This paper presents a method for an autonomous target approach to buildings for MAVs using image based visual servoing. For this purpose the target is detected by image processing and later on tracked by an optical flow algorithm. To guide the MAV to the target, set velocities are calculated using a two dimensional visual servoing algorithm. To translate these set velocities into practice, two control approaches are tested. The first one transforms the set velocities into set angles without using direct control feedback. The second, more sophisticated approach uses measurements from visual pose estimation to directly correct the navigation filter states. This additional measurement leads to a more precise estimation of the velocity in the navigation filter which is used for a direct control loop feedback in the target approach control.
展开▼