Fraction of costs for deployment (no heavy safety, no design phase, less weight and momentum), overall costs, different gripper, train a new use case in around 4,000 hrs video material.

Think of it as a progression of collaborative robots such as cobot Sawyer. And for most new use case, you don’t need millimeter accuracy, high payload or speed to answer your potential remarks.

It most likely will move into a sort of dual-arm humanoid for stationary use cases (such as Jaka K1). Mobile will likely specialize to fit use cases (wheel, legs)