The project proposes the development of a hybrid, machine-vision-supported navigation  system for UAVs that ensures safe and precise operation in GNSS-jamming and spoofing  environments, with an initial focus on large-scale solar park inspection and linear  infrastructure surveying. Building on PX4/ArduPilot and QGroundControl, the system fuses  GNSS, inertial sensors and computer vision (including SLAM-based local mapping and AI based object detection) to monitor GNSS integrity, detect interference, and seamlessly  switch to vision-centric guidance when needed. The solution will provide robust path  following capabilities even in GNSS-denied conditions, while remaining drone-agnostic and  exportable as a modular payload and software stack. The activity will deliver integrated  hardware–software prototypes validated on multiple UAV platforms, together with tools  and documentation that enable immediate exploitation in commercial solar park services  and pave the way for broader adoption in infrastructure, security and defence markets.