It won't be a problem for long. Once the autonomous car arrives, the days of human driven cars at least on public roads will be short-lived. The cost in human lives and reduced road capacity would be far too great to allow otherwise.
Another difficulty in making this concept work well is that most of the cars in the lot will have been parked by humans, so they won't be perfectly centered within each space. Will the autonomous parking agent squeeze your car into a tight space so that the drivers on either side of you bang your car when they open their doors?
Letting self-drive cars circle round a parking lot waiting for a space to become free does not feel like a good solution.
While the parking spaces do not need to provide info to a server under this scheme you have the complex problem of visual recognition of when a space is free or occupied.......and you have lots of cars burning fuel......
And what happens when a space becomes free and two automated vehicles go for the same spot.......I suppose some sort of CDMA back off.....otherwise known as "the tank gets the parking bay."
In fact there are cities in Europe of mulling the idea of putting sensors in all parking bays and then providing available parking information to cars in return for receiving complete download of the CANbus information from the vehicles.
A trial along these lines was conducted in Eindhoven.
Safety will always be a concern with automated cars but I don't think this is just a marketing plot. For average drivers, how often would do they have long drives versus needing to find a parking spot. It is easy to see why automated parking will have a larger market.
But self parking car is truly useful if you can just drive to your favorite restaurant and jump out and let the car find its own spot (and yes, come back to you when you want it again). The car can drive itself around and around the block to find parking and just keep circling until a spot opens up. (Ha! Don't need all parking lots to provide vacancy info to a server this way.)
Of course, you can also see that in this use case safety concern is even greater: like mixing advanced algorithms with jaywalkers.
If a car can find parking spot by itself, that would be just awesome! but imagine the amount of work involved. All parking lots have to keep updating the data real time. The car needs to have access to that data system. The mobile service provider, bandwidth issues. But yes its a good idea and why not. When in today's date smartphone are getting so common this will be another customer service.
Autonomous driving is a good concept and I believe that it will be well received once the consumers realize "what it can" and "what it cannot do." What I have noticed about technology over the years is that all technology that is a good idea on paper, sometime can be a bad idea once it is implemented. But autonomous driving technology sounds very exciting to me.
Nearly as daunting as the development to create an automobile able to find a parking spot and park itself is the R&D behind an autonomous squeegie guy robot. Using a combination of millimeter wave radar and machine vision thousands of these marvels will patrol urban parking lots applying a mixture of rain water runoff and road grit to the windows and headlights of parked cars. Upon completion of the squeegie operation the robotic squeegie guy will use the telematics systems of advanced automobiles to ask for any loose change that may be in the cup holders.
I learned to drive in Boston and got a lot of practice parking (my record was 32 round trips to get into a small parking space). Driving in circles looking for a spot provided a lot of time to reflect upon issues associated with parking. I wonder how the computerized parking system will address 2 issues. 1/ A large box needs to be lifted out out of the passenger side and a small person needs to get out of the driver's side. Therefore, how do I park off center in the stall. 2/ There is a sinkhole in the middle of the automatically selected parking place from the reservation system.