Automatic Landing Control of a Multi-Rotor UAV Using a Monocular Camera.

The reliability of autonomous landing of the UAVs in an unknown or unprepared environment can be improved by the application of image-based sensors. This paper investigates the landing control of a multirotor UAV by controlling the optical flow and estimation of vertical distance supported by a low-cost monocular camera. Landing control and vertical distance estimation using just a camera as the sensor, makes the proposed approach well-suited for emergency scenarios in GPS denied environment. To develop the optical flow-based controller strategy, an appropriate nonlinear model is proposed by combining the optical flow equations and the kinematics of vertical landing. In the control loop of the multirotor landing, an estimation of the vertical distance is required, which is provided by the application of the developed model and the optical flow derived from camera images. RLS algorithm and the EKF have been applied to estimate the vertical distance. The controller output is determined by using the NDI algorithm, the augmented state-space form of the equations, and the estimated states based on optical flow. An experimental setup is developed for the tasks of optical flow extraction, vertical distance estimation, and control. Additionally, the developed estimation and controller strategy are applied to the nonlinear dynamics of a quadrotor to demonstrate the applicability of the proposed models and algorithms for landing control. According to the results, the proposed optical flow-based control strategy can support a smooth landing for the multirotor UAV while its performance is dependent on the quality of estimation. There is a steady-state error in tracking the optical flow due to the estimation error of the vertical distance. The EKF-based algorithm has better performance in terms of estimation accuracy respecting the RLS, and therefore supports a better landing performance. [ABSTRACT FROM AUTHOR]