Unmanned aerial vehicles are now part of the standard repertoire of today's armed forces. Although there is also a need for land-based systems, unmanned ground vehicles are not yet used in rough terrain without appropriate traffic infrastructure due to the not yet established technological maturity. Only early technology demonstrators are used for collecting first operational experiences. In the commercial (civil) sector, technical development is well advanced and virtual testing of highly automated vehicles is already taking place. In the military sector, the state of the art still comprises manned or teleoperated vehicle guidance. The special challenges for the commissioning of automated, military vehicles in the field include robust obstacle detection, intelligent path planning and safe maneuver execution. In order to develop these functions efficiently and to test them in a variety of scenarios (e.g. in different weather conditions, with obstacles or vegetation), virtual validation methods are required. 

The main objective of the project VIVALDI is to provide an operational and fully functional simulation solution for the virtual validation of automated vehicle controls for off-road driving, whereas a vehicle model, a sensor model, a terrain model as well as the autonomous algorithm serve as input.

This solution will be demonstrated using prototypical models and concrete application scenarios and can be used or being adjusted to different vehicle, sensor and terrain models by BMLV or its technological partners after the end of the project. The main objective will be achieved through the efficient further development of the results and methods of the FORTE project MOSKITO. Based on relevant preliminary work and a state-of-the-art analysis, the project addresses the following challenges and research questions:

1. Which methods and tools must be used or further developed to validate and optimize autonomous vehicle control in the field by means of simulation?
2. Which existing methods and tools can be further developed in order to keep the required development effort low?
3. How can a computationally efficient "closed-loop" simulation solution be implemented to process up to eight high-resolution, virtual camera streams simultaneously for vehicle control?
4. Which KPIs (Key Performance Indicators) are most suitable for virtual validation and quality testing of autonomous vehicle control models in the field?

The result of the project is a modular simulation solution to be used by the BMLV for research, development and testing. The flexible design of the simulation enables the integration of different vehicle, sensor, control, and environment models. The provided platform can be adapted and configured as needed.