Ukraine's army doesn't have as extensive human resources as Russia, so it is teaching its machines to do more of the work on their own. Ukrainian developers are building autonomy into aerial and ground drones so the systems take over tasks the operator once did, the defense outlet Oboronka reported.
Across reconnaissance, bombing, interception, and ground robots, Ukrainian firms are automating the steps that once tied up a pilot, allowing a single operator to run multiple drones at once.
Automating reconnaissance and mission planning
The heaviest early demand for autonomy is in reconnaissance. Farsight Vision's FSV Platform builds real-time 3D terrain maps from drone imagery for planning assaults and strikes, and models where enemy electronic warfare and radio dead zones lie, so FPV pilots can avoid losing their link.
Meanwhile, Sine Engineering offers "autonomy modules" that let a drone navigate on its own. A Mavic flies to an observation point, works for 30 to 40 minutes, and returns, with the pilot only turning the camera, so one operator can run several drones.
"70 to 80% of the need for autonomy right now is reconnaissance," Sine Engineering CEO Andrii Chulyk said.
Autonomy for bombers and FPVs
Heavy bomber drones waste most of a pilot's time in transit. A bomber's target can sit 35 to 60 kilometers from launch, over 25 minutes each way, and a mass raid needs a pilot per drone, so Lasar's Group uses route-planning software to tie up to five bombers to one pilot, who connects for a few minutes to line up each strike.
A single pilot can carry out bombing runs one after another with a group of bombers, preserving personnel for other missions.
Autonomous FPVs are pricier but far more capable: Chulyk estimated an AI drone that finds and destroys a target in a swarm would cost $2,000 to $3,000, against about $700 for a pilot-flown night FPV with auto-capture.
Sine Engineering argues one automated FPV can replace up to ten simpler drones in hard missions, while The Fourth Law's Maksym Savanevskyi favors adding auto-capture to cheap existing drones, which he says makes a radio-controlled drone two to four times more effective.
"Fire and forget" for interceptors and robots
The hardest problem for interceptor drones is the "last mile", spotting a small target against open sky. The Fourth Law's TFL Anti-Shahed module scans the sky and detects targets at twice the range of the human eye, marking a Shahed so the pilot knows where to fly.
Savanevskyi predicts Ukrainian makers will field automatic Shahed-killing solutions by year's end. That points toward "fire and forget" interception, said Yevhen Lesin of the 412th Nemesis unmanned-systems brigade, though he warned against autonomy for its own sake, noting automated launch from ground robots creates technical problems and needs more people, not fewer.
Ground-robot autopilots remain among the field's hardest tasks: DevDroid's Oleh Fedoryshyn noted it took Tesla ten years to make the car autopilot work. At the same time, frontline dirt roads, cratered and unmarked, are nothing like marked highways, so DevDroid's robots now cross only short link-dead stretches on their own before the operator resumes control.
Read also
-
Reports credit North Korea’s newest missile of terror with 2.5-ton warhead and 900-km range. Expert explains why those numbers can’t both be true
-
Zelenskyy awarded Vikings star, ballistic missile CEO took stage, and strike drone got signed with messages to Russians: Ukraine’s major UNITED24 platform turns four
-
Evacuation is “force majeure that no one is ready for” in most Ukrainian units, and robots are too slow to fix it, commander says
