Eliminating the driver is the half-cure
I've been watching Google's effort to produce a driverless car with mild interest, mild because it's no panacea, with interest because my degree and first work out of university was transportation engineering. (One of my classes was on traffic flow theory; I was the only student to take it that year.)
One claim is that the driverless car will reduce congestion; the counter-claim is that it won't. Let's take a look at what causes traffic to jam.
The capacity of a road consists of several variables. Let's take the ideal example of a freeway between major city A and suburban bedroom community B. Here is how to reduce traffic jams:
- Increase the speed limit (the faster the limit, the more cars get from A to B per hour)
- Increase the number of lanes (the greater the number of the lanes, the more cars)
- Reduce the distance between cars (the shorter the distance or "tailgating," the more cars)
- Reduce the size of cars (the shorter the vehicles, the more cars)
This list, however, contains a inverse problem: faster speeds require cars to drive further apart (the two-second rule for safe braking distance). Drivers naturally tail-gate during rush hour, increasing the capacity. The traffic jam is created when one driver brakes, causing the driver behind to brake slightly more, naturally. The overbraking creates a chain reaction to the point that one car in the line finally comes to rest. (After this the traffic jam jerks along, as drivers speed up too quickly and brake too much.)
You can see how Google engineers think they can solve the traffic jam problem through driverless cars: small, fast cars can drive with near-zero distance between them. Trains don't experience traffic jams.
But there are other factors that jams up traffic on freeways:
- Each onramp adds the near-equivalent of a lane of cars, reducing capacity
- Narrow bridges or tunnels eliminate the equivalent of one or more lanes, reducing capacity
- Accidents and road construction remove one or more lanes; emergency vehicles slow traffic
And then there are the human factors:
- Tunnels have a curious effect on drivers, in that they hit the brakes at the bottom of the tunnel, reducing capacity
- Accident sites cause drivers to slow as they turn their heads to gawk
- A single slow vehicle (careful driver, or heavy truck) effectively blocks an entire lane for some distance
- As capacity increases, more drivers take the roads or commute from futher distances
You can see that road filled with driverless cars helps in some of these additional situations, but not all of them -- and especially not when the road is not a freeway. Additional factors hinder capacity in ways that driverless cars are helpless to change:
- Red and yellow traffic lights; stop and yield signs; traffic circles
- Pedestrians needing time to cross roads; pedestrians interfering with right turns on red
- Cyclists sharing lanes with cars
- Protest marches and fund-raising runs
- Busses stopping in lanes; busses being slow to start at green lights; busses taking up as much space as 2-4 cars
(It is ironic that transit increases congestion, which is proven here in Canada each time bus drivers go on strike: rush hour traffic immediately flows more smoothly.)
Unless governments mandates driverless cars, segments of society will continue to enjoy the experience of driving cars, making the advantages moot. Perhaps solutions could include:
- Hybrid-control cars (which already exist) that optionally take away control from drivers, such as determining the intercar distance and doing parallel parking
- Specific lanes designated for cars in driverless mode, just as today there are car pool lanes
I know that on that annual one-day, 13-hour road trip to the in-laws, I wouldn't mind the car taking on the task of doing most of the driving on those long, largely-empty roads in Western Canada, while I snooze.
Comments