Volvo Self Driving Car... Fail

[EldrickOnIce]

If I could predict the future, I would be rich - but I'm not. However, my gut tells me:
The automobile will become an even greater status symbol of the wealthy, because the cost of a driverless car will be significant. As a result, there will be a significant shift in the way we approach vehicle ownership. Unless one drives as part of employment, we operate a vehicle maybe 2 hours a day? So, matching needs based on driving habits - maybe 100 individuals share 25 driverless vehicles. Whether I commute to work or to transit, the vehicle no longer sits there for the day (or in my driveway/garage). Instead it is used by later starting commuters, retired, elderly, etc, etc, etc. I would have no idea what kind of a real number of users per vehicle would actually look like, but I find the concept of it really interesting.
I expect somebody will get rich perfecting something along this kind of 'lease' arrangement.

[BlackArcher101]

Originally Posted by Flash Walken

If the only cars on the road are driverless, they will compensate for the one computer that may have a malfunction.

It's the same concept behind herd immunization.

Honestly, bumpy roads? Do you think Taber is more or less of a bumpy drive than this?:

Stop using space shuttle programs and aircraft as examples of why computers will work in this application. If you want that level of hardware, you and I nor the government will be able to afford to put that many cars on the road. Just look at the effort and testing involved into just one little piece. TCAS for aircraft, a method of communication between aircraft to prevent crashes, is a relatively simple device compared to what would be required for self driving cars. That still took 20 years to really get right.

[MrMastodonFarm]

Originally Posted by Flash Walken

So this has gone from a thread illustrating the faults of driverless technology to providing irrefutable evidence towards the faults of human drivers.

Maybe time for a thread title change, "Idiot motorist is an idiot, injures pedestrian."

Originally Posted by Flash Walken

If the only cars on the road are driverless, they will compensate for the one computer that may have a malfunction.

It's the same concept behind herd immunization.

Honestly, bumpy roads? Do you think Taber is more or less of a bumpy drive than this?:

Flash sums up and wins the thread with back-to-back posts right here. Some of these posts, especially from polak are so simple it hurts.

And to the guy who can't believe these could be used for the blind? Umm yeah, of course! They can also be great for the severely handicapped who can't get around, it's freedom baby, yeah.

My mind immediately went to this video I watched days before.


Fredick Brennan has brittle bones disease and after watching that the sense of confinement and lack of a sense of freedom would terrify me. A driverless car would open the world to anyone with such a physical disability, the way it opened up the world to the rest of us when we were teenagers.

Instead of relying on a Handibus that may or may not show up, might show up and leave, might fault him for not being exactly on time, Brennan could have his own driveless car that his wheelchair could roll right into and strap in.

[polak]

Originally Posted by Flash Walken

If the only cars on the road are driverless, they will compensate for the one computer that may have a malfunction.

It's the same concept behind herd immunization.

Honestly, bumpy roads? Do you think Taber is more or less of a bumpy drive than this?:

Again, that relies on every car being self-driven at once. I'm talking about that awkward first adapter phase that tech has to get through. Don't want self-driving cars to go the way of the segway do you? That was supposed to revolutionize the world too.

The computers in the satellites aboard that space shuttle probably have a lot more money invested in to their stability than what a person who is currently driving a Ford Fiesta can afford. They're also probably not turned on...

[Coach]

Originally Posted by BlackArcher101

Stop using space shuttle programs and aircraft as examples of why computers will work in this application. If you want that level of hardware, you and I nor the government will be able to afford to put that many cars on the road. Just look at the effort and testing involved into just one little piece. TCAS for aircraft, a method of communication between aircraft to prevent crashes, is a relatively simple device compared to what would be required for self driving cars. That still took 20 years to really get right.

Yeah, because technology never evolves and becomes cheaper and/or more efficient, along with smaller and more portable.

I can't carry around a computer 1000s of times faster than one that took up the space of a whole room 30 years ago in my pocket. Not at all.

Seriously do people think things just get to a point and stop?

[undercoverbrother]

Originally Posted by polak

Again, that relies on every car being self-driven at once. I'm talking about that awkward first adapter phase that tech has to get through. Don't want self-driving cars to go the way of the segway do you? That was supposed to revolutionize the world too.

The computers in the satellites aboard that space shuttle probably have a lot more money invested in to their stability than what a person who is currently driving a Ford Fiesta can afford. They're also probably not turned on...

All cars are now is computers with wheels, even a Ford Fiesta. The step to getting a car that is self-driving isn't as far as people think (IMO), the wide spread implementation of the self-driving car is farther off (IMO).

I can remember when ABS (a computer braking you vehicle for you) was a big selling point. "Does it have ABS?" "Do you need to pump your brakes with ABS?".


You make it sound like the current fleet of vehicles on the road are all Model T's.

[BlackArcher101]

Originally Posted by MattyC

Yeah, because technology never evolves and becomes cheaper and/or more efficient, along with smaller and more portable.

I can't carry around a computer 1000s of times faster than one that took up the space of a whole room 30 years ago in my pocket. Not at all.

Seriously do people think things just get to a point and stop?

No, but they certainly won't happen overnight like some people seem to think. I'm point out that something like this is going to have to be reliable and if you want the same reliablity as aircraft, you won't be getting it that quickly.

I'm NOT saying it won't happen am I?

These things take time to get right, to get tested, to get approved by government (after they all rewrite laws), then to get implemented and then revised again when something unexpected goes wrong.

Right now they are just in the technology feasibility phase. It won't be until the real serious development begins where things will get serious. Tesla is just beginning that phase and look how long it's taken, but yet we don't see wide adoption due to costs.

It's going to take a ton of finessing to get perfect (though perfection is not possible). People are saying Polaks reasoning and replies are over simplistic, I see it the other way around. He's at least thinking of the hurdles that have to be overcome, giving a sense of reality. Meanwhile the true simplistic answers and rejecting those realities and just saying everything will work based on things like ABS, traction control, blah blah blah. The complexity of having a fleet of self driving cars operating perfectly in any condition/environment is magnitudes greater than those simple technological advances, but people are thinking it's simple.

There are extremes being thrown around in this thread, from it'll never happen to it's happening now! There's a middle ground here. Yes, it WILL happen, but with the current environment of ensuring things work properly and government requirements, you can be sure it won't be very quick either. Unless of course you want to put the countries best engineers into it and have the government pour a crap ton of money into it, like the Apollo program, in order to get an expedited schedule. Even with that though, you are not guaranteed to have anything go wrong, as such with any technological advancement. Hell, we can't even get airbags or throttle linkages correct in mass production.

[Russic]

EldrickOnIce made the point I was going to make: it's not the government that'll force you car lovers to stop driving, it's the insurance companies. Once the data starts rolling in and we find that the group saying they drive better than machines are x% more likely to kill someone, those rates are going to explode.

Does it suck for those that love to drive? For sure, but a twitter joke (that I'll have to paraphrase) summed it up well:

"So there was a time when people drove their own cars?"
"Yep."
"And nobody died?"
"Oh God no, millions died"

It puts into perspective how we'll one day look back on conversations like this and be astounded that anybody didn't agree.

[Simon96Taco]

Originally Posted by codynw

And yes, I genuinely enjoy driving. Even if they perfect self driving cars, I don't want one. Driving a car is fun.

I'd like to be able to turn the feature on and off. Driving a car is not fun in stop and go rush hour traffic. I'd love my car to deal with that by itself.

Give me a twisty, empty, open mountain road, and damn you car, I'll take it from here!

[Coach]

@ BlackArcher

You were saying to stop using space-age technology as the basis for why this could/will work. I'm saying you are carrying space-age technology around in your pocket.

I don't think anyone's saying it'll happen overnight, but we're not as far off from that capability as someone like yourself makes it seem. My phone can pinpoint a Car2Go within a few feet of where it is. You're telling me that if every car used a similar GPS and were linked with each other that something like this couldn't work? Maybe it starts with as soon as you exit onto a high-speed/high-traffic road, the computer kicks in and evolves from there.

Someone such as yourself seems to use the complexity of the issue as a reason that it won't or shouldn't happen at all. It's the same issue I have with people in regards to fossil fuel vs alternatives. "It's too expensive/inconvenient to implement on a wide scale, so lets just not."

I understand people who love driving, I enjoy it myself. I don't think old-fashioned cars will ever go away, but they will end up much more as art pieces than anything else.

As someone who is 25, I can definitely see this type of thing going down in my lifetime.

[Dan02]

Originally Posted by BlackArcher101

No, but they certainly won't happen overnight like some people seem to think. I'm point out that something like this is going to have to be reliable and if you want the same reliablity as aircraft, you won't be getting it that quickly..

It wouldn't need to be as reliable as airplanes to save potentially millions of lives. 1.24 million people died in car crashes in 2010 alpine. If self driving cars could reduce that 90% would you selfishly resist the change just because you are more comfortable being in control?

[BlackArcher101]

Originally Posted by Russic

It puts into perspective how we'll one day look back on conversations like this and be astounded that anybody didn't agree.

Except no one has disagreed that they won't save lives.

[Itse]

A smartphone is probably more advanced/complicated than a spaceship, technologically speaking.

As for reliability of computers, think of the string-and-bubblegum legacy mess that is Windows. It's essentially the lowest comparison point for computer reliability. I can't even remember when my laptop last crashed. These days you can easily run one for months without rebooting.

Computers are VERY reliable now, and could easily be made more reliable if there was an actual need for it.

[BlackArcher101]

Originally Posted by MattyC

@ BlackArcher
You were saying to stop using space-age technology as the basis for why this could/will work. I'm saying you are carrying space-age technology around in your pocket.

That technology we are carrying around was developed from the space tech and the level of design that went into it wasn't that great (due to low risk). Self driving cars will not be tech revised from something else, it is a new tech and currently unproven. Developing it like new tech (such as a space program) is not feasible (due to costs), unless someone has a good amount of money (cough google cough) and are willing to take a good loss on it for a while.

Originally Posted by MattyC

I don't think anyone's saying it'll happen overnight, but we're not as far off from that capability as someone like yourself makes it seem. My phone can pinpoint a Car2Go within a few feet of where it is. You're telling me that if every car used a similar GPS and were linked with each other that something like this couldn't work? Maybe it starts with as soon as you exit onto a high-speed/high-traffic road, the computer kicks in and evolves from there.

No, GPS wouldn't work, it's not accurate enough, nor reliable. Each car will have to have a passive/active system that communicates with other cars and detect it's surrounding environment. I do like the idea of having it kick in on freeways. Perhaps temporarily they can do that to get it implemented quicker as side roads and compact streets are where the biggest hurdles will be. Add those features as DLC later

Originally Posted by MattyC

Someone such as yourself seems to use the complexity of the issue as a reason that it won't or shouldn't happen at all. It's the same issue I have with people in regards to fossil fuel vs alternatives. "It's too expensive/inconvenient to implement on a wide scale, so lets just not."

No, I'm using the complexity of the issue as a reason to why it won't happen quickly or all at once. I'm not saying it won't happen period.

Originally Posted by MattyC

I understand people who love driving, I enjoy it myself. I don't think old-fashioned cars will ever go away, but they will end up much more as art pieces than anything else.

As someone who is 25, I can definitely see this type of thing going down in my lifetime.

Sadly you are right. I just hope it doesn't happen in my lifetime. I expect it to be slowly made available, with only the rich buying it at first, then gradually being made affordable as the rich subsidize further advances to bring the cost down and reliability up. After the tech is developed fully, the last hurdle will be how it's adopted by governments throughout the world.

[EldrickOnIce]

Morgan Stanley sees sales to begin by 2026 and reach 100% market saturation two decades later, resulting in global saving of $5.3T annually - $1.3T annual savings in the US alone.
Other research firms tend to agree on the timeline, predicting 50-75% of new vehicle spending on driverless cars by 2035.

For what predictions like that are worth...

http://linkback.morganstanley.com/we...1#docId:582572

[BlackArcher101]

Originally Posted by Dan02

It wouldn't need to be as reliable as airplanes to save potentially millions of lives. 1.24 million people died in car crashes in 2010 alpine. If self driving cars could reduce that 90% would you selfishly resist the change just because you are more comfortable being in control?

Sadly this is a concept I don't agree with, the one of "Good Enough". As as designer, I've always gone with the idea of "If a persons life is at risk if this thing fails, then it damn well better work 99.999% of the time". Unfortunately risk vs monetary reward was invented and companies like GM risk peoples lives to make money.

