I think safety is fine, especially in cities with not too crazy weather.

It's more that if an autonomous car with no human stalls, due to a traffic light that's stopped working, or an oddly shaped plastic bag in the road, it's stuck, and would cause a traffic jam. Not sure how to resolve such a scenario, can't rely on someone remotely taking control, as mobile data doesn't have 100% coverage, so a human who is allowed access to the car would somehow need to get thorough the traffic jam to move the vehicle on. Can't think of an efficient way to resolve this situation.

First, some baselines: humans suck at driving too. Next, coverage isn't an all or nothing thing, it'd be quite easy for an autonomous uber like service to restrict itself to areas that permit remote control when cars need assistance. Last I'd encourage reading some of the literature behind what folks are doing: it's a very different sort of system, one that blends a lot of complex probability estimates. It's not like everyday coding where a human developer needs to anticipate every combination of corner cases.

Machine learning tools also cannot cover corner cases, if they're not present in it's training data. Communication with human agents (other cars, police officers) is also extremely tough.

It's not as if human driven cars never get stuck in an intersection. I imagine the solution will be much the same: it will be moved manually if possible or towed if not. If a person is deliberately fucking with traffic by deliberately messing with an autonomous car, they'll also probably be fined.

I mean the other day on my daily commute there were two cars right in the middle of a freeway, one attempting to jump start the other and they were at it for a while. This is not shockingly new ground.

But nobody's asking you to use any of these systems yet. they're in development, not products for sale. There's no benefit to anybody in questioning whether google knows that snow exists, or if uber understands that traffic stops unexpectedly sometimes. Obviously everybody involved in developing self-driving cars understands the most basic definition of the problem.

If google actually starts selling a self-driving car and doesn't give any information about whether it can deal with a policeman-controlled intersection, then you can be concerned. Until then, implying their self-driving car programs haven't considered the basic problems that any uninformed commenter on the internet can think up is not incredibly helpful.

It only needs to be a better driver than you for you to be safer, and its that already.

I can guarantee you that on a one to one driving skills shootout between you and the software you will loose 98% of the time.

That may well be the case, but it does not follow that a 98%-complete self-driving car would actually be a viable product. It's not good enough that the robot can beat a human in a head-to-head skills test; the robot also has to be as versatile as a human, across an unpredictably wide range of circumstances. That's hard.

If a autonomous car can pass a standard driving test it should be allowed to drive. They always do, every time.

Depending on where you live, "can pass a standard driving test" isn't necessarily saying very much.

In any case, I'm not talking about whether it "should be allowed to drive", I'm talking about the level of capability an autonomous car would have to reach in order to become a viable product.

No its not, that is kind of my point. It only needs to be better at driving than the average human to be technically viable. And yes, it may very well be allowed on Indian or Chinese roads before its allowed in America, but that's human fiction at work not demonstrable ability.

I'm not talking about technical viability! I'm talking about product viability. Can this hypothetical device be manufactured and sold for a profit? Can it be used, profitably, for some purpose, for which it will be cheaper or in some way preferable to existing solutions? It is not enough that this autonomous car can drive around the block without crashing; it has to be good enough, reliable enough, efficient enough, that you can build a profitable business around the unique service it can provide. That is much harder than simply passing a standard driving test.

Yes of course it is enough, if that is your definition of 'viable product', that should be self-evident. It only needs to be better than a human at driving for it to be a viable product. Are you prepared to pay the insurance premium on a non-autonomous transport appliance in 10 years time ? I'm betting you won't.

"Driving" is not one single monolithic problem. "Driving" is a whole array of complicated problems. The "standard driving test" exercises a very limited subset. "Better than a human at driving" is a very broad and poorly-specified claim.

Hell yes I will still be using a "non-autonomous transport appliance" in ten years - it's called a "motorcycle" :) - and I don't want your high-tech self-driving future if it has no room for that.

Clearly 'driving' to you is an expression of personal freedom - not body transport. And that's good and well, but finding unrelated justifications to defend what is essentially a hobby is silly.

People will continue to spend stupid amounts of money on hobbies. I'm not going not hand in the keys to my Landcruiser or my CRF250. But I don't leave those parked at the train station as a rule either.

You asked me a personal question, so I gave you a personal answer, somewhat flippantly as I disagree with your premise.

That remark was separate from and unrelated to the actual point I've been trying to make: there is a very significant gulf between the notion of a "self-driving car" as a form of automation built into a vehicle which is still under the control and responsibility of a human driver, and the notion of a "self-driving car" which is a fully autonomous robot built in the shape of a car, that must be capable of doing something reasonable in all situations without depending on a human. The latter is a much harder problem than the former, and we don't yet know whether it can be solved in a way that yields a practical, profitable transportation system of the sort many people like to fantasize about when they discuss (for example) Google's autonomous vehicle program.

Seeing as there are already people trying to use their Tesla autopilot as autonomous capabilities, I'm pretty sure they'll be viable as soon as they are road legal.

Viable for what, exactly?

It is very clear that autopilot will be a successful feature and we can reasonably expect continued improvement in human-driver-assist technologies for years to come.

