The consortium project aimed to create a platform‑independent assistance system for long‑range flight platforms that performs real‑time video image analysis. The core of the system is a VPX‑based hardware platform that hosts a video processor and a convolutional neural network (CNN) module. The video processor is built around an FPGA that implements innovative video‑compression techniques with a Region of Interest (ROI) capability, allowing high‑resolution transmission of multiple selectable image patches while keeping latency low. The CNN module, supplied by Rockwell Collins Deutschland GmbH, performs object detection in real time, enabling situational awareness for operators on the ground.

Elma Electronic GmbH was responsible for the design, fabrication and integration of the VPX hardware. The hardware package includes an aluminium chassis optimized for strength, weight and IP protection; a custom VPX backplane that distributes high‑bandwidth signals with preserved integrity; an I/O board that provides signal and power interfaces with required filtering to meet electromagnetic compatibility (EMC) standards; and a power supply unit tailored to the specific power envelope of the video‑processor payload. The design process began with a detailed requirements specification (Pflichtenheft) that guided the development of each subsystem. Elma produced a functional laboratory model that integrated the VPX backplane, I/O board, chassis and power supply, and performed extensive verification tests in a system integration laboratory (SIL). A flight‑ready model was then assembled, incorporating the final board assemblies and the video‑processor module from Collins. This flight model was qualified through experimental flights with the LUNA NG drone, confirming the system’s operational capability under realistic conditions.

Performance results from the laboratory and flight tests demonstrate that the VPX system meets the stringent size, weight and power (SWaP) criteria required for airborne deployment. The ROI feature allows selective high‑resolution data streams to be transmitted without saturating the available bandwidth, while the FPGA‑based compression achieves low latency suitable for real‑time decision making. The integrated CNN module successfully identifies objects in the video stream, providing actionable information to the ground operator. Although specific numerical throughput or latency figures are not disclosed in the report, the successful completion of flight tests indicates that the system satisfies the performance targets set by the consortium.

The project was organized into eight work packages covering requirements definition, backplane development, I/O board design, mechanical design, power supply development, hardware construction, component verification, and system verification. Elma’s contributions spanned the creation of the VPX hardware specification, the design and integration of the backplane and I/O board, and the verification of the complete system. Rockwell Collins supplied the video‑processing module and the CNN hardware, while the Fraunhofer Institute, Eberhard‑Karl‑Universität Tübingen and EMT (now Rheinmetall Technical Publications) provided complementary expertise in signal processing, machine learning and system integration. The consortium leader was Rockwell Collins Deutschland GmbH.

The project faced delays due to the COVID‑19 pandemic and global supply‑chain disruptions that affected the procurement of electronic components. Despite these challenges, the consortium completed the development and flight qualification of the VPX system within the allocated timeframe. The project was funded under the European Union’s Horizon 2020 programme, supporting the development of advanced airborne video‑processing and AI capabilities for long‑range flight platforms. The final outcome is a fully integrated, platform‑independent VPX system that can be embedded into existing drones, enabling real‑time video analysis, ROI‑based data transmission and distributed neural‑network‑driven object detection.