Corridor 3D Navigation of a Fully-Actuated Multirotor by Means of Bee-Inspired Optic Flow Regulation

Abstract

This paper deals with the issue of autonomous indoors navigation related to Unmanned Aerial Vehicles. Here, we simulated two hexarotors: a fully-actuated one that maintains level its attitude, and therefore that of the visual sensors; and an under-actuated one. Both vehicles were meant to fly forwards in a tunnel while reacting to the irregularities of the terrain, adopting a bee-like behavior based on a nonlinear optic flow regulation. The dynamic models are provided by means of the Newton-Euler equations, nonetheless, the unit quaternion representation is used for a suitable treatment of the rotational motion. The attitude stabilization depends on the knowledge of the quaternion itself, moreover, and due to the non-linearities related to the translational optic flows, visual guidance relies on the implementation of adaptive integral sliding mode controllers to accomplish a triple direct regulation (forward, side and lift commands). We compared both vehicles performance using detailed numerical simulations, validating the concept that a fully actuated hexarotor permits to improve the optic flow based navigation task.