In the military context, autonomously operating ground vehicles (UGV) for reconnaissance, logistics and medical evacuation tasks represent an essential future perspective. The special challenges for the operation of UGVs in terrain include robust obstacle detection, intelligent path planning and navigation using passive sensors only as well as efficient maneuvering in a wide range of military relevant terrains, weather, and visibility conditions. The objective of the project "TERRA-V" is to provide a comprehensive, AI-supported simulation platform for the development and optimization of perception, path planning and control algorithms for autonomously operated military vehicles in military-relevant environments. For this purpose, detailed simulation models for a vehicle system, an environment, various weather and visibility conditions, and a passive sensor system are fused together. In the case of the vehicle model, the geometrical data as well as the details and characteristics of the body, chassis and powertrain are parametrized in detail. In the sensor suite, both the characteristics and the position of the individual sensors (e.g. cameras and IMU) are simulated in detail. The autonomous driving algorithm is implemented in the vehicle. A significant innovation of the project is the use of artificial intelligence to generate critical, militarily relevant testcases for the robustness testing of autonomous systems. This involves combining complex interactions in military operations, including terrain, obstacles, weather, lighting/visibility, and other actors such as vehicles and personnel, to induce errors or undesirable behavior in the UGVs. This method is used to test and optimize so-called "edge cases". Another essential research task is the development of militarily relevant Key Performance Indicators (KPIs). These include quantitative assessment metrics for (1) functional safety to avoid collisions, (2) the quality of effective path planning in terms of travel time and energy consumption, (3) the preservation and safety of personnel, cargo, and materials, and (4) the quality of environmental perception and localization without GNSS. During simulation, a set of KPIs is calculated for each scenario and compared with target values. The algorithm is iteratively optimized until all target values are met in all critical maneuvers. The vehicle, sensor, and terrain models can be exchanged, modified, or expanded without limitations, enabling the TERRA-V simulation platform to be used for all military-relevant scenarios, including cross-country movement. The focus is particularly on operation in GNSS-denied environments using exclusively passive sensor technology. The project culminates in a software package with a user-friendly interface that incorporates the described performance features. System validation is conducted using an autonomous "System under Test" to demonstrate performance.