Design of control system for UAV-based video recording and tracking

Abstract: The core objective of this project was to design a controller for a drone (or quadcopter) that would allow the drone to keep a target in the center of a video filmed by a camera mounted on the drone. This was done by controlling the pitch of the gimbal, on which the camera was placed, and the yaw of the entire drone combined with using an image recognition program that could identify the target. For this project the DJI Mavic 2 Zoom was used. Although designing a PID controller was an important part of the project it was relatively easy and emphasis was instead put on system identification and minimizing the impact of a large delay. The large delay was present in sending images from the drone to the controller and could be as long as 1 second. A Smith predictor was used in order to reduce the influence of the measurement delay in the control while a Kalman filter was used in order to estimate the position of the target based on the old measurements. The system consisted of a computer connected wirelessly to a mobile phone, and the drone connected wirelessly to the remote controller which in its turn was connected by wire to the mobile phone. The end result was a system that could keep a moving target, with constant velocity, well in the center of the video but suffered from overshoots for a target making quick turns.