How can we feel safer in a fully autonomous vehicle?

agichat

#1

Are you ready to kick back and read the news or take a nap while your car takes you anywhere you want to go?

While not necessarily far from the truth in the near future, the statement above is nowhere near as simple as one might conceive. We’ve all heard about the main players in this game, but it is more than just giving the market an Autonomous Vehicle (AV) that is fully capable and tested - it’s about making the people feel and know they’re safer by making use of them.

Let’s address the elephant in the room - there have been tragic stories of fatal accidents caused by AVs in the recent past, and those are losses that we’re still having trouble processing. This is a factor we must weight and consider heavily for the future of AVs on our roads - but in an era where traffic-related fatalities are steadily climbing, scientists and engineers are teaming up in an effort to reduce this statistic by minimizing factors of human error such as fatigue, impaired judgment or lack of civil responsibility driven by negative emotion.

There is room to argue that autonomous vehicles haven’t had enough time on the road and haven’t been thoroughly tested in the same conditions some of us face each and every single day. Still, even though an AV has a limited, strict, data-driven vision of the world around it, it won’t be driving you around drunk, nor distracted by a text message on the mobile phone - it will constantly be reading the data input from its many sensors and making decisions in a split-second.

Looking past the fact that AVs have a very crude view of the world around them, we cannot deny that having numerous and various types of cameras, radar, and lidar (a laser-guided radar) along with a ton of processing (and horse) power will help the machine process all the data found around it. Such sensors allow a clear picture that charts out navigation smoothly, cross-referencing it with GPS for better positioning recognition. Us, humans, rely purely on our senses as input, and our memory as GPS - which we all know too well may prove unreliable.

As AVs evolved over the years, it became clear these machines had obvious advantages over humankind regarding reaction-time to avoid a collision, the ability to see in low-light conditions which enable the detection of potential obstacles and possible communication with other vehicles by sharing reports/logs about the roads travelled.

It is true that automated vehicles will not take the decisions most of us would take and some might even question the morality of letting the machine choose what’s right for us as sometimes the “right” choice may not be the one we’re expecting.

But over 90% of the crashes in the United States alone are due to human error, which makes one think: why are people so fearful of giving AVs a chance for safety? Not only that, but why are people not willing to trade the comfort of the familiarity they have with their current lifestyle for the even greater potential comfort this segment has to offer?

Nearly three-quarters of Americans are still afraid of riding in self-driving cars, but looking at the facts it seems that humans are very close to losing the title of best drivers to AI. Although that is already making rides much safer for some, the majority still fears to travel in such vehicles.

Maybe if there was actually someone in the driver’s seat they would feel a bit more relaxed? Well, there is already a self-driving robot chauffeur! It that can fit in a suitcase and make most vehicles autonomous. Although still lacking emphatic skills, could these handy driving “companions” become the solution for the most apprehensive ones in the near future?

Let us know what do you think in the comments and stay tuned for the next issue!


#2

I have heard a lot of chatter as to why people are so resistant to autonomous vehicles and I think most of it has to do with security. A lot of people feel someone who knows more about technology could hack their car and take control leaving them vulnerable. This might be a bit irrational and paranoid but I see posts on social media and discussion forums bringing up this point more than any others I see.


#3

A much more advanced vision architecture. Alexey would agree. Current DL architectures are mostly blind! And this even caused deaths and the lucky fool Elon Musk didn’t get sued. Basically that woman died because some clueless behaviorists think that RL is a good enough formulation of an intelligent agent, if they thought we needed passable scene representation supposedly such massive errors could be prevented. It’s sad how scientific ignorance can cause (other) people to die.

There are many other possible improvements but the lack of an effective vision system is the most visible flaw IMO.


#4

Absolutely, the primary reason this isn’t already in wide spread roll out is the infrastructure to support it is not quite in place.

Though with implementations of Quantum cryptography the issue of habitability will be reduced somewhat, but history has taught us that nothing is every 100% hack-proof, and there is always a point of failure somewhere.

Though this is of course true of many of our current systems that we rely so heavily on, many of which if they were hacked would cause major disruption, and cyber war is a very real threat that we face every single day.


#5

In line with everything else the technology used will improve, not to mention that when autonomous vehicles are interconnected, they will gain the advantage of the collective leanings of all vehicles and so be able to make better decisions.

I don’t think we will ever get to a zero mortality rate just by the nature of reality being chaotic. however, what I can say with absolute certainty, as we transition to autonomous vehicles there will be fewer deaths by a huge factor as compared to manual driving.

Globally nearly 1.25 million people die in road crashes each year, on average 3,287 deaths a day.

source: Association for safe international road travel


#6

Using the figures, of average cost of vehicle insurance in the US at $940.00 per year
and the estimated vehicles registered in the US in 2016 at 270,000,000 , means the companies haul in, per year

$ 253,800,000,000.00. :grimacing:


#7

Disruption due to technological change is nothing new, history is littered with examples of closed businesses and even in fact whole industries being superseded almost overnight.

The transportation industry is no different…

Think about all the industry that went into maintaining the horses in a city setting, the food that was required, the blacksmiths, the tackle shops, the vets, the waste management, the stables, the hay, the grooms and many other elements too.

All these industries totally ravaged and disappeared over just as ten year period. This occurred more than 100 years ago, so what of the next ten years, when we transition to autonomous vehicles??

Automotive Insurance companies may very well find themselves out of business.


#8

Great picture to represent the suddenness of change Tim, in general I agree with the sentiment here - that the primary force that can make us feel safer would be smarter and and more powerful AI itself. Since the pace of technological change is increasing itself, I’d give it 2 to 3 years before I’ll feel safe :slight_smile: