Monday, May 6, 2013

Self Driving Cars, a Necessary Step Towards Flying Cars

At an SAE conference last February Google product manager Anthony Levandowski stated that Google's self-driving cars (SDCs for the rest of this post) would be ready in five years. Despite the myriad concerns and technical difficulties surrounding SDCs, I believe widespread acceptance is inevitable provided vehicle-accident fatality rates don't fall dramatically in the near future. Safety, convenience and efficiency will drive the adoption of SDCs.
Google SDC

In 1944 it was predicted by Oliver Parks that flying would become as ubiquitous as driving. Up to this point in history, however,  trends have been in the opposite direction. The cost of personal flight has outpaced inflation by a significant margin and the number of certified pilots as a percentage of the population has been slowly falling for the last 30 years.

Why hasn't Oliver Park's prediction panned out? It's not because we can't make airplanes that are cheap enough and good enough. Most light aircraft are less complicated and less sophisticated than most cars on the road today, given equal scales of production there is no reason why modern light aircraft shouldn't be equal to modern cars in cost. There are two main reasons that flying cars have failed:

1. Pilot training. As a flight instructor, I know first hand that learning how to fly is immensely more difficult, expensive, and time consuming than is learning how to drive. Flight training to the private pilot standard typically takes 8 months and $8,000 for a committed student. This cost in time and money is not feasible for most people... though if a national culture of flying were developed it's possible that this factor could be overcome...

2. The limited nature of the current Air Traffic Control (ATC) system.The current ATC system is scheduled to be slowly replaced by the "NextGen" system to increase efficiency, capacity and safety. Here "increasing capacity" means that the new system will be able to control maybe 12,000 flights (optimistic) simultaneously whereas the old system's capacity topped out at about 6,000. Currently there are about 5,000 flights in the air over the US at any given time. 5,000, 6,000, 12,000...
Ghetto Flying Car
These numbers are quite small compared to the number of flights that will occur if flying became as ubiquitous as driving. Currently, there are at least 125 million vehicles registered in the US (The way different states report registrations makes it difficult to tell exactly, real number probably ~double the given figure), if we take this minimum number of 125 million and assume that each car is driven for a half hour each day that puts at least 2.6 million cars on the road at an average time.  If each one of these cars became an aircraft, current ATC capacity would be exceeded by about 50 thousand percent.  A radical overhaul of air traffic management would have to occur before flying cars could become ubiquitous. And this new system would have to be immensely more flexible than the current one because... what if mom forgets the milk and has to go back to the store? What if you decide to make an impulse buy at McDonald's? To accommodate such a dense and dynamic traffic system immense amounts of very high quality data would be needed. Data about where you are, where you are going, where other traffic is and where it is going, where obstacles are and what the rules are etc. etc. Fortunately this is exactly the kind of data that is needed for self driving cars as well, recent rumors suggest that Google's self-driving cars collect and process roughly 1 Gb of data per second, wowza. It is easily imaginable that self-driving cars will eventually form a network, communicating with each other to prevent accidents and reduce traffic congestion without the need for a central control system. An extension of this system to more fully three dimensional control seems quite straight forward.

Yes! in one elegant swoop, the data collecting and interpreting tools that will power self driving cars will also solve the two biggest obstacles to ubiquitous personal flight. If self driving cars become ubiquitous, I think it is safe to assume that  the long delayed advent of ubiquitous personal flight will follow. Yayyy!

Fantastical Future Flying Car Utopia