The assertion that data augmentation is an excessive complexity for a simple grasp demonstration like that of the Festo Bionic Handling Assistant is understandable, because it is true that to show a robot can catch a red ball in a very controlled environment, minimal calibration would suffice. However, this view does not account for the real conditions where the robot might be deployed, where lighting, angle, or even object color could vary. Data augmentation, by simulating these variations, prepares the system for robustness that a few examples cannot provide. Imagine a taxi whose GPS only knows the direct route from your house: it would be useless once you change your starting point or if there is a deviation.