The idea isn't new (hell, it was developed by a Dutch traffic engineer in the seventies as well), but it's nice to see more empirical confirmation. To put it in a wider context, I think it's an interesting idea that as we become more efficient at a certain task, we don't necessarily spend less time on said task.
I'll wager that "self-driving cars" are going to be out of reach of the average consumer for quite some time to come. Timeshare in a fleet of self-driving towncars or vanpools? Not so much. The other great thing is as the number of autonomous vehicles increases, the synchronization of roads increases such that congestion drops.
Self driving cars will help congestion, but probably not as much as people think they will. First, we need to think about the sources of congestion. There are a few, and only some are human caused. The human caused sources include traffic accidents and unsound driving practices. Obviously accidents can mess things up on the road very significantly, but the effects are transient and infrequent. Poor driving probably leads to some congestion also, such as when people fail to merge despite warnings. The other, major source of congestion is simply a problem of physics, specifically the conservation of mass. Traffic can be thought of as an incompressible fluid, so that mass flux must equal zero over any closed loop (in steady state). More simply stated, there's only so much stuff you can pack into a given space. The only way self driving cars can alleviate any of this congestion is by making cars drive closer together, thus compressing the fluid. Even if they halved the safe driving distance, the gains would only be marginal, because each car needs to slow down to accommodate any new cars entering the freeway. Therefore, traffic jams may be more enjoyable due to our ability to read while enduring them, but they certainly will still exist. I wonder whether it will affect the flow of traffic in cities. That is, will self driving cars stop at red lights if they're certain that no pedestrians, other cars or obstacles are present? This could speed up travel times, and increase fuel economy greatly.
This is true when communication is transmitted particle-to-particle. However, before long, self-driving cars will likely be networked such that every car in a locale is communicating with each other, more like a quantum condensate. When the light turns red, every car will accelerate at the same moment, without a transmission delay along the string of cars. The same would go for merging traffic. All cars involved will likely be aware of the merging car's intent before it even gets on the ramp. It's possible that once you set your destination, current traffic will begin to optimize for your trip. I imagine that cars will be packed tighter in the left lane, and that lane change optimizations will take place as cars are intended to be added or removed.Even if they halved the safe driving distance, the gains would only be marginal, because each car needs to slow down to accommodate any new cars entering the freeway.
That's not possible unless all the cars have stopped maintaining an equal distance between them that they would have when they are operating at full speed. It's not as if the guy driving the car 10 cars back doesn't see the light turn green at the same moment that the person driving the first car in line does. The delay happens because space needs to be maintained between cars. Self driving cars aren't going to operate two feet from one another. Even if the engineering allowed them to operate at some minimal safe working distance, say, 10 feet, the way engineering works is by building safety factors into the system, because there are always contingencies we can't foresee. Depending on the nature of the project, a safety factor can vary from 1.5 to 20 in some circumstances. As to your other point, even if the cars are "aware" that another car is going to merge 5 minutes from now, they still need to adjust their speed to account for the space. Just because they do it gradually instead of abruptly doesn't mean that they can drive any faster than space limitations account for. You'll note that conservation of mass is a law of physics, and it can't be violated by computers, even if they're really, really, really smart.When the light turns red, every car will accelerate at the same moment, without a transmission delay along the string of cars.
Actually, I think it's often the case. But even when it isn't the guy usually doesn't press the gas until the brake lights on the car ahead of him turn off. There's a significant amount of delay introduced by people, especially if someone is reading a text message at the light. Also, humans will slow down more than they have to (as they have no way of knowing precisely) when they see traffic slowing ahead, which leads to the accordion effect. Self-driving cars will only slow down as much as necessary. They also won't gawk. :) No, but I think there is a lot of slack to be found in human driving behaviors, especially when the system can be optimized as a whole.It's not as if the guy driving the car 10 cars back doesn't see the light turn green at the same moment that the person driving the first car in line does.
You'll note that conservation of mass is a law of physics, and it can't be violated by computers, even if they're really, really, really smart.