Autonomous or self-driving cars have sparked quite a bit of interest and debate in the last year or so. There are the potential benefits, such as safety and productivity on the commute. And potential pitfalls such as a hacker attack weaponizing the cars or the massive amount of the workforce that could be displaced. With the tech-car company Tesla Motors making headlines with their “auto-pilot” mode on their Model S to more mainstream companies such as Toyota and Nissan looking to make self-driving cars, the momentum is there. However, Advanced Driver Assistance Systems (ADAS) generate massive amounts of data that needs to be collected and analyzed, creating a new Big Data challenge.
Autonomous or self-driving cars have sparked quite a bit of interest and debate in the last year or so. There are the potential benefits, such as safety and productivity on the commute. And potential pitfalls such as a hacker attack weaponizing the cars or the massive amount of the workforce that could be displaced. With the tech-car company Tesla Motors making headlines with their “auto-pilot” mode on their Model S to more mainstream companies such as Toyota and Nissan looking to make self-driving cars, the momentum is there. However, Advanced Driver Assistance Systems (ADAS) generate massive amounts of data that needs to be collected and analyzed, creating a new Big Data challenge.
With ADAS people like to think about the flashy items like fully autonomous vehicles though this covers much more ground such as Adaptive Light Control, Adaptive Cruise Control, lane departure warnings, traffic sign recognition, automatic breaking, and self parking. According to the Society of Automotive Engineers (SAE), there are six levels of automation (pictured in the chart above). Several companies are working through these levels as they focus on the final goal of self-driving cars. In the mean time, they are generating petabytes of data that needs to be stored and gone through to gain insights and speed up research and development.
As most probably know, having a car drive to a destination by itself isn’t as simple plugging in the GPS coordinates and letting it use Google Maps to get there. There are hundreds of issues involved including traffic, other drivers, weather, detours, accidents, objects in the road (even people or animals), and the map just being wrong. These are all things the human mind can deal with as they come up with the driver putting little effort into doing so. ADAS on the other hand needs to monitor sensors, cameras, radar, and light conditions. And as car companies look to move up in SAE levels they need to evaluate and improve their vehicles, some times millions of miles worth of data.
Dell EMC has been looking into this issue and have provided some interesting numbers to make one stop and think. For example a state-of-the-art front looking radar (FLR) operating at 2,800MBit/s and it may need to capture 200,000 miles of data. At 60MPH that is 3,333.3 hours of data, at 2,800MBit/s (or 1,260GB/h) that results in 4.2PB for one sensor, some cars may have up to a dozen. Again this is for one car; a company may have an entire fleet of cars about testing for SAE qualification. This data doesn’t even take into consideration the regulatory requirements of storing the data, which can last for years or decades.
Car manufacturers and ADAS supply companies need a large scale out storage for this data. Dell EMC’s solution is a large scale out NAS that is simple with high capacity, efficient data retention, and high availability. Of course this describes their Isilon product and the above research they conducted also makes Isilon a more attractive option for those looking to store and analyze the above data. While Isilon isn’t the only option available it is one of a handful that can handle such massive amounts of data.
Big Data is a buzzword(s) that has been tossed around a lot in the past year or so with no real clear definition of what it was only that it was some nebulous information that was extraordinarily important. Putting a specific set of parameters around the concept paints a better picture of not only what big data truly is, but the potential benefits that can be reaped by capturing and leveraging it.
Sign up for the StorageReview newsletter
