Self-Driving Commentary: Evolution and Attitudes

This post belongs to a series I’m calling Self-Driving Commentary, in which I spout and ramble on whatever is rattling around in my head pertaining to self-driving cars. Enjoy!

I’ve decided that rather than try to enumerate in a single post all of my feelings on self-driving cars and the future I imagine they’ll usher in, I’ll choose a topic from an existing article and discuss my feelings on that particular topic. It seems there’s no shortage of articles discussing the benefits of self-driving cars in the areas of safety (I can’t tell you how many times I’ve read that each year human error causes roughly one million auto fatalities worldwide, thirty thousand in the US alone), environmental impact (nine out of ten cars will be taken off the road, and the quicker potential for 100% electric vechile adoption), and convenience (all that extra work and Twittering we’ll be able to do, or sleeping - I vote for sleeping). I’m sure I’ll touch on those topics here and there, probably tangentially to whatever other topics I may discuss, throughout. For now I’d like to discuss a topic I found in this article (found through the ever-so-scientific process of Googling “self-driving cars will”): Top misconceptions of autonomous cars and self-driving vehicles, particularly the point that driver assistance systems will evolve gradually into fully autonomous cars.

The article states:

To summarize: Driver assistance systems can not evolve continuous driving capability gradually! At the moment we entrust them to drive continuously they require a huge, discontinuous jump in capability which will place their capabilities very close to the capabilities of fully autonomous vehicles.

I agree completely that full automation can’t possibly be simply the convergence of an amalgam of discrete driver assistance features. (Although I don’t necessarily agree that the reasoning can be based so simply on the instantaneous nature of driver assistance features - which could probably be fairly easily enhanced to operate more continuously. e.g. The foundations of instantaneous lane assist might be easily adapted to a continuous lane assist feature.) There are far too many potential inputs to consider all at once, which is why full automation is being achieved through machine learning which allows for more generalized behavior through, basically, mimicry.

My point about these driver assistance features is how they promise to gradually shift public attitudes toward autonomous vehicles (and already have done so). This was a realization I made at Thanksgiving dinner, pointing out to my family that every single damn car commercial is touting some sort of driver assist feature, if not multiple features. I’m not exaggerating - every single one! How could full autonomy not be just over the horizon?

Sure, these features can’t just all be tossed into a bowl, mixed up, and - presto! - full autonomy. But the public at large doesn’t have to know that. All they need to know is, “Well, shucks! My car’s been parking itself, keeping me in my lane, and stopping me before I hit things for years now. Why shouldn’t I trust it to drive all by itself?” That’s what I think driver assist is doing for full autonomy.