Learning control of a flight simulator stick

W.J.R. Velthuis, Theodorus J.A. de Vries, Koen H.J. Vrielink, G.J. Wierda, André Borghuis

    Research output: Contribution to journalArticleAcademicpeer-review

    Abstract

    Aimportant part of a flight simulator is its control loading system, which is the part that emulates the behaviour of an aircraft as experienced by the pilot through the stick. Such a system consists of a model of the aircraft that is to be simulated and a stick that is driven by an electric motor. To make the simulation as realistic as possible, the simulator stick should behave in the same way as the stick in the real aircraft. However, due to the properties of the motor and the stick, small irregularities can be felt when the stick is moved, which do not occur in a real aircraft. Probable causes of these irregularities are cogging in the motor and small imperfections in the transmission. Both disturbances have a reproducible nature. Because the disturbances are reproducible, feedback error learning control is used for control. The learning controller consists of two neural networks. One neural network is used to compensate the unknown friction and is operated in feed-forward. The other neural network compensates cogging and imperfections in the transmission and is operated in feedback. Experimental results showed that the learning controller is able to compensate the disturbances.
    Original languageUndefined
    Pages (from-to)29-34
    Number of pages5
    JournalJournal "A"
    Volume39
    Issue number3
    Publication statusPublished - 1998

    Keywords

    • METIS-111785
    • IR-78690
    • Intelligent control
    • Adaptation
    • spline networks
    • feedback error learning
    • flight simulator
    • neural control

    Cite this