabling scalable, high-fidelity vulnerability discovery and exploit validation earlier in the acquisition lifecycle. The second employs a human-AI teaming model during operational testing, where agentic AI functions as a force multiplier for accredited Red Teams, increasing assessment depth, speed, and coverage without proportional increases in manpower or cost. By adopting agentic AI as a test instrument rather thana productivity aid, the T&E enterprise can raise assessment fidelity, reduce late-stage acquisition risk, and preserve the credibility of cyber survivability evaluations against modern, AI-enabled adversaries.

Testing AI-Enabled Systems: Lessons Learned from Cognitive Electronic Warfare
Instructor: Dr. Sandeep Patel, KBR

Artificial intelligence (AI) is rapidly moving from research environments into operational military systems, including command and control, intelligence analysis, cyber operations, autonomous platforms, human-machine teaming, and electronic warfare (EW). These capabilities offer the potential to increase speed, adaptability, and mission effectiveness, but they also expose a growing challenge for the test and evaluation (T&E) community: traditional assumptions about deterministic behavior, repeatable conditions, bounded environments, and fixed performance measures do not easily apply to adaptive systems whose behavior may change with data, context, and mission interaction. Cognitive Electronic Warfare (CEW) provides a compelling and operationally relevant case study for this problem. In contested electromagnetic environments, CEW systems must sense, interpret, learn, and respond while balancing mission performance, operational trust, and human oversight. As a result, CEW highlights many of the same questions that will shape the future of AI-enabled operational test across autonomy, cyber, decision support, and human-autonomy teaming.

This presentation draws on insights from the Test and Evaluation Threat Resource Activity (TETRA) CEW Research Compendium and related technology roadmapping activities supporting the Department of War (DoW) T&E community. Across CEW research efforts, workshops, technology assessments, and operational test discussions, recurring gaps emerged: representative data generation, threat modeling and simulation, verification and validation (V&V) of learning-enabled systems, explainability and operational trust, assessment of adaptive behavior, real-time test analytics, and realistic environments for human-system and system-threat interactions. While improved algorithms are important, they are not sufficient for rigorous operational assessment. The T&E enterprise will need threat digital twins, representative data ecosystems, live-virtual-constructive (LVC) environments, AI V&V methods, digital engineering integration, and credible approaches for measuring trust, risk, and mission impact in adaptive systems. Although grounded in CEW, the lessons extend to the broader challenge of testing AI-enabled systems before they are fielded in operationally consequential environments. Rather than presenting a finished answer, this presentation identifies the critical gaps, investment priorities, and collaboration opportunities needed to move AI-enabled capabilities from promising demonstrations to trusted, testable, and mission-relevant operational systems.

Testing at the Speed of Change: Forging Operational Advantage through Continuous T&E
Presenter: John E. Robert, Douglas C. Schmidt, Nickolas Guertin

Software-defined and AI-enabled systems are changing not only how military capabilities are developed and deployed, but how they must be tested and evaluated. Capabilities can increasingly be reprogrammed, reconfigured, and enhanced through software updates, new AI models, changing data, and modular components. Operational capability may therefore change in hours or days rather than over traditional development and certification cycles. The resulting challenge is fundamental: when systems can change faster than traditional T&E processes can assess them, T&E itself can become a constraint on delivering operational advantage. Meeting this challenge requires moving beyond episodic evaluation of relatively static products toward continuous, adaptive, risk-informed T&E that produces trusted evidence at the speed of operational decision-making. Modern capabilities increasingly operate within interconnected systems-of-systems comprising independently evolving sensors, software services, autonomous platforms, networks, AI-enabled decision systems, human operators, and physical components. These elements form dynamic mission threads and kill chains whose configurations and behaviors may change continuously. Exhaustively testing every combination is impossible. T&E must instead determine which changes and interactions create the greatest mission risk and continuously evaluate interoperability, resilience, cybersecurity, AI behavior, emergent effects, and mission-level consequences under operationally relevant conditions.

This presentation proposes a future T&E model for forging operational advantage at the speed of system change. The model combines continuous testing embedded within software and AI delivery pipelines; digital engineering and high-fidelity digital twins for rapid experimentation; AI-assisted test design, execution, and analysis; formal methods for verifying critical properties; and independent, risk-informed evaluation focused on the operational consequences of change. AI thus becomes both an object of evaluation and a means of accelerating T&E itself. Together, these capabilities transform T&E from a downstream gate into a continuous source of trusted evidence connecting development, experimentation, and operational evaluation. The result is a T&E enterprise capable of accelerating innovation without sacrificing effectiveness, safety, security, or resilience—and of turning technological speed into decision and operational advantage.

Track 8 – Applied Test

A Synchronized, Spatially Co-Registered Multispectral Scene Projector for Multiband Seeker Hardware-in-the-Loop Testing: A Single-Panel Quantum-Dot-on-MicroLED Approach
Presenter: Yajie Dong and Manuel A. Triana

Modern guided-weapon seekers increasingly fuse information across several spectral bands—ultraviolet, visible, shortwave infrared, and midwave infrared—to discriminate targets, reject countermeasures, and operate in adverse conditions. Hardware-in-the-loop (HWIL) test capability has not kept pace. Today’s infrared scene projectors are largely single- or dual-band, and multiband scenes are produced either sequentially or by combining separate emitter arrays through beam-combining optics; both approaches introduce temporal and spatial registration errors between bands. Because band-fusion and counter-countermeasure algorithms depend on the temporal and spatial correlation between bands, a projector that cannot present co-registered multiband imagery cannot faithfully exercise the very algorithms that make a multiband seeker valuable. The result is a concrete and growing gap in seeker test and evaluation.

