Dynamic Image-Based Output Feedback Control for Visual Servoing of Multirotors
Xuetao Zhang,Yongchun Fang,Xuebo Zhang,Jingqi Jiang,Xiang Chen
DOI: https://doi.org/10.1109/tii.2020.2974485
IF: 12.3
2020-12-01
IEEE Transactions on Industrial Informatics
Abstract:This article proposes a novel adaptive image-based output feedback visual servoing approach to control a multirotor to the desired pose by using a minimum onboard sensor suite, which consists of an inertial measurement unit and a monocular camera. Different from "perspective moment," a new type of image feature is designed as "rotated perspective moment," whose dynamics is independent of roll, pitch, and yaw rates. On this basis, a nonlinear adaptive observer is designed to estimate the scaled linear velocity, which is more accurate, since the observer does not involve noisy angular velocity measurements. Then, a novel image-based output feedback controller is proposed with the designed image features and the observer, wherein the new saturated integral terms of linear and angular velocity errors are introduced into the controller design, respectively, to compensate system uncertainties. As a result, the steady-state error is decreased considerably. In addition, without the assumption of the separation principle between the observer and the controller, the small-angle approximation, or the time-scale separation assumption, the error signals of image features, attitude, velocity, and observer estimation can all converge to the origin asymptotically, which is proven by rigorous Lyapunov analysis. Comparative experiments are conducted to show the superior performance of the proposed approach in terms of more accurate velocity estimation, smaller steady-state errors, and stronger robustness.
automation & control systems,computer science, interdisciplinary applications,engineering, industrial