But people keep on handwaving past the very important gap between "automation which assists a human operator" and "autonomous vehicle which gets around on its own", as though doing well at the first problem necessarily makes the achievement of the second problem inevitable.

Driver assist technologies make all kinds of sense but they do not fundamentally change the nature of our transportation systems. They just make it easier to do a better job as a human driver.

Getting to the point where most wheeled vehicles are self-piloted autonomous robots with no human driver interaction...? I still think that's a really hard problem.

>Viable for what, exactly?

Sales. That's what makes a viable product. One that will sell. Despite elsewhere in this comment chain you saying "I'm talking about product viability not technical" I can't help but think that your response to me was about technical viability.

People are already willing to buy and trust their cars to drive for them right now, when they're specifically being told "don't trust it to drive for you". The second it's legal to advertise "let it drive for you!" autonomous cars will sell regardless of their ability to perform well in unregular circumstances.

If "self-driving car" just means "I'll drive for you, most of the time, while you sit behind the wheel", then we are talking about nothing more than a better automobile. Of course that will sell, but in that case the world of the future looks pretty much the same as the world of today, with fewer crashes.

But people keep talking about "self-driving car" as though it is going to be a robot capable of running errands on its own, with no human driver and not even necessarily any human passenger. The question is still open: can this be done well enough, enough of the time, solving a valuable enough class of problems, that producing such car-shaped robots will actually become a significant industry? Furthermore, will the problems solved by such hypothetical car-shaped robots be significant and widespread enough to justify the hype currently being lavished on them?

Meeting the hype expectations of people who practically worship technological change is a pretty high and imo unreasonable bar for viable product.

Personally I'm of the opinion that because we've been making tools for 2.5 million years and don't seem any happier than the average animal, the very concept of a new tool making all that much of a difference is absurd.

I agree that autonomous cars are happening, but there are a ton of assumptions about human capabilities embedded in a standard driving test. It's not nearly rigorous enough greenlighting autonomous vehicles.

Sure, but put a tumbleweed in front of the driverless car and it's stuck. Or wood with a nail sticking out/generic movable obstacle, a human can remove it, the auto-car cannot.

Yea so ? Harping on the 2% issues does not negate the benefit you have the rest of the time: It can swerve in microseconds, you cannot. How many times will your nail removal skills save you or your kids live compared to software that knows that rolling the car to the left have a 87.34 better chance of occupants survival that the head-on collision with the truck that just swerved in 0.4 seconds ago ?

Forget it, driver-less s cars will be the norm in the future. You won't be able to afford the insurance associated with a meat-driven car.

My nail removal skill will prevent gridlock on the road. If an auto-car gets confused due to a 2% issue, it becomes a brick, clogging the road. If multiple auto-cars do this in a city, the result is standstill.

You make it sound like people are gonna be strapped into their autonomous vehicle like Hannibal Lector. Seriously, a couple pieces of debris are not some thing where we go "oh crap, we're done, guess the whole city will just sit here all day."

Can it not drive around the tumbleweed? Or, if it's completely walled in by them, make an assessment of their density and likely harm caused, and decide to just run over them?

If we're crediting the car with being able to handle merging into a crowded multi-lane highway and dealing with idiot human drivers who don't indicate or hold lane position very well, it seems a bit odd to arbitrarily presume that it won't be able to identify non-hazards like said tumbleweed.

Yes, a computer drives better than me. But that's not it's competition.

Does it drive better than me driving while being supervised continuously by a computer? I still make the important decisions, and my mistakes are pointed out by the computer.

The competition for self driving cars are not humans, but humans being adviced by driver assistance programmes

No sure why you make that distinction, or what the benefit of that will be. I want to read HN on the back seat on the way to work, not "supervise" the driving appliance.

Op has a good point. You might want to read the web in the back seat, but if you supervising the computer makes it safer, will the public let you doze in the back seat putting your life and the lives of others at increased risk?

If I have to supervise, there is no point. Having a human intervene in what should be 100% automated is putting my life and others at risk. Look at road death statistics: the vast bulk of incidents are are caused by humans sucking at driving - why should anyone want to continue doing that ?

I would say that humans should drive and computers should supervise and correct obvious mistakes, that however is status quo right now. It doesn't sell as futuristic fiction

Well, one way to solve the problem is to simply bring the car to a halt, then wait until the problem is solved by itself (or by the user).

And if you're going 70 down a busy interstate?

The number of times that "you can't stop / brake / slow because someone behind you will run into you" comes up in these threads is always alarmingly high.

Indeed, it reminds of this article, which thankfully didn't end up killing anyone:

https://www.wired.com/2015/07/hackers-remotely-kill-jeep-hig...

> Miller and Valasek’s full arsenal includes functions that at lower speeds fully kill the engine, abruptly engage the brakes, or disable them altogether.

Are there seriously cars on the road with complete fly-by-wire brakes which have no pure mechanical backup? That's horrifying. How did we go from dual redundant hydraulic systems that work (if in a reduced capacity) even when the ignition is off, to something at the mercy of some code that can be bamboozled over 3G?