Robust Visual Landing Control of Quadrotor on a Moving Platform: A Sampled-Data Approach With Delayed Output and Disturbances

Abstract

This article presents a position-based visual servo approach to allow a quadrotor to visually land on a moving platform, addressing perception delay and disturbances using only sampled-data output feedback. Without the platform’s model prior, the relative position of the quadrotor and the platform is determined by capturing the AprilTag on the platform by an onboard camera. The limitation in the camera’s sampling frequency yields only discrete output with time delay, emphasizing the requirement for a sampled-data method, since continuous system theory is not applicable in this scenario. In addition, disturbances arising from the unknown platform motion, wind resistance, and attitude tracking errors are also unavoidable. To mitigate these issues, a sampled-data time-delay extended state observer (TDESO)-based predictor is developed, capable of actively predicting the current states and disturbances. Using these predictions, a composite sampled-data controller is devised that incorporates disturbance feedforward compensation, thus enhancing the system’s robustness against disturbances. Rigorous Lyapunov analysis is provided, offering a guarantee that the states of the sampled-data control system converge asymptotically to a bounded region, even in the presence of perception delay and disturbances. The effectiveness and practicality of the proposed algorithm are supported by simulations and experimental results.