Autonomous Vision Based Landing Strategy for a Rotary Wing UAV

Abstract

This paper investigates solutions to achieving autonomous landings with multirotor UAVs (Unmanned Aeriel Vehicles). The landings presented in this paper require reliability and high levels of accuracy. The UAV can not rely on traditional GPS for localisation, instead, it makes use of specialised landing pads and onboard cameras. The solutions presented rely on the assumption that landing targets are static. The paper presents the design of 3 different landing pads and evaluates their effectiveness in a simulation environment designed to replicate the real-world and physical UAV as close as possible. The robustness of the proposed solution is evaluated by varying external factors. The paper also addresses state machine design to ensure reliable and safe landings. Other methods of achieving accurate landings are also discussed. Statistical simulation results are obtained using a fusion between Gazebo simulator, ROS and PX4 autopilot flight software.