On Traffic

June 1, 2009

The way that freeway traffic flows is quite inefficient. Each individual driver has to react to many outside stimuli without any knowledge of the intent of the intent of the source of the stimuli. For example - a driver could swerve into an adjacent lane, but if the driver is only swerving to avoid a board in the road and returns to their lane almost instantly, a driver in the lane that car just entered does not “need” to slow to avoid a collision. If the car swerved into an adjacent lane and meant to stay there, however, a nearby driver might have slow or even stop to avoid a collision.

Freeway traffic flow can be improved. If we look to the internet for inspiration, we know that there does not have to a central authority to dictate where each packed (vehicle) has to go to maintain an efficient flow. Instead, those decisions are left to routing stations (the internet backbone) to direct packets in the most efficient manner. To improve the freeway system, we need a moderately centralized source of information as to what routes between here and there are clogged, which are empty, and which are fastest. We’ve already started down that path, but it’s a hodge-podge of systems. We have proprietary traffic data, independent of GPS navigators, and no simple way to integrate everything.

Second, the caterpillar effect is a huge problem for freeways near their carrying-capacity. It starts when one driver decellerates or brakes to avoid a car in ront of them. This causes the car behind them to brake harder to avoid a collision, which then has a ripple effect. In heavy traffic, the chain reaction can eventually bring traffic to a stop. This is why you get traffic jams for no apparent reason. Also, it’s why you shouldn’t weave between lanes, because you just cause jams behind you.

So, to eliminate this from happening, we must remove the ability for drivers to make independent decisions. Vehicles must be able to communicate with each other, and relay what they are doing, as well as that they plan to do. Basically, we need self-driving cars.

Third, we need a way to share work among vehicles. I don’t have exact data, but I suspect that trains are much more efficient than cars or trucks in delivering stuff from here to there. If we could allow cars to connect together to share work, we might be able to reduce overall work and save resources. Imagine big trains of cars.

For example, let’s say that cars A, B, and C are leaving Seattle traveling south. A is going to Tacoma, B is going to Olympia, and C is going to Portland. They couple together until the train reaches Tacoma, at which point A disconnects and proceeds to it’s final destination. Car D joins the train, heading to Centralia. In Olympia, B leaves. In Centralia, D leaves, and E joins also heading to Portland.

This may or may not save work, but it would definitely increase the carrying capacity of freeways, because cars in the train wouldn’t have to worry about being to close to the car in front or behind of them – they’d be connected. The main reason I suspect that it would save gas is from decreased air resistance.