Analysis of flight control and trajectory planning for autonomous ship landing using small-scale UAVS

Abstract

Due to the potential to expand flight envelopes and increase flight safety for sea based rotorcraft, there has been a drive to produce reliable autonomous ship landing algorithms. These algorithms are inherently complex and must be validated by extensive experimentation. Experimentation at model scale offers a controllable test bed that can be used to isolate the effects of individual parameter variations, helping gain insight into the limitations and vulnerabilities of a complex landing algorithm. In this paper, the development of an autonomous landing control mode suitable for use in scaled experiments of this nature is presented. The main goal in algorithm design was to create a landing solution representative of other methods published in the literature, while also allowing for easy variation of parameters during testing. The resulting algorithm features explicit model following controllers, quadratic programming trajectory generation, and deck motion prediction based on autoregressive models. Additionally, a scaling method based on Froude scaling is proposed and results from 230 scaled flight tests to a virtual deck are discussed. During testing, reference tracking bandwidths and jerk limitations were progressively reduced and deck prediction accuracy was degraded. Results show that the landing algorithm performs well for scaled moderate to high sea states and deck motion predictions help compensate for reduced aircraft mobility. It was also found that reduced tracking bandwidths effect landing accuracy, but that inclusion of reasonable jerk limitations can be important to prevent the autonomy algorithm from overcompensating for artificially reduced model scale bandwidths.