Self-driving cars may crank out data at 1GB per second
Self-driving cars, which some experts have predicted will be readily available within five years, will come with a myriad of sensors creating machine-to-machine data at the rate of 1GB a second, according to one strategist.
Mark van Rijmenam, a big data strategist and founder of BigData-Startups.com, believes the sensors in self-driving cars will also will provide great opportunities to spot mechanical problems before they happen—and even schedule repairs.
Hitting the road
Last year, Google CEO Sergey Brin said self-driving cars will be a reality for "ordinary people" in less than five years. Last fall, California's governor signed into law a bill allowing the vehicles on its roads. Tests are also taking place in the UK.
Among others, GM plans to introduce a semi-automated Cadillac driving system in 2015.
"With the amount of cars worldwide to surpass one billion, it is almost unimaginable how much data will be created when Google's self-driving car will become common on the streets. But Google is not the only company working on self-driving cars," Rijmenam wrote.
"By 2020, there will be quite a few offerings out there for autonomous vehicle functionality that you can buy at a reasonable price point," said Thilo Koslowski, an analyst with Gartner.
Rijmenam predicts that all car manufacturers are likely working on self-driving cars. Mobileye, a Dutch company that specializes in inexpensive cameras that assist self-driving cars, has raised $400 million.
"The self-driving car from Google already is a true data creator," Rijmenam said in a blog post this week. "It uses all that data to know where to drive and how fast to drive. It can even detect a new cigarette butt thrown on the ground and it then knows that a person might appear all of a sudden from behind a corner or car."
If a self-driving cars does produce 1GB per second, it would, on average will create about 2 petabytes of data a year, according to Rijmenam. He came to that calculation based on driving 600 hours per year in a car, which translates into 2,160,000 seconds or about 2PB of data per car per year.
Koslowski doesn't agree that autonomous cars will produce 1GB of data per second. While large amounts of data may pass between internal components in an automated car, it won't be stored or even shared because the data will only be used by the car for driving purposes.
"You might have a high-end vehicle like a [BMW] 7 series or [Mercedes] S Class producing a gigabyte of data in an hour that's meaningful and you'd want to analyze to some extent ... but it's not 1GB per second," he said.
Surpassing today's smart cars
Nevertheless, autonomous car technology will increase exponentially the amount of data being produced compared to what cars today create.
For example, cars in the future will have more infrared sensors, inexpensive video cameras and laser-based radar to detect objects around them, Koslowski said. Cars will likely even talk to each other, "see" the velocity of nearby vehicles and react when they turn or brake suddenly. And with computer algorithms and predictive models, a car will be able to predict where other vehicles are going and measure the other drivers' skills—potentially protecting drivers from others' bad moves.
Koslowski also sees a day when automobile data will be uploaded into a cloud storage system that the government can use to make roads safer.
Autonomous cars that have sensors will also be able to identify mechanical problems in real-time and proactively address them. For example, a driver would be notified of a pending mechanical issue before a problem develops, and the car would be able to schedule a maintenance appointment without driver assistance, Rijmenan said.
Event data recorders in 95 percent of new cars already track the behavior of the driver and the performance of the car.
In the future, the data collected "will help car companies to quickly pinpoint the areas for upgrading and adjust the car appropriately. Time to market for new cars will be shortened," Rijmenam wrote.