A common robot task is to pick an item from one place and set it in another. It sounds simple until the items vary in position, orientation or type. Vision-guided systems use a camera to locate each object and work out how to grasp it before moving.
The hard part is variety and clutter: overlapping items, reflections and unfamiliar shapes all complicate the view. Progress here is what lets robots handle mixed, disordered goods rather than only identical parts arriving in a fixed, predictable position.