Self-driving vehicles could struggle to eliminate most crashes

[Dadhawk]

Originally Posted by Spuds

Incentive = Money.

and Price = Sales.

Right now, the cost of current levels of autonomy add substantially to the price of a lower end vehicle where the majority of sales are, and where there is a lot of sensitivity.

You raise the price of the cars in the bottom 50% of the market, you eliminate new buyers who move to used, where autonomy isn't available, therefore driving full adoption further into the future.

The average age of a vehicle registered in the US was 11.6 years in 2016 and growing (I've seen it reported as high as 11.8 years in 2019).

That creates a very long transition period, and even longer if the manufacturer isn't the "responsible party" in an accident under full autonomy.

[Irace86.2.0]

Certain levels of autonomy are not expensive. Even full autonomy doesn’t need to be expensive and definitely will be cheap in the future. Expensive stuff didn’t stop the mass adoption of more expensive things like automatic transmissions or SUVs, but level 5 autonomy is a few years away, yet you are right that mass adoption will take some time. Many things will influence that.

[Dadhawk]

Originally Posted by Irace86.2.0

Expensive stuff didn’t stop the mass adoption of more expensive things like automatic transmissions or SUVs, but level 5 autonomy is a few years away, yet you are right that mass adoption will take some time. Many things will influence that.

Mass adoption of automatic transmissions took decades. GM started selling the hydromatic in 1940 (or maybe '39), others had automatic transmissions back as far as the early 1900's, but even into the 80's more than half of all vehicles sold were still MT and in 1995 it was still 25% or so.

SUVs have also been around forever (the longest selling car moniker is "Suburban" after all) but I agree adoption rates grew exponentially in the last couple of decades after the minivan fell out of favor.

[Irace86.2.0]

Originally Posted by Dadhawk

Mass adoption of automatic transmissions took decades. GM started selling the hydromatic in 1940 (or maybe '39), others had automatic transmissions back as far as the early 1900's, but even into the 80's more than half of all vehicles sold were still MT and in 1995 it was still 25% or so.

SUVs have also been around forever (the longest selling car moniker is "Suburban" after all) but I agree adoption rates grew exponentially in the last couple of decades after the minivan fell out of favor.

My point was that expense hasn't stopped people from mass adopting something they want. The same argument can be said of paying for smart phones and more expensive data plans, cable TV, etc. Expense is a big factor, but there is nothing overtly expensive about autonomous capabilities. The cost is in developing the neural network, but that isn't inherently expensive either.


https://fortune.com/2016/03/14/self-driving-car-honda/
https://automobiles.honda.com/sensing


The S curve for mass adoption depends on the price, availability, etc for what you are defining. I'm not really concerned about trying to speculate about mass adoption. That will depend on interest, government regulation, price (but that is already cheap), public acceptance, etc. I'm more interesting in talking about when the technology will reach viability for full/level 5 autonomy. I think we are there already or within a year or two of being there.

Experiments have been conducted on automated driving systems (ADS) since at least the 1920s;[9] trials began in the 1950s. The first semi-automated car was developed in 1977, by Japan's Tsukuba Mechanical Engineering Laboratory, which required specially marked streets that were interpreted by two cameras on the vehicle and an analog computer. The vehicle reached speeds up to 30 kilometres per hour (19 mph) with the support of an elevated rail.

Technically, self-driving tech has been in development in one form or another for almost as long as automobiles have been around, so it isn't like we need more time. We are there.


https://en.wikipedia.org/wiki/Self-driving_car

[why?]

Originally Posted by Tcoat

Fully autonomous driving will take a full infrastructure and standardization before it can ever be completely effective. Individual cars looking for lines on the road and relying on radar signals will never be able to do it.

Potholes not yet but even the collision avoidance on the wife's Outback will pick up anything larger than a good sized cat. Considering that is far from cutting edge stuff then it is feasible.

This is so true. There will have to be some sort of transmitters in many things to go along with much better gps and communication between everything. If the road tells the car exactly where it is and where the car should be, autonomous driving becomes much easier.

Originally Posted by funwheeldrive

Why don't we just teach people how to drive?

Because people don't want to pay that much money, and some people will never be able to drive. There is a gigantic difference in aptitudes needed for driving all across the USA. If you live in farm country in the middle of nowhere where there is absolutely nothing to hit, you don't need to drive as well as if you live in the middle of a city.

Originally Posted by Irace86.2.0

There is a difference between knowing how to drive and using that knowledge. For instance, I think alcoholics that drive while intoxicated know they shouldn't be driving, and I'm sure people eating a Whopper, while talking on their cellphone know that distracted driving isn't the best way to drive, and I'm sure people know not to speed or recklessly drive, and they know not to fall asleep at the wheel, and they would like to not have a seizure or stroke while driving, and I'm sure people didn't intend to have that spider fall on their face, and I bet that truck driver intended to change the brakes of his semi last stop, and I'm almost positive that senior citizen would have checked her blind spot before changing lanes if she didn't have a kyphotic spine that is in the shape of question mark. I could go on and on.


The neural learning for Tesla's autonomous driving should follow a S curve, where there is a huge and rapid increase in the development of the net followed by a leveling off. The problem is how long it will take to get to 99% then 99.9% then 99.99%, etc. Eventually there will be a crossover point where the risk of harm is much greater to drive without self-driving enabled, and the payout for a claim if it fails will be rare enough to justify the problem like how it is with manufacture recalls right now. My thoughts are that the rate of neural development will put level 5 autonomous driving available within the next year or two and that it will take a year or two for enough data to come in to support mass adoption just like ABS, backup cameras, seat belts, etc.

