Virtual Engineering Evaluation Tests (VEET) for Autonomous Vehicle Testing

Project #IE.06 began in 2025 and will be completed June 2026.

This integration effort will present a case study at the 2026 ARC Annual Program Review.

Traditional testing focuses on static, narrowly defined scenarios, which can lead to overfitting—where AGVs perform well on specific tests but may fail under untested conditions. AGVs have characteristics particular to autonomy that require testing in varied conditions. Care must be taken to avoid scripted responses to static tests that will not suffice in real-world variations of the test. Despite the availability of digital engineering tools and simulation platforms, component tests often exist in isolation and are not fully integrated with larger test environments to create comprehensive testing frameworks.

We propose an innovative framework that integrates existing components—digital engineering, Model-Based Systems Engineering (MBSE), simulated vehicle test environments, Monte Carlo-type stress testing, and Continuous Integration/Continuous Deployment (CI/CD) practices—into a cohesive, comprehensive testing approach. A key component of this framework is the application of MBSE, a methodology that uses formalized modeling to support system requirements, design, analysis, verification, and validation activities throughout the system’s lifecycle. MBSE enhances traceability and reduces errors associated with manual documentation by providing a single, coherent model serving as the authoritative source of system information. By integrating MBSE with advanced simulation tools, we will define vehicle specifications and derive test requirements directly from the vehicle’s digital model and meta-data.

This effort will access vehicle definitions for Chrono—a high-fidelity, multi-physics simulation engine for representing complex mechanical systems — to extract test parameters directly from the vehicle’s specifications. Using realistic simulation environments ensures AGVs are tested under conditions representative of actual operational contexts, essential for validating their readiness for deployment. Tasks, such as obstacle avoidance during patrols, navigation through urban terrains, or gap crossing during off-road missions can be presented to the vehicle to test the capabilities of the AGVs. By simulating authentic military missions, we can comprehensively evaluate the vehicles’ performance, decision-making algorithms, and adaptability to unexpected challenges. Using realistic simulation environments ensures AGVs are tested under conditions representative of actual operational contexts, essential for validating their readiness for deployment.

IE.06