NON-SYMMETRIC MEMBERSHIP FUNCTION FOR FUZZY-BASED VISUAL SERVOING ONBOARD A UAV

摘要

This paper presents the definition of non-symmetric membership function for Fuzzy controllers applied to a pan & tilt vision platform onboard an Unmanned Aerial Vehicle. This improvement allows the controllers to have a more adaptive behavior to the non-linearities presented in an UAV. This implementation allows the UAV to follow objects in the environment by using Lucas-Kanade visual tracker, in spite of the aircraft vibrations, the movements of the objects and the aircraft, update has been tested in real flights with an unmanned helicopter of the Computer Vision Group at the UPM, with very successful results, attaining a considerable reduction of the error during the tracking tests.