Even if it is technically possible in the next year, I doubt it would be legal. They have to prove it can do everything and not cause accidents.

[Dadhawk]

Originally Posted by why?

There will have to be some sort of transmitters in many things to go along with much better gps and communication between everything. If the road tells the car exactly where it is and where the car should be, autonomous driving becomes much easier.
.

Improving GPS would mean coming up with a completely new technology. Even with WAAS, which is used in aviation, the error can still be the width of a traffic lane or greater. GPS is just not reliably precise particularly in urban areas.

[WolfpackS2k]

Good points above.

Also it's worth pointing out that building an SUV doesn't cost more than building a (similarly sized) sedan. If anything it costs less. But consumers are tricked into thinking they're a more premium product and thus pay more.

If someone that's used to only driving around farm country can't handle driving in a big city...then don't. If its a big city they don't need their car anyway. This isn't rocket science. If you can't grasp basic properties of physics then you shouldn't drive. People are lazy. (also I'm not talking about handicapped or elderly)

[funwheeldrive]

Originally Posted by why?

This is so true. There will have to be some sort of transmitters in many things to go along with much better gps and communication between everything. If the road tells the car exactly where it is and where the car should be, autonomous driving becomes much easier.

If only there were some kind of transportation system where a large amount of people could be taken to common points throughout a city autonomously and safely. 🤔

[Yoshoobaroo]

Originally Posted by funwheeldrive

Why don't we just teach people how to drive?

I don't know. We've never properly tried it. It could work!

[Irace86.2.0]

Originally Posted by why?

Even if it is technically possible in the next year, I doubt it would be legal. They have to prove it can do everything and not cause accidents.

You are right. I'm sure the government would need to make a new set of tests like crash tests that a full autonomous vehicle would have to perform before being able to declare the car is fully autonomous. If a manufacturer like Tesla pushes an autonomous update to cars that is Level 5, but they say it isn't fully autonomous and requires drivers to maintain awareness and check in via steering wheel feedback then wouldn't that give them the ability to pass on the legality until the government sorts things out?

Either way. We are pretty much at the testing phases for Level 5:

https://www.traffictechnologytoday.c...r-journey.html

Originally Posted by funwheeldrive

If only there were some kind of transportation system where a large amount of people could be taken to common points throughout a city autonomously and safely. 🤔

You mean where someone takes the railway into the city then a taxi to the bus terminal then a bus to the epicenter then a taxi to the destination? Crap. How do I carry this Ikea cabinet and lamp back?

[Red-86]

Originally Posted by funwheeldrive

Why don't we just teach people how to drive?

We’ve been trying to do that for about 100 years now, and the roads are still full of incompetent, inattentive, dangerous drivers!

[Red-86]

One thing to bear in mind is that the purchasing power and infrastructure to implement fully autonomous vehicles around the entire world is likely much further away than the leading edge of the tech in select wealthy countries. IOW, even if Tesla and others develop fully autonomous vehices fairly rapidly, most of the millions of new cars sold around the world each year will continue to be human operated for basic economic and logistical reasons. And even after the majority of new vehicles sold in, say, the USA or EU are fully autonomous, it may be a few more decades after that before the same is true throughout developing countries.

IOW, I think the autonomous vehicle revolution will be a very slow burn, rather than a rapid changeover. And that suits me fine - I enjoy driving, even if it is just a trip to the shops. Admittedly, it’s mostly fun because I prefer sportscars. If I could only drive boring SUV people movers, then it loses some of the appeal.

[Irace86.2.0]

Originally Posted by Red-86

One thing to bear in mind is that the purchasing power and infrastructure to implement fully autonomous vehicles around the entire world is likely much further away than the leading edge of the tech in select wealthy countries. IOW, even if Tesla and others develop fully autonomous vehices fairly rapidly, most of the millions of new cars sold around the world each year will continue to be human operated for basic economic and logistical reasons. And even after the majority of new vehicles sold in, say, the USA or EU are fully autonomous, it may be a few more decades after that before the same is true throughout developing countries.

IOW, I think the autonomous vehicle revolution will be a very slow burn, rather than a rapid changeover. And that suits me fine - I enjoy driving, even if it is just a trip to the shops. Admittedly, it’s mostly fun because I prefer sportscars. If I could only drive boring SUV people movers, then it loses some of the appeal.

Basic economic and logistical reasons could be the very reason we move to autonomous driving faster.

Society might see a greater cost savings to go autonomous. Healthcare costs could go down from fewer accidents. Insurance rates could drop. Damage to streets, trees, poles, etc would go down. People would be more productive instead of living with long term disabilities. Less people would need to be prosecuted for DUIs. Police duties would decline.

Most people would love to commute to work or home with the ability to nap, to talk to family, or to catch up on social media, work, emails, etc, and now they would. Businesses could benefit from not having to pay truckers and from running the trucks more often. Deliveries of all types could arrive in super small cars (smaller than Smart cars or old Minis). Taxis could be summoned and cheaper.

Cars could be made without steering wheels, without traditional readouts and without pedals. Cars could be constructed with completely different shapes and interiors like seats that don't face forward.

The technology would be very disruptive.

[WolfpackS2k]

Originally Posted by Red-86

We’ve been trying to do that for about 100 years now, and the roads are still full of incompetent, inattentive, dangerous drivers!

Plenty of other countries have succeeded...