How Autonomous is our Self-driving Vehicles?

Distinguishing the hype from the truth in self-driving systems, and understanding the current limitations, possibilities and everything in between

Remember the old joke, “What’s the difference between a software salesman and a used car salesman? The used car salesman knows when he is lying”. With self-driving vehicles today, the car salesman has automatically become the software sales man and vice versa, are we to expect the hype associated with software sales or the verity with car sales.

With technologies taking over our world and machines controlling how we do things, these days we mostly can’t even see a limit to technological innovations. It is often hard to discount any ideas floating around for the fear of being thought of as ill informed, how would you doubt the existence of mind controlling robot when Siri is seamlessly controlling your web searches. So, a lot of “hype leaders” are getting by, because they are depending on the rest of us being too afraid to ask the tough questions or express doubt, and even when we do the explanations, we get are mostly too ambiguous or too technical to grasp. Today, I would be taking you down the unknown road of understanding what we have right now with regards automated driving, its control and everything in between.

To understand how far we have attained and to be able to tell the hype from the truth, It would be fair to deconstruct the mechanism of the self-driving vehicles, let’s look at what goes on under the hood, and from that point get the knowledge of what the car does and what it can’t. And What can be more basic to this topic than the concept of Artificial intelligence.

“Artificial intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions”

Jake Frankenfield

The major setback with artificial intelligence right now, lies mostly with misinformation. Our perspective on the topic of AI and the future of robotics have been clouded by a lot of science fiction (Sci-Fi) and propaganda, that we fail to see, most applications around us that already make use of these technologies. We harbor fears of robots taking over the world when as common as our email spam controller makes use of such technologies. If you may know, AI technologies around us ranges from the spam collector in our emails, Netflix and Amazon suggestions, and our popular Assistants Siri, Alexa and Cortana.

Ranging from the spam collector to the speech recognition systems, our knowledge on the subject of AI have been based on collected data which are made to be understood by machines and then implemented accordingly.

Machines perform what is programmed and cannot make the judgment of right or wrong. These systems cannot take decisions if they encounter a situation unfamiliar to them, they either perform incorrectly or breakdown in such situations.

“Artificial Intelligence would be the ultimate version of Google. The ultimate search engine that would understand everything on the web. It would understand exactly what you wanted, and it would give you the right thing. We’re nowhere near doing that now. However, we can get incrementally closer to that and that is basically what we work on”

Larry Page

There is no smart box at the middle of these systems or some magic but a preprogrammed algorithm, data collection and interpretation systems.

Basically, how this translates to your un-manned vehicles is that everything should go just well as far as everything you encounter falls within the sector premediated by the makers of such systems. But before you conclude that the usage of these unmanned contraptions is bad, A look at the types of AI should clear some doubts.

In the world of Artificial intelligence, we have categories Narrow AI, and broad AI. The Broad being the type that replicates human beings (robots) and everything that goes with it, and the broad being the systems that perform specific tasks.

The narrow AI are normally designed around specific tasks, this has enabled them attain efficiency in their tasks. Thank goodness driving vehicles fall within the realm of the narrow AI. The Vehicle doesn’t need to detect how you are feeling or how your kid did in school to take you to your destination.

You might have heard of another word always occurring with un-manned vehicles and this is sensor. These are a fancy name for measurement devices, though an advanced version of this. The sensing systems range from as large as the Global Positioning Systems (GPS), to the high definition cameras. But this is not all, the system comprises of a lot of presence detectors, monitoring it’s every move. And all these are fed collectively to the AI algorithm. As already mentioned, this narrow AI technology is an algorithm formed from data coalition and interpretation, this algorithm is then implemented with real time data that are gotten while you move in the vehicle.

Think of it this way, The AI are algorithms that have been tested with buttloads of data, the result is an algorithm, a template we call the “AI for self-driving”. When we run our vehicle unmanned, the vehicle gets real time data and imputes them into this template (AI) to get a result. This result then guides the vehicles along the path it should go.

Because of the advantage of self-driving cars having a narrow scope (Narrow AI) in the world of artificial intelligence, the producers are able to implement as many real-life scenarios that you may encounter, such as a dog jumping into the road or a jaywalker strolling outside the crosswalk.

Well, despite my optimism for this technology I would like to look also at the down sides. An weather disturbance to the technology GPS can cause disturbance on the system, even some users have complained of some discrepancies on snow covered roads. Sure, just like many things around us, this technology still requires more input and in fact, a lot is been done every day to make it better.

My take on the adaptation of this technology, is that in so far as an unforeseen eventuality doesn’t occur on the road, you are free to remain on auto-driver all day long, but if like me you are very cautious of your safety, It wouldn’t hurt to take the wheel and ride the street.

In conclusion, AI itself has limitations in the replication of human thoughts, but for specialized systems that just drive you to your destination these limitations are greatly reduced, but this doesn’t totally eliminate their existence. The thought of AI and all the propaganda surrounding us keep clouding and preventing proper understanding of the proper functioning of this technology making us likely to forget that these systems are a mere physical implementation of our large data and complex algorithms. As of this day, our autonomous feat has just been data driven, and while we’ve tried to copy our human functioning it keeps eluding us, so, our autonomous vehicles should rather be called programmed vehicles or data analysis machines, but this doesn’t sound very appealing does it?

I believe AI can achieve the technological feat of complete automation and self-awareness. But finally the choice of using AI technologies lie with the consumer, personally I don’t see why I wouldn’t try this technological capability at least once (Maybe when I’ve got the chance to), but until then, it would do us all a great benefit to tell the hype from the truth, and know that right now, we just have data driven machines, and maybe sometime in the future we would get there, maybe….