Vision based autonomous gap-flying-through using the micro unmanned aerial vehicle

A vision-based enhanced PID control algorithm for the gap-flying-through problem using a low-cost multirotor platform is the main topic of this paper. We designed an image processing program to detect the target rectangle gap in each video frame captured by onboard camera of the drone. During the process, some manipulation such as dilate and erode are implemented to reduce the noise. To control the motion of the drone, we applied a PID control algorithm with several enhancement like segmented coefficients, restricted integral saturation and integral separation. The current version of our system is based on an AR. Drone quadrotor. Our program is executed on a laptop, which cooperates with the drone through Wi-Fi connection and Robot Operation System (ROS). Our system have been evaluated with a series of indoor experiments under a relatively harsh circumstances, with success rate of 81.67%, the effectiveness of our system has been demonstrated.