Training an AI model with images of apples does not guarantee that the Festo robot knows how to handle them without crushing.
There is a clear difference between visually recognizing an object and knowing how to interact with it physically, especially if it is fragile.
Recognition is part of the problem, but it is not sufficient; it lacks force sensors, haptic feedback, or motor control algorithms.
It's like having an IKEA assembly plan without screws or the Allen key to assemble the furniture; capability depends on other factors.
Without this discipline of physical engineering, the robot would see the apple but crush it because it doesn't know how to adjust its grip.