Protocol for Vision-Based Tracking and Proportional Control in Quadcopter Follow-Me Applications

This paper presents the design and implementation of a real-time Follow Me protocol for a quadcopter using vision-based tracking and proportional control. The system enables a drone to autonomously follow a moving human subject using bounding box detections from an onboard AI-based object detection camera stream. The design extracts the width of the bounding box surrounding the target and uses it as a reference for distance. A proportional control algorithm maps the deviation of the observed width from a predefined ideal width into a corresponding pitch velocity, which is then converted to PWM signals to drive the drone. The control logic ensures that the drone maintains an optimal distance from the subject by dynamically adjusting its forward and backwards movement. Experimental results demonstrate a linear and monotonic relationship between the bounding box width and the drones pitch signal, validating the accuracy and responsiveness of the tracking system. The proposed system operates robustly in real-time and can be integrated into lightweight UAV platforms without requiring GPS or external localisation systems.