Do you want to get a demo?
Leave a request and our managers will contact you.
By submitting the application, you agree to the personal data processing policy.
EMIVN drones: Interceptors for defense
EMIVN is developing autonomous interceptor drones to protect the airspace. These systems detect, track, and neutralize enemy UAVs in real time by integrating computer vision, radar, and guidance algorithms. Suitable for military, security, and critical infrastructure tasks ranging from local defense to scalable networks.
We adapt technologies to the client: speeds up to 7.5 m/s, multi-drone scenarios, work in noisy conditions. We are ready for field tests on real hardware.
We adapt technologies to the client: speeds up to 7.5 m/s, multi-drone scenarios, work in noisy conditions. We are ready for field tests on real hardware.
Key technologies
The systems are based on ROS 2, PX4 Autopilot, and the Gazebo simulator. Hardware base: Raspberry Pi 5, Hailo-8L accelerator, CSI cameras (130-160° overview), compatible flight controllers.
Computer vision uses YOLOv8m (mAP@0.5 : 0.76 at 1024px) for detecting intruder drones from different angles. The OpenCV Nano tracker provides stable tracking, while the IBVS algorithm provides autonomous approach while keeping the target in the center of the frame.
Multisensory fusion: visual data + radar (azimuth, range, noise/delays). The trajectory prediction model filters out noise and compensates for delays for forward guidance.
Computer vision uses YOLOv8m (mAP@0.5 : 0.76 at 1024px) for detecting intruder drones from different angles. The OpenCV Nano tracker provides stable tracking, while the IBVS algorithm provides autonomous approach while keeping the target in the center of the frame.
Multisensory fusion: visual data + radar (azimuth, range, noise/delays). The trajectory prediction model filters out noise and compensates for delays for forward guidance.
Development achievements
From November 2025, a full simulation environment with multi-vehicle scenarios has been deployed: interceptor vs. the intruder follows a circular path. Datasets of 3000+ images have been collected, marked up in CVAT, synthesized with augmentations.
Integrated ROS radar node: TCP connection, conversion of polar coordinates to NED, error covariance matrix, auto-reconnect. GUI on PySide6: video stream, manual target capture, flight parameters.
The tests confirmed: successful interceptions, recovery after a miss/loss of contact, resistance to noise. The actual drone assembly is complete, the camera is calibrated to 30 FPS.
Integrated ROS radar node: TCP connection, conversion of polar coordinates to NED, error covariance matrix, auto-reconnect. GUI on PySide6: video stream, manual target capture, flight parameters.
The tests confirmed: successful interceptions, recovery after a miss/loss of contact, resistance to noise. The actual drone assembly is complete, the camera is calibrated to 30 FPS.
Advantages for defense:
Reliability: Detection + tracking + real-time prediction, even with radar delays.
Autonomy: Re-aiming, final maneuver (ramming).
Scale: Multi-neural simulations, integration with networks.
Fast adaptation: The client's tasks include intelligence, interception, and patrol.
EMIVN provides a full cycle, from simulation to field tests. Invest in the defense of the future through tokens.
Autonomy: Re-aiming, final maneuver (ramming).
Scale: Multi-neural simulations, integration with networks.
Fast adaptation: The client's tasks include intelligence, interception, and patrol.
EMIVN provides a full cycle, from simulation to field tests. Invest in the defense of the future through tokens.
Directions
Contacts
© 2026 EMIVN Corporation - All rights reserved.
Legal information
Languages
