russian forces are expanding the types of unmanned aerial vehicles they equip with onboard target acquisition and guidance systems based on machine vision, according to Serhii Flash. What began as relatively common computer vision suites for FPV attack drones is now being adapted for larger, fixed-wing strike platforms such as the Molniya, a worrying step that blends cheap massed strike concepts with autonomy and extended range precision.
Machine vision guidance for FPV drones has been visible since early 2024, when both sides began fielding systems that could lock onto and track targets visually, raising hit probability against small and moving targets. The core advantage is twofold: visual algorithms can keep steering a munition to a visible aimpoint when radio links die, and they can operate around or through short range electronic warfare bubbles that would otherwise blind remote pilots. That combination increases both effectiveness and survivability of low cost kamikaze drones.
Read more: Ukrainian Drones Strike Belgorod Dam, Flood russian Positions and Cut Supply Lines Near Vovchansk
Adapting the same capability to airplane type drones like the Molniya is technically harder but tactically valuable. Fixed-wing strike UAVs fly faster and farther, and adding onboard computer vision allows them to continue terminal guidance after losing uplink at low altitude or beyond line of sight. In practice this means the enemy can attempt long range strikes with higher confidence of impact even in contested EW environments. It also enables strikes against moving or fleeting high value targets.
Reports also point to experiments with fiber optic spool systems and other hardened datalinks for the Molniya variants, a development intended to preserve control in dense EW conditions. Together, vision based autonomy plus resilient connectivity creates a layered solution: maintain remote control where possible, but fall back to onboard visual targeting when links fail, effectively extending both the reach and the reliability of cheap strike UAVs.
There remain open questions about scale and operational prevalence. Fielding advanced vision stacks at scale requires hardware, software, testing, and logistics; mass equipping of entire drone fleets takes time and supply. Public indicators so far suggest pockets of deployment and experimentation rather than a fully generalized capability, but even limited employment can shape tactics and force defenders to adapt rapidly.
Countering this threat requires a multi domain response. Traditional EW must be complemented by optical and multispectral countermeasures: active dazzlers, laser/IR dazzers, rapid deployment smoke and obscurants, and decoy signatures to confuse vision nets. Kinetic options, point air defenses, loitering interceptors, directed energy systems and man portable air defense, remain essential, but defenders must also invest in algorithms that can spoof or jam machine vision systems (adversarial patterns, light modulation) and in sensors that provide fused cues (radar + EO/IR) to improve detection and tracking.
Tactically, ground units and critical sites should adapt operating procedures: reduce predictable visual signatures, employ mobility and concealment, and train personnel to recognize and react to autonomous terminal approaches. At an operational level, procurement priorities should include resilient communications, rapid response air defense layers, and counter autonomy tools. Intelligence collection to map where vision equipped drones are used will be crucial to allocate limited defensive assets efficiently.
Read more: TAC's New "Active" Q-NET System Looks Familiar and Not Exactly Unique
