Autonomous Person-Tracking Camera System

This project is an in-progress autonomous camera system designed to detect and track a person in real time. The system combines a Raspberry Pi, YOLO-based person detection, live video streaming, remote web controls, and motor-driven pan-tilt movement to keep a target centered in the camera frame.

Fixed cameras require manual adjustment and cannot actively follow a moving person. The goal of this project is to build a camera platform that can detect a person, estimate their position in the frame, and automatically aim toward them using a motor-driven pan-tilt mechanism.

I completed the computer vision and motor control portions of the project, including YOLO person detection, target center tracking, live video streaming, remote web interface control, and pan-tilt movement commands. The remaining work is focused on building the physical enclosure, improving the mechanical aiming mechanism, and packaging the system into a more finished prototype.

I used YOLO for person detection because it provides real-time object detection while remaining practical for a Raspberry Pi-based system. I built the tracking logic around the difference between the detected person’s center point and the camera frame center, then used that error to command pan and tilt corrections. I also added a browser-based interface so the system can be viewed, tested, and controlled remotely during debugging.

The project currently supports person detection, target tracking, live video feedback, remote web viewing, and motor-driven aiming control. The next stage is mechanical: designing and building the enclosure, refining the pan-tilt mechanism, improving wire routing, and making the system more durable and presentable as a complete prototype.

Design the enclosure, build the pan-tilt mounting structure, improve mechanical stiffness, clean up wiring, reduce tracking lag, and tune the control behavior for smoother person-following performance.