KAPGAN Combat UAV

National champion and winner of the "Most Original Software" award at Teknofest 2021 Combat UAV Competition — I was the sole developer of all three software units.

2020 – 2021 Tek yazılım sorumlusu (3 birim)

Video

Overview

KAPGAN is the autonomous combat UAV we built with Team ÖZTÜRK for the Teknofest 2021 Combat UAV Competition. I single-handedly coded all three software units — the AI unit (the "eyes"), the main management unit (the "brain") and the ground control station (the "operator").

The Problem

The combat mission requires recognising a rival UAV from the camera, locking onto it and tracking it for a set time — fully autonomously, within seconds, and stably on a windy field. That demanded a robust perception-to-action autonomy chain.

AI Unit

I trained a Faster R-CNN model with Python and OpenCV to recognise drone bodies and detect target UAVs from the camera feed in real time, streaming detections to the main management unit.

Management Unit & Ground Control Station

I wrote the main management unit in Python with multi-threading: it evaluated detections and sent autonomous engagement commands to the flight controller over MAVLink. The ground control station was built in C# / .NET WinForm with a layered architecture and a MaterialSkin UI, including telemetry sharing with the competition server and a module that emulates that server for testing.

Outputs & Results

We won the national championship and the "Most Original Software" award. The system recognises and locks onto a target in 0.1 s and tracks it for 10 s, flies stably in 30-35 km/h winds, stays airborne for 25-35 minutes, has a 40+ km autonomous-control range and streams live video from 5 km.