Modelling and performance analysis of a machine vision-based semi-autonomous aerial refuelling

A critical aspect in the design of Semi-Autonomous Aerial Refuelling (SAAR) control schemes for Unmanned Aerial Vehicles (UAVs) is the availability of accurate measurements of the relative UAV-Tanker distance and attitude. In this effort, the attention was focused on the development of an accurate modelling of the SAAR manoeuvre and on the development of a Machine Vision-based scheme for the estimation of the tanker-UAV relative pose. The developed MV scheme is based on markers installed on the surface of the tanker, and performs specific tasks as Feature Extraction, Feature Matching, and tanker-UAV relative Pose Estimation. The accuracy/robustness of the overall scheme was evaluated in the event of markers occlusion, in presence of inaccuracy in the positioning of the markers on the tanker aircraft, as a function of the level of attitude and GPS sensors' noise and as a function of the data Transmission Delay (TD) between aircrafts.