The majority of the world seems to have reached a consensus that our current transportation systems are not sustainable. Most notably, the transportation system in America accounts for about one−third of the country's greenhouse gas emissions. For one thing, we rely too much on cars.
For this reason, there is a large push in many places for public transportation, which is a good thing. However, the sheer number of people whose lives are based around cars make for very frightening transitional costs, leading researchers to also look into ways to make cars more fuel efficient.
But finding new fuel sources is only one part of improving the way we drive cars. Most readers will be familiar with the fact that there are two miles per gallon (MPG) statistics: highway and city. The highway value is always higher since maintaining one speed is more efficient than constantly stopping and starting. Even on a highway, stop−and−go traffic can result in worse fuel efficiency than estimated for the city. The solution lies (as I usually claim it does) in computers and programming algorithms.
The reason city traffic is so inefficient is, of course, traffic lights. Traffic lights are supposed to be timed to the speed limit so that if you maintain the speed limit, you'll never need to stop. However, lights are not always timed properly. All too often, drivers find themselves stopping at every light despite staying on the same street, even though no cars are coming from the other direction. Pressure sensors that change lights only when necessary do exist, but they are expensive to install.
Luckily, a cheaper alternative does exist: thinking. Traffic lights are not magical boxes that randomly change color; they are programmed by humans. And statisticians can use computer−recorded traffic data and mapping software to time lights such that they stop cars on busy, important roads less often. The timing can even depend on the time of day, so that a route normally used by commuters is given priority in the morning and afternoon but not at midday. This low−budget solution is not implemented nearly as often as it should be, and I'd encourage any reader who knows of a poorly timed light to say something to the town's department of transportation.
A simple improvement on this, proposed by a University of Texas professor, is to use what computer scientists would call a scheduling algorithm. In essence, any car would be able to inform a traffic light that it is coming and wants a green light. Police departments use such a system, but that's basically an on−off switch.
To make this work for regular drivers, the light has to be smart. It has to set a green light for the busier street without making the other street wait too long. It also has to know what the other lights are doing so it doesn't send cars to an already full intersection. Additionally, it can tell a driver its decision so that, instead of speeding through a red light, you can simply coast until it turns green.
Tackling the problem on a larger scale requires the science fiction−style self−driving car. The military encourages most of the research in the field so fewer soldiers have to risk their lives in convoys, but it can revolutionize consumer gas efficiency too. The EU−based SARTRE project aims to create road trains.
Instead of fully automating driving, one expert drives. All other cars use onboard collision sensors (already on many cars) to closely follow but not hit the car in front, minimizing the air resistance (headwinds) each car encounters. The researchers hope for 20 percent greater fuel efficiency from this automated version of auto racing's drafting technique.
--
Ben Schwalb is a member of the Class of 2012 who majored in computer science. He can be reached at Benjamin.Schwalb@tufts.edu.
