Measurement Based Stabilizing PID Controllers for Camera Gimbals

摘要

Stabilizing gimbals are ubiquitous within the unmanned aerial and ground system robotics communities as they permit acquisition of a variety of sensing modalities, typically vision, while removing undesired rapid orientation changes that a vehicle might experience. In this paper, we leverage a commercial gimbal controller to construct a highly reconfigurable rapid prototyped gimbal setup and compute a practical set of stabilizing PID controllers for the gimbal. The novelty in this research is that the method constructively determines the set of stabilizing PID controllers using only direct data measurement without additional knowledge such as the identification of the plant model, order of the system and number of poles and zeros. The set guarantees stability and, thus, is particular to each plant to be controlled. This feature is an attractive complement to a legacy commercial gimbal controller tuning method in which the parameter tuning ranges are common for all possible hardware specifications of the gimbal irrespective of system stability. We compare the set of stabilizing PID controllers, called the stabilizing set, with a sample outcome of the auto-tuning algorithm provided with the commercial gimbal and verify that the stabilizing set contains the auto-tuning result. The geometrical visualization of the stabilizing set enables comparative analysis of the controllers, along with guide possible directions at which retuning should be aimed.