Autolanding a Power-off UAV Using On-line Optimization and Slip Maneuvers

Abstract

This paper tackles the final stages of autolanding a fixed-wing Unmanned Air Vehicle (UAV) in a power-fail emergency scenario. We applied the principle of Nonlinear Model Predictive Control (NMPC) to plan trajectories that respect the aircraft's limits, and react on disturbances and environmental changes in real-time. However, thanks to judicious problem formulations, our algorithm optimizes--in each time step--the whole horizon of the landing trajectory, and settles a dynamic compromise between the different touchdown requirements. Furthermore, the algorithm is powered by a novel method that utilizes slip maneuvers to achieve adequate energy control notwithstanding the absence of specialized drag control devices. Thus, precise touchdown point is attained even in the presence of strong wind shear and large initial errors. All design parameters are optimized off-line over a wide spectrum of conditions using Genetic Algorithm (GA). Monte-Carlo simulation of the optimized design shows excellent landing performance and high success rate on both flat and sloping terrains.