Hello and thanks for reading.
There was one scene from the movie "Blow" that I remember getting a bit of a kick out of. As the popularity of cocaine had taken off, Johnny Depp's character had a unique problem...where to store all of the cash that they had. He and his friend had it piled up in every room of the house in boxes. While most of us would never do his occupation, the idea of having that much cash to store is a problem that many of us would greatly accept.
This scene came to mind when I heard some of the numbers regarding how much data a driver-less car generates (and often needs to transmit). Simply....where do we store all of this data? Do we have any idea how many extra server farms we may need to pull this off?
Here is what I am talking about (the lines in bold are thanks to our friends at the Financial Post):
Self-driving car prototypes already generate mind-bogglingly massive amounts of data. The one being tested by Google’s parent company Alphabet Inc., for example, collects about one gigabyte of data per second, or a feature-length, high-definition film’s worth of data every five seconds.
If the world’s one-billion-plus cars were self-driving, each one would generate about two petabytes — or two million gigabytes — of data every year on average, according to an estimate by big data strategist Mark van Rijmenam.
Ok, to put your head around this....the top iPad stores 256GB of storage. So, each car would fill up just shy of 8000 iPads worth of data each year. That is each car. When you determine that there is 1B cars on the market, and now add in all of the trains, planes and other vehicles, well, it doesn't even start to register in my head.
Where do we even start to begin to start to be able to store this information? Ok, let's go one step further....imagine we did store it....how do we make this useful? The reality is that 99.9% of that data will be either redundant or will have no use to anyone. We would need data filters that are way more advanced than we have now.
As well, most of the data from driver-less cars needs to be streamed to/from the vehicle in real-time. The vehicle needs to be updated at break-neck speeds to be able to make the correct decisions, I think we all get why. So, how do we begin to transmit all of this data? I have heard of trials involving advanced WiFi systems in towns, but are we going to extend this to all roads on the market? The sheer logistics of that would be staggering, not to mention that it would swamp the component market so much that no one would have available components to build anything else.
The point of this blog was just to mention the amount of data generated from self-driving cars. It wasn’t intended to cover many other issues such as security and roaming capability – i.e. could a tourist from Mexico's system work in Texas? It’s clear there is a lot more work to be done to get these systems off the ground, so don't expect to see it anytime soon.