Zero visibility autonomous landing of quadrotors on underway ships in a sea state

摘要

Unmanned aerial vehicles (UAV) are a valuable resource and have many applications in marine environments. The UAV's capabilities can be further augmented through increased autonomy. One potential area for more autonomy is landing on the stern of underway ships. This requires good position estimates that are independent of GPS and in many harsh environments, vision. We demonstrate a position estimation and tracking technique that uses peer-to-peer acoustic range measurements coupled with the relative inertial measurement unit estimates from the ship and UAV to create a robust state estimation. A Stewart-Gough platform is used to emulate the motions of a ship stern, where the UAV would land, in different sea-states and a state-of-the-art motion capture system provides the ground-truth positioning for the UAV and the Stewart-Gough platform. We demonstrate a pose estimate with a mean-squared error (MSE) of 9.41 cm (23.8% of landing area length) and a “ready-to-land” position tracker with an MSE of 12.05 cm (30.4% of landing area length). The proposed extended Kalman Filter used to fuse the measurements is more than adequate. The zero visibility autonomous landing algorithm works well and the next stage will be validation at-sea.