This paper presents a synchronized, spatially co-registered multispectral scene projector that emits all bands simultaneously from a single panel. Every band originates from one ultraviolet microLED pump exciting a cascade of quantum-dot converters on the same panel, so the outputs are inherently time-synchronized and co-registered—co-registration is a property of the pixel physics rather than something reconstructed downstream with optics. The architecture is solid-state and room-temperature, requires no cryogenic cooling, and is built on commercial microLED-display manufacturing and inkjet quantum-dot patterning. This enables a near-term retrofit prototype and a path toward frame rates of one kilohertz or greater and high apparent-temperature operation across four to five bands from the ultraviolet through the midwave infrared.

We describe the projector concept and its modeled radiometric performance, and we focus on the attributes that matter to the HWIL test community: inherent per-pixel co-registration with near zero inter-band timing jitter; spectral adaptability by changing patterned quantum-dot inks to match a given seeker’s bands without redesigning the panel; and drop-in integration into existing HWIL architectures without modifying scene-generation software. Two delivery configurations are addressed—single-aperture co-axial projection through a reflective collimator for a multiband seeker, and dichroic-split synchronized paths for separate-aperture seekers—so the same panel serves both current and future test articles. We also outline a staged development path in which a retrofit of a commercial ultraviolet-microLED quantum-dot display de-risks the architecture before scaling to a custom high-frame-rate, high-resolution driver. The approach directly targets the multiband seeker HWIL gap and offers the test community a compact, room-temperature route to co-registered, spectrally adaptable scene projection.

Modernizing End-to-End Weapon System Evaluation Through a Distributed Live-Virtual-Constructive Test Environment
Presenter: William A Reed

The 96th Test Wing (96TW) has begun a major range investment at Eglin Air Force Base to modernize how advanced weapon systems are developed, tested, and fielded. Current weapons test focus on isolated components of the entire Kill Chain and do not provide an absolute test of the Long-Range Kill Chain (LRKC). In addition, traditional Major Defense Acquisition Program (MDAP) paradigms have historically relied on siloed, sequential developmental testing (DT) followed by operational testing (OT). This legacy approach often uncovers critical integration failures too late in the lifecycle, driving up costs and delaying operational deployment. In response to Department of Defense initiatives focused on rapid capability delivery, developmental and operational tests are increasingly being combined within Integrated Test and Evaluation (IT&E) frameworks. Furthermore, as modern systems become highly interconnected across sensors, networks, and kinetic platforms, testers urgently require operationally representative simulation environments capable of evaluating entire mission kill chains prior to costly open-air test events. The 96 TSSQ is developing and executing a portfolio to provide the Long-Range Weapon Kill Chain (LRWKC) full end-to-end test capability. To address this strategic shift, 96TW is developing a distributed Live-Virtual-Constructive (LVC) ecosystem. This ecosystem is designed to connect hardware-in-the-loop laboratories, software simulations, command-and-control systems, high-fidelity sensor representations, range instrumentation, and operational assets into a synchronized, synthetic battlespace. This proposed architecture is intended to execute complex scenarios repeatedly under controlled conditions. This enables the early evaluation of interoperability and performance essential for merged DT/OT, while reducing reliance on expensive live-fire events and increasing statistical confidence in Measures of Performance (MoP) and Measures of Effectiveness (MoE).

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited 96TW2026-0040 Integrating geographically distributed facilities requires precise timing synchronization, low-latency communications, and coordinated simulation execution. The environment supports live-to-virtual interactions through an over-the-air test datalink architecture, which aims to allow real-world weapon system elements to interact dynamically with synthetic sensors, C2 nodes, and mission-level effects.

The virtual-domain modernization effort incorporates several key enabling programs currently under development: • The core engine: Designed to provide the simulation environment, OTA test datalink, and the master sim controller to synchronize disparate simulations to a common, overarching timeline. • Compute, Storage, & Security: Formulated to manage centralized data storage, data services, and security, utilizing logical isolation to ensure secure, digital partitioning of distinct program data within a single-classification, multi-tenant infrastructure. • Network Modernization: Structured to establish the persistent, high-throughput network infrastructure. • Nextgen motion flight table: Planned to integrate physical hardware with highfidelity motion and simulation resources. This paper presents the architectural design, developmental approach, and technical challenges of establishing this distributed LVC environment. We highlight synchronization, orchestration, facility integration, and the projected role of virtualized operational testing in accelerating acquisition. The resulting framework provides a developmental pathway toward more efficient, operationally representative testing, supports the Department of the Air Force’s and Joint Force broader objectives for digital transformation, merged DT/OT, and rapid capability delivery.

Adapting OCUDU For Test Range Operations
Presenter: Dr. Achilles Kogiantis, Joshua Morman

Supporting airborne and high-speed unmanned systems in test ranges requires wireless networks engineered for extreme distance, velocity, and mobility—challenges that extend well beyond conventional 5G deployments.

This presentation outlines the capabilities needed to adapt O-CUDU architectures for 5G and FutureG test range operations. We will explore how to enable very large airspace cells and support high-mobility UEs by maximizing existing 3GPP mechanisms, including extended Cyclic Prefix, RACH formats, and extended window searches. The discussion will also introduce potential beyond-standard enhancements. Specifically, we will focus on addressing gNB-side Doppler compensation and deploying advanced mobility management. By building upon standardized techniques like conditional handover, such enhanced mechanisms can improve link reliability and seamless transitions in highly dynamic scenarios.
Additionally, the talk will highlight the structural overlap with Non-Terrestrial Network (NTN) standards, demonstrating how NTN-driven advances can secure robust connectivity for dynamic range users.