Algorithm detects fast and furious
Published: Wednesday, December 7, 2011
Updated: Thursday, December 8, 2011 03:12
Ohio was host to 1,021 accidents in 2009, according to statistics from the National Highway Traffic Safety Administration. Imagine how many lives would be saved across the county if there was a way to predict accidents fast enough for people to react. Even if a person lands in the hospital, that's one less person in the graveyard.
Researchers at the Massachusetts Institute of Technology have begun developing an algorithm that predicts which cars were most likely to run red lights. The algorithm detects the car's deceleration rate and its distance from a light. Using this data, it determines which cars are most likely to obey traffic laws and which cars are most likely to disregard them.
Interestingly enough, the algorithm worked 85 percent of the time, according to an article on Popsci, a magazine that highlights popular science. It successfully identified potential violators seconds before they reached the light. Researchers contend the algorithm detects potential violators in enough time for others to react. This technology would be able to prevent vehicle collisions. In doing so, this handy piece of work could save lives. The traumatizing and discouraging videos shown in driver's education of people running red lights may not be a future concern if researchers can find a way to implement this in cars.
As of now, researchers are developing a system that enables vehicle communication. Vehicles would have computers that transfer data, such as one car's speed, to another car. It could potentially warn a driver not to go on a green light because somebody is about to run a red one. The technology would warn drivers in enough time to prevent a rise in insurance costs, costly vehicle repairs and possible trips to the hospital. If researchers can perfect this technology, it could be integrated into future cars.
While this would reduce the number of accidents, the use of technology in vehicles poses an interesting future for traffic laws. The algorithm detects which cars are most likely to violate the law, but they don't detect why someone is about to run a red light. At each intersection, researchers should collect data to study how many people run that red light. It could simply be because the light is poorly timed; people might have to slam on their breaks to catch it and when they can't they run it. In instances when the road is slippery from ice and water or a car has poor suspension, poorly timed lights pose more of a risk than a safeguard. Other times, a person could have a health problem or be part of a funeral brigade; the algorithm wouldn't be able to detect this. If law enforcement adapts this algorithm technology, they should be sensitive to matters in which a person really had no choice.
Researchers should spend more time testing this technology before it's widely released. Once it is refined, vehicle owners should be taught how to maintain the system, and it should be checked to ensure that it can't be hacked or manipulated. If widespread manipulation could occur, it could mean more deaths than lives saved. It's also going to be prone to bugs and false alarms, and it's best the company spends at least a year monitoring and developing the technology before it's implemented into cars.