Variant PID controller design for autonomous visual tracking of oil and gas pipelines via an unmanned aerial vehicle

Abstract

This paper presents an improved autonomous navigation control mechanism for a UAV for accurate tracking and inspection of the ground laid linear horizontal structures such as oil and gas pipelines. At first, the Canny Edge Detector (CED) and the Probabilistic Hough Transformation (PHT) are used to identify the structures based on visual data collected by the onboard camera. Then suitable geometrical parameters are extracted from the collected structure information to design controller mechanism of the UAV for autonomous tracking along the identified linear structure. For autonomous tracking, the development of an appropriate controller is significant because it affects the overall response time, sensitivity, accuracy of the tracking performance by the UAV. Through analyzing the previous research results of pipeline recognition and the UAV navigation by velocity mapping [1], it can be seen that the overall lateral correction of the navigation path based on the conventional PID has obvious disadvantages such as a slow response time, UAV oscillation, angular and lateral instability etc. Especially for a low-altitude tracking, the tracking target is prone to be out of camera view very quickly. Thus a new controller with variant coefficients is designed for a better tracking performance. Combined with the previous research achievement, this paper emphasizes the principal and the allocation of the newly designed variant PID controller.