Project

Sunflower Tracking System

Real-time rocket tracking with YOLOv8 object detection and dual-axis PID control.

Role

Computer vision and controls

Stack

YOLOv8, OpenCV, Python, PID control

The Problem

Amateur rockets move fast. Manually tracking one through a camera viewfinder is basically impossible once it clears the rail. We needed footage good enough for post-flight analysis, not just a blurry dot disappearing into the sky.

What I built

Sunflower is a camera tracker that locks onto a rocket and follows it autonomously. It runs YOLOv8 for detection, feeds bounding box positions into dual-axis PID loops, and drives a custom stepper-motor gimbal to keep the rocket centered in frame. Won 2nd place at Launch Canada.

Highlights

Implemented YOLOv8 detection and tracking on live video feeds.
Tuned dual-axis PID loops for smooth and responsive control.
Integrated mechanical gimbal assemblies with software control.

Back to work

Media