Fuzzy Gain-Scheduling Based Fault Tolerant Visual Servo Control of Quadrotors

Abstract

When military and civil missions such as transportation increase, fault tolerant control of unmanned aerial vehicles will be an obligation. Although onboard sensors provide information about the status of a quadrotor, the camera is not included in the list. In this study, visual servo control of quadrotors as a popular method for motion control is addressed. we address a visual servo control system for quadrotors as a popular method for motion control. The feature motions in the image plane are analyzed to reveal the relation between the actuator faults and these motions. Four AI fault approximators, a neural network, an extreme learning machine, a linear support vector machine, and a long short-term memory are used to approximate actuator faults of a quadrotor while using feature inputs. The results are convincing and the approximation results are used by a fuzzy logic unit to provide gain-scheduling based fault tolerant control. The proposed system shows sufficient results as a visual servo system for fixed and moving feature targets while providing fault tolerance.