Which stability-enhancement methods appear most effective for quadrotor UAVs?

Abstract

The use of quadrotor unmanned aerial vehicles (UAVs) in a wide variety of tasks, ranging from environmental monitoring to agriculture spraying of crops and disaster zone rescues, benefits from the small size, ability to perform vertical takeoff and landing and better maneuverability. However, the lightness of their structure and high level of nonlinearity make quadrotors very vulnerable to the influence of different kinds of disturbances such as wind gusts, sensor errors, thermal drift and actuator biases. Hence, one of the most important research areas related to quadrotors is concerned with their stability and attracts a lot of scientific interest due to both practical importance and possibilities to contribute to the state-of-the-art. This research focuses on various control system approaches that have been proposed in order to improve the stability of quadrotors, including classic PID, LQR, and modern approaches like SMC, backstepping, fuzzy logic and neural networks control. The comparative analysis touches upon the issues of nonlinearity handling, performance in the presence of disturbances, and timely computation. In particular, the results obtained during this research suggest that modern control strategies show better performance when working in a noisy environment, while classical approaches still have some value in terms of their simplicity, less requirements in computing power and other implementation aspects. As to the final conclusion, there is no universal best method of controlling a quadrotor. At the same time, the research shows that modern approaches offer better options in case of performing difficult tasks, while classics are good enough for low-level tasks.