UAV & Autonomous Systems · 2026
Autonomous Target Lock with YOLOv8n
YOLOv8n-based computer vision that detects rival UAVs from the onboard camera in real time, plus Python software that tracks the target and performs the lock-on.
What I did
- Used a YOLOv8n-based model so our UAV could recognize rival UAVs from the camera feed.
- The model was optimized with TensorRT to run in real time on an NVIDIA Jetson.
- Wrote the lock-on algorithm in Python that keeps tracking the detected target inside the lock area.
- A lock-on to another UAV was achieved in flight testing and recorded on video.
Tech stack
- Python
- YOLOv8
- OpenCV
- TensorRT
- NVIDIA Jetson
Related projects
UAV & Autonomous Systems · 2026
Ground Control Station & Competition Server Link
A ground control station that works alongside QGroundControl, plus a MAVLink bridge that streams our UAV telemetry to the competition server. Used without issues at the TEKNOFEST 2026 Fighter UAV final.
Details →UAV & Autonomous Systems · 2026
In-Flight Maneuver with Offboard Mode
Control software that takes a UAV flying a mission into offboard mode mid-flight via Python, steers it, and hands it back to the planned mission when stopped.
Details →UAV & Autonomous Systems · 2026
No-Fly Zone Avoidance with A* and Pure Pursuit
A hybrid guidance architecture that combines A* path planning with Pure Pursuit path tracking to avoid no-fly zones received from the competition server.
Details →