There are really a bunch of parts here, with a lot of unknowns. But I'll take a stab at it anyway. (For those who are wondering, the headings below (except the last one) are based on the original Question. The phrasing of the question has been changed, but the main points are still, IMHO, applicable.)
Centralized or decentralized?
I would use the term distributed. In a way this is actually similar to today's air traffic control system. There is no one master control center. Instead, each airport and each area within a country (I believe sector is the proper term) controls a section of airspace. With the significantly higher traffic volume that comes from low-cost airplanes combined with autonomous flying (the cost of airplanes and the training required for pilots are both limiting factors today), I envision a multi-level distributed system. Basically:
Mini-Sectors - Very small areas - think "cell phone tower range" - perhaps a few square miles each, would have low-level control systems tracking every plane with speed & trajectory to handle takeoffs/landings within the area and coordinate traffic to prevent mid-air collisions. The existing TCAS would be enhanced to work with both these small areas and all aircraft to keep traffic flowing smoothly & safely despite the much higher congestion than today's airspace. Each of these area control systems would communicate with all aircraft under its control, with each immediately neighboring area (typically 4 - 8 areas) and with the next level up - the sectors.
Sectors - Larger areas would communicate with all smaller areas under their control and with neighboring larger areas. These sectors might be 50 - 100 miles on a side (i.e., 2,500 to 10,000 square miles). They would coordinate flight plans for all traffic inside them and direct aircraft between the smaller areas for safety (primary) and for efficiency (shorter flights = less fuel and less time in the air). The sector controllers would also notify each smaller area of relevant restrictions due to weather or congestion.
Regions - The largest areas would cover large sections of a country or entire countries - perhaps 6 - 10 to cover the 48 contiguous states and manage overall congestion, larger routing issues (e.g., traffic from Chicago to New York might have several air corridors and only at the Regional level could you make the decision which route the next plane should use).
Each of these levels would have the capability to service surrounding areas if needed - i.e., redundant with automatic failover. In addition, if the larger system (Regions) totally fails then the Sector controllers would direct all aircraft to land within their Sectors or the neighboring Sectors. If all Sectors (i.e., including redundant failover neighboring Sectors) fail within an area then the Mini-Sectors would place all aircraft into a failsafe mode, essentially large holding patterns until everyone is down safely.
How is priority determined?
Size and source. Your jumbo jets with hundreds of passengers will get priority over the single-engine 4 passenger Cessnas due to the fuel usage/number of people affected. Government & military will have priority as well when necessary. But most of the time everyone will be the same - just like buses, trucks, private cars and government vehicles (except for the occasional siren-for-emergency) do on the roads. If anything, it will be easier in the air because, except for takeoffs & landings, 3-dimensional travel provides a lot more flexibility.
How does it deal mid-flight destination change, or time changes?
No problem at all. Think about driving: If you are relying on GPS-based routing today, you punch in a new address and the system in your car or phone tells you where to turn based on traffic congestion, travel speeds, etc. No different in the air. The Mini-Sector would take the information and deal with the small stuff (and safety!) while handing off to Sector and/or Region to get you a new overall route if it is a big change.
What laws and regulations would need to change / be removed?
As with autonomous driving, the biggest change will be allowing the plane to fly itself instead of a pilot being in control. The catch is that there are millions of licensed drivers but far fewer licensed pilots, so this will become a sudden change once the technology is ready instead of a gradual Level 0/1/2/3/4/5 change as is happening with cars. Pilots today work with levels 1/2/3, maybe 4 (not sure). But to have ordinary people who have no pilot training fly, they can ONLY use Level 5.
The other change is maintenance issues. Cars, at least in the USA, have minimal maintenance requirements. Some countries (and some individual states in the USA) have stricter requirements, but nobody has anything close to the requirements placed on aircraft. Those requirements are a very important thing and not likely to change. But a lot of automation and new infrastructure will be needed to be done to support maintenance of millions of new aircraft.
Where are the landing strips? Is it possible to make vehicles stored in a personal garage, or do people have to travel to an airport to fly their airplane?
Airports, but many new airports will be needed. New airports will be designed with the requirements and limitations of small autonomous aircraft in mind. They won't need 10,000 foot runways. But they WILL need efficient service areas for routine maintenance and large hangars - perhaps multilevel given the number of aircraft involved and the limited space available around cities. A typical trip will be to take your Tesla from home to the airport, hop in a plane and fly 50 - 200 miles to your destination and then take an autonomous Uber to your destination. Convertible car/plane is not really practical for a lot of reasons.
Ride Sharing/Rentals
One likely possibility, even with < $50,000 aircraft, is to have ride-sharing or rentals in place. This is already planned for autonomous cars and would work even better for planes because of the maintenance and other issues involved. But thanks to the lower cost of the planes (and therefore lower amortized cost per ride) and the autonomous flying (allowing flexible on-demand flights), the cost becomes competitive with other forms of medium to long-distance travel without the hassles of big airports.
