A great deal of the labour inside a warehouse or stockroom is not skilled work at all - it is walking. Filling an order often means travelling long distances up and down aisles to fetch items from where they are stored, then carrying them to where they are packed; restocking means moving goods the other way; and across a large space, staff can spend a substantial part of every day simply covering ground with things in hand. This walking is tiring, time-consuming, and uses people for something that requires stamina rather than skill or judgement, which is precisely the profile of work that robotics has taken on with real success. Warehouse robots - machines that move stock around, or that bring goods to a worker rather than sending the worker to the goods - are among the more established and convincing practical applications of the technology.
The idea that makes them effective is often to reverse the usual arrangement: instead of a person walking to the stored goods, a robot brings the goods to the person. A machine retrieves the shelf or container holding the needed items and delivers it to a worker who stays in one place, picks what is required, and sends the rest back, while the robot fetches the next. The walking, which was the bulk of the time and effort, is transferred to a machine built to do exactly that tirelessly, and the person is left doing the part that actually benefits from human hands and eyes - selecting, checking, handling and packing. Other arrangements have robots simply moving stock from place to place around the building on their own, taking the transport off people entirely.
The value, as with the most persuasive practical robotics, lies in the machine taking a monotonous, physically-demanding portion of the work so that people's time and skill go to the parts that need them. Running a warehouse well involves plenty that a robot does not do - judgement about how to handle particular goods, care with fragile or valuable items, dealing with the exceptions and problems that constantly arise, checking that an order is right. Handing the sheer distance-covering to machines does not remove those human contributions; it clears away the walking that was crowding them out, so the same people accomplish more of the work that matters and spend less of their day as, in effect, carriers moving between shelves.
This is once more robotics being useful through narrowness rather than grand ambition. A warehouse robot does one well-defined, repetitive thing - move stock, or bring it to a worker - and its usefulness comes precisely from how much dull, tiring, distance-covering that one thing removes from people's days. It does not run the warehouse, exercise judgement over the goods, or replace the workers, all of which stay firmly human; it takes the walking. For a business whose operations involve significant fetching and moving of stock, that is a grounded and well-bounded gain: not a warehouse that runs itself, but one where the exhausting business of covering ground is handled by machines suited to it, and the people are freed for the work that actually calls for a person.