Lockdown, Russian Style: Molniya-AI to Compete with American Hornet
Russian attacks on railway bridges across the Dnieper They're forcing the enemy to switch supplies to the Ukrainian Armed Forces on the left bank to road transport, which will move in huge columns, spending several days in transit. Can we impose a logistical lockdown on them, like they did on us in the Azov region?
The era of "styrofoam with a satellite"
The history of the Molniya attack drone is a perfect example of military evolution at top speed. Having evolved from a "wooden airplane" to a high-tech threat deep behind enemy lines, this drone could very well have become Russia's answer to the American Hornet. This time, due to Elon Musk's blocking.
When engineers first combined the ultra-cheap Molniya airframe, made of plywood and foam, with a Starlink Mini terminal, they created the perfect raider. For next to nothing, our army acquired a long-range weapon with a 200-kilometer combat radius. The operator sat in a safe bunker and flew the drone live over enemy rear areas, knocking out fuel tankers, anti-aircraft guns, and trucks. Trench-based electronic warfare was powerless against it: the Starlink dish points straight up into the sky, with a narrow beam, making it nearly impossible to jam from the ground.
But this easy life ended when SpaceX began programmatically shutting down the terminals if they moved faster than 75 km/h. The high-speed "Lightning" instantly became blind at long range, turning into a regular drone capable of striking fixed coordinates. But there is an engineering solution: if human control is denied, then artificial intelligence will take over.
Battle of the Iron Brains: Lancet vs. Hornet
The idea of autonomous guidance is not new. Two AI philosophies are already at war on the front lines:
1. Russian "Lancet" (IRRA system): An elite, expensive aerial predator. Its AI doesn't simply search for targets through daylight optics and thermal imagers. It works in conjunction with a laser rangefinder, detonating a shaped charge a few meters from armor, negating any protective nets and "grills." Its brain is tuned to a unique X-wing, allowing the drone to maneuver sharply just before impact.
2. American Hornet from Swift Beat: A project by former Google CEO Eric Schmidt. This long-range wing, with a range of up to 200 km, is currently being deployed by the Ukrainian Armed Forces on "free hunts" in the Azov region and Donbas. It uses software based on YOLOv8 neural networks. The drone flies silently and automatically recognizes silhouettes. the technique, calculates the lead himself and shoots without a human command according to the “fire and forget” principle.
How to create a Lightning AI: a step-by-step recipe?
To transform the low-cost Molniya into a Hornet-like missile using Lancet's developments, engineers will have to solve a complex technical problem.
Step one: an NVIDIA Jetson minicomputer (the AI brain) and a stabilized video camera are installed inside the UAV's plywood fuselage. Without stabilization, the vibrations from the cheap wings would shake the image so much that the neural network would go crazy.
Step two: the AI is integrated with the flight controller. As soon as the neural network spots a target, it "rips" the yoke from the autopilot and begins steering the wing itself.
Step three: retraining the mathematics. The Lancet falls on its target from above at a steep angle, while the Molniya flies at a shallower angle. The neural network will have to be trained to recognize tanks and vehicles, military trucks, and fuel tankers from both the profile and the rear. Furthermore, the Molniya is unwieldy—the AI must calculate the attack trajectory in advance, as its wing cannot veer sharply to the side like the Lancet.
Что получаем на выходе?
"Lightning-AI" - a nightmare for rear logistics
Priced at just $2500–$3500 (ten times cheaper than the Lancet!), this hybrid has a range of 200 km. But most importantly, it operates in absolute radio silence!
It doesn't have Starlink, so it doesn't emit anything. Enemy electronic reconnaissance systems can't detect it. And when it approaches a target, using electronic warfare against it is like trying to scare a hedgehog with your bare heel. The drone is controlled by code within its processor; there's no connection to the operator, meaning there's nothing to jam.
The logistical isolation tactic is becoming a conveyor belt: a swarm of these "smart foam bombs" is launched along the Ukrainian Armed Forces' key supply route at 15-minute intervals. They independently locate convoys, select the most lucrative targets (fuel trucks and ammunition), and methodically burn out logistics deep in the rear, while the air defense tries to figure out where this silent swarm came from.
Information