If a major point of self driving cars is to save lives, then why not make them reliable enough so you decrease that number further to 99.99%?

[Dan02]

Originally Posted by BlackArcher101

Sadly this is a concept I don't agree with, the one of "Good Enough". As as designer, I've always gone with the idea of "If a persons life is at risk if this thing fails, then it damn well better work 99.999% of the time". Unfortunately risk vs monetary reward was invented and companies like GM risk peoples lives to make money.

If a major point of self driving cars is to save lives, then why not make them reliable enough so you decrease that number further to 99.99%?

I never stated the goal wouldn't be 100% reliability. But every tech has a starting point. So do you keep them off the road entirely or star a large scale roll out while there are still kinks but at the same time way safer then the alternative?

[BlackArcher101]

Originally Posted by Dan02

I never stated the goal wouldn't be 100% reliability. But every tech has a starting point. So do you keep them off the road entirely or star a large scale roll out while there are still kinks but at the same time way safer then the alternative?

Indeed. Which is why I hope Gen 2 is implemented on the road instead of Gen 1. Never buy a first gen

I admit though, my confidence level in the manufacturers is weak. The automatic cruise control feature on a Ford stops working with the slightest bit of dirt covering the sensor, preventing cruise control from even being able to be used. And it happens on every damn trip. Hopefully that's just an isolated case of a manufacturer cheaping out or they have now fixed those bugs.

[polak]

I also don't buy this "even if it's not reliable it'll still be safer than humans" angle.

Sure, 50 years from now, if everything goes smoothly and every car on the road happens to be a self driving car, we might have a network that can compensate for a faulty car, but even then, how do you think the world will react to the first time one of these cars has an error and slams in to a median and kills all of it's passengers? Right now when we hear about an accident, we have the nice reassurance and ability to tell ourselves that "hey, that won't happen to me, that driver was just an idiot, I won't do that". The moment that goes away you enter a much different world of comfort. Think about the fear people get from flying but now imagine you had that same feeling every time you got in your car. That's exactly what will happen if these cars crash with any sort of regularity.

So even though it might be safer overall, I think the lack of control is far more off putting than the actual statistical risk of being in an accident, much like flying.

I think you guys are kidding yourselves if you think this transition will be so smooth. We've also seen how slow the world is to adapt to infrastructure needs for new driving technology. I work for a fuel retailer and distributor and our CEO flat out laughed when someone asked about EV charging stations being rolled out in our retail network. It'd be such a great thing for the world for these electric cars to catch on, but we still don't do anything about it, why? Cause it takes a lot of benefit for us to "fix" something that isn't necessarily broken.

I personally think self driven cars will have to be perfect and completely autonomous out the gate for them to catch on.

[jammies]

Originally Posted by BlackArcher101

The complexity of having a fleet of self driving cars operating perfectly in any condition/environment is magnitudes greater than those simple technological advances, but people are thinking it's simple.

Nobody thinks it's "simple", they're saying it's *possible*, and that objections like "well what about when it SNOWS, huh?" aren't questions that haven't already been considered. If a human driver can handle a situation, then there is no reason a computerized system can't theoretically handle it - end of argument.

Unlike many fields, driving requires no creativity, it is entirely rule-based, which is something humans aren't particularly good at, and computers are. You don't consciously drive a car by deciding things using your high-end mental faculties, you do it by habit, reflex, and experience, none of which are irreducible to logic trees.

I personally doubt very much that the end system is going to use GPS or track other cars as networked entities, except as secondary or navigation systems. Each car will be autonomous and will behave much like fish in schools or birds in flocks, both of which are becoming better and better understood and reduced to algorithms. Unlike planes, you are working in only two dimensions and in restricted space, plus you have far more other objects in close proximity, so the challenges in making auto-pilots for planes are not really more than tangentially applicable.

Again, nobody is saying it'll be easy. Just not impossible due to arguments based on weather, time-limited decision making, or other purely technical challenges. The only really valid arguments are ones like I made as a joke - how do you go outside the parameters of point-to-point travel, like slowly rolling down the Red Mile honking your horn at pedestrians? What about taking a truck off-roading, or an ambulance onto the tarmac at an airport? Those are the kind of free-form, difficult to define problems that actually pose a real challenge.