Fuzzy Visual Servoing for Micro Aerial Vehicles

We present a visual servoing approach for controlling a micro aerial vehicle which is based on monocular camera input and a fuzzy logic controller. Visual pose estimates are obtained using a state-of-theart visual SLAM approach combined with marker-based metrical initialization. In comparison to related work, we demonstrate that our system tolerates quickly changing velocity measurements which are commonly observed in visual pose estimation, and that it is well suited for autonomous inspection tasks because it does not require a nadir view onto the scene.