All three controllers are tested with the same sequence of four position set-points as shown in the top row of Fig. 7–9 for the controllers III-A–III-C and exemplarily in Fig. 6 for the decoupling controller from Section III-C. No additional interpolator is used between the desired set-points. Note that for the model-free controller from Section III-A the set-point for the end-effector irEE,d is commanded, while for the coupled (Section III-B) and the decoupling controller (Section III-C) the desired position of the UAV’s center of mass irCM,d is commanded. Every set-point is sent as soon as the previous waypoint is reached within 1cm accuracy. Hence, the duration of the trajectory segments for the three different controllers varies, as can be seen in the top row of Fig. 7–9. With the model-free controller it takes about 19s to execution the complete trajectory sequence, the coupled controller needs approximately 17s and the decoupling controller is slowest with about 20s. The mean squared error (MSE) of the deviation from the four position set-points is very similar for the model-free (MSE: 0.028), the coupled (MSE: 0.029), and the decoupling controller (MSE: 0.029). However, the path of the UAV’s center of mass in Fig. 5 reveals a deviation from the direct connection between two waypoints of up to 8cm for the decoupling controller.