As I’m sure all of you have heard, companies such as Uber and Google have been working on self-driving cars. With these self-driving cars, the end goal is to no longer have need for a driver at the wheel, and therefore give what would have been the driver the ability to be just like any other passenger in the car.
Now, I have covered self-driving cars in this blog before (take this post for example), but this time is different – the risks are coming more-so to light than what they have before.
Accidents have occurred with these self-driving vehicles before (take this article for instance), but so far, in the majority of cases, these accidents had some human error involved.
So why the sudden shift of tone? Well, scientists have been conducting tests on what are called “computer vision algorithms”, which are used to figure out what each object is, whether that is a traffic signal or a person. These tests have not been showing great results either – they have been able to find ways to degrade the ability to recognize these objects with these algorithms. What does that mean? Unfortunately, this means that with a self-driving car, if this ability is degraded, there may be cases that the car could potentially hit a pedestrian. That’s not good!
What should we do now? The best thing to do right now is to conduct further research in order to limit this risk as much as possible. It’s not going to be entirely perfect (that’s impossible in the tech industry), but if this further research can limit this risk to where it is very unlikely that such an attack to its algorithms can occur, that would be a great step towards self-driving cars becoming a more sought-after reality.
That’s all for now, until next time…keep that tech gear ticking.