



01 / 04
SENSOR FUSION
Five-step autonomous pipeline from first sensor contact to mission reset. Each stage executes in seconds. The operator retains override authority throughout.





Detection
Dual-sensor scan - optical and thermal cameras sweep the airspace. The AI identifies aerial objects against sky, terrain, and clutter. Motor heat signatures light up against cold sky in any visibility condition.
DOZOR AI is VARTA's autonomous interception module. It combines dual-sensor input — high-resolution optical and thermal cameras — with onboard neural-network classification, inertial navigation for GPS-denied operation, and direct MAVLink integration with ArduPilot and PX4 autopilots.
Once deployed on an interceptor, DOZOR scans the airspace continuously. When it detects an aerial object, the classifier distinguishes threat type — enemy FPV, reconnaissance copter, loitering munition, fixed-wing UAV, or friendly platform — before any engagement decision is made. In assisted mode, DOZOR presents the target and requests approval. In fully autonomous mode, it executes the full kill chain: detect, classify, track, engage, and reset. Multiple intercepts per sortie are possible without returning to base. The operator retains override authority at every stage.
Key applications include point defense of mobile formations and fixed assets, airspace patrol in contested zones, and saturation defense against FPV kamikaze swarms. Because DOZOR is drone-agnostic and integrates through SYLA's 4-channel firing control, it plugs into existing Ukrainian FPV platforms rather than requiring purpose-built airframes.