Asserting that the robotic hand grasps an apple as a simple extension of grasping a small red ball grossly underestimates the inherent complexity of maximal performance.
A ball is a uniform, predictable object, but an apple, with its irregularities and fragility, requires force calibration and tactile recognition that go far beyond.
To achieve optimal yield, such as picking fruits without damaging them, a re-engineering of algorithms and sensors is needed, not just a slight "specialization".
It is the difference between holding a pebble and lifting an egg without breaking it: one is a basic task, the other requires extreme precision to avoid total loss.