Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Alphabet's Self-Driving Cars to Get Their First Real Riders

bloomberg.com

Post reply on HN
Print view View original

261–270 of 411 posts

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#261

Post by ajross »

jonmc12 wrote: post #146

Am I the only one feeling the press + big tech narrative of "driverless vehicles = safety" is way ahead of itself? OK, "driver error is blamed for 94% of crashes", but what is the validation strategy to show where driverless cars can quantifiably improve? In 2016, Philip Koopman of CMU discusses that there are non-trivial engineering challenges of achieving safety in NHTSA Level 4 vehicle automation: https://users.ec…

> what is the validation strategy to show where driverless cars can quantifiably improve?

That seems like an odd question to ask. There are huge reams of traffic safety data collected every year by the NTSB and others. Traffic accidents are one of the top two or three treatable public health issues in the modern world, and they get very significant public funding for their study.

Honestly I think the question has to be the reverse: what is is about "driverless" safety data that makes you think it won't be well-measured by the existing "validation" regime?

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#262

Post by dorfsmay »

loup-vaillant wrote: post #215
stevenh wrote: post #205

Earlier quoted context omitted.

Imagine ransomware infecting your car that gives you 30 minutes to pay or you'll be rammed into a tree at 90mph. I don't care about any contrived and well-funded narratives about how unthinking, unfeeling, driverless cars will somehow save lives. No thanks, I'd rather take my chances continuing to share roads with human drivers who are careful because they fear their own death just as much as I do.

The correct answer to that is very simple, and very expensive: just make correct software. Such malware are only possible because the targeted system has vulnerabilities to begin with. One just has to ensure the absence of such vulnerabilities, possibly using machine checked proofs. One obvious approach is to properly isolate the driving software and sensors from external input.

+ a mechanical override.

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#263

Post by monkmartinez »

MarkMc wrote: post #238
stevenh wrote: post #205

Earlier quoted context omitted.

Imagine ransomware infecting your car that gives you 30 minutes to pay or you'll be rammed into a tree at 90mph. I don't care about any contrived and well-funded narratives about how unthinking, unfeeling, driverless cars will somehow save lives. No thanks, I'd rather take my chances continuing to share roads with human drivers who are careful because they fear their own death just as much as I do.

Check out this graph about airplane accidents: http://cdn.static-economist.com/sites/default/files/images/2... Since 1970 total airplane casualties have halved, while there are now 7 times the number of passengers. Much of this improvement has been due taking decisions away from pilots and into the hands of 'unfeeling' automated systems.

And look at the cost of planes and those systems. Absolutely out of reach for 97% of the world.

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#264

Post by regulation_d »

jonmc12 wrote: post #146

Am I the only one feeling the press + big tech narrative of "driverless vehicles = safety" is way ahead of itself? OK, "driver error is blamed for 94% of crashes", but what is the validation strategy to show where driverless cars can quantifiably improve? In 2016, Philip Koopman of CMU discusses that there are non-trivial engineering challenges of achieving safety in NHTSA Level 4 vehicle automation: https://users.ec…

I feel like the question is not if "driverless vehicles = safety" but when. So, all other things being equal, the quicker we get to that point, the more lives we can save.

Obviously not all other things are equal, and potentially we could create greater loss of life by attempting to get self-driving cars on the road too quickly. But that is the nature of problem: how do we avoid moving too quickly while recognizing that the current danger that is human drivers should be removed from the equation as quickly as feasible.

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#266

Post by nommm-nommm »

stevenh wrote: post #205
jonmc12 wrote: post #146

Am I the only one feeling the press + big tech narrative of "driverless vehicles = safety" is way ahead of itself? OK, "driver error is blamed for 94% of crashes", but what is the validation strategy to show where driverless cars can quantifiably improve? In 2016, Philip Koopman of CMU discusses that there are non-trivial engineering challenges of achieving safety in NHTSA Level 4 vehicle automation: https://users.ec…

Imagine ransomware infecting your car that gives you 30 minutes to pay or you'll be rammed into a tree at 90mph. I don't care about any contrived and well-funded narratives about how unthinking, unfeeling, driverless cars will somehow save lives. No thanks, I'd rather take my chances continuing to share roads with human drivers who are careful because they fear their own death just as much as I do.

You're not looking at the whole picture here. What's your odds of being killed by a drunk/inattentive/inexperienced driver vs odds of being killed by randomware? Road fatalities are already the number 1 killer of some age groups, its estimated that over a million people worldwide die on the roads every year. Around 40,000 in the US last year. If we trade 40,000 road deaths for 10,000 ransomware deaths, it's a massive improvement.

Ignoring that, that ransomware wont' even exist for a number of reasons, the biggest being it wouldn't be profitable. Ransomware authors aren't going to get a payout, very few people are going to be able to buy Bitcoin in 30 minutes while unable to leave their car even if they wanted to. Not only that, it isn't remotely difficult to have a manual, physical, kill switch that physically stops the car in a life or death situation. Even if you didn't have a kill switch, you could simply call the police and they can lay down a spike strip.

>who are careful because they fear their own death just as much as I do.

LOL! That's a good one! Have you actually even been on the roads before!?! When I worked drive thru I had people who were so drunk they could hardly form a sentence come through my drive thru. That's just ONE example.

You share the road with these upstanding citizens who are clearly very careful and concerned for their own wellbeing: https://www.youtube.com/watch?v=DcnuIWNv8lw https://www.youtube.com/watch?v=9hddt4bWNns

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#267

Post by saosebastiao »

teach wrote: post #240
saosebastiao wrote: post #169

This is an important step for testing, and I'm surprised that this is being spun as a move towards commercialization. As of right now, waymo has only been doing what could metaphorically be called unit testing. That is, they test the cars behavior in very controlled but unrealistic environments, looking for very specific responses. The accident rate that they've incurred is likely ridiculously skewed: they've been dr…

> they test the cars in very controlled but unrealistic environments Are you sure about this? Google has had self-driving cars in Austin for years and I'm pretty sure there's never a good time to be driving here.

Pretty damn sure. Even if an engineer drives his normal car to work, hops in a waymo car, and drives 9-5, and then drives a normal car back home, they've already constructed a scenario that is heavily time-weighted towards lighter than human-representative traffic experiences. Even small biases that you don't even think are meaningful can have a big impact on outcomes.

But it doesn't stop there. Google hasn't tested snow at all. I'd be surprised if their testing locations represented rainy weather that is representative for more than 10% of the US. They don't test freeways, and by extension, do not test onramps and offramps. In California, they don't even go over 25 mph. Under those conditions they have been ostensibly pretty good at not causing accidents, but how good are they at avoiding accidents that are not their fault? This means accidents caused by other drivers and acts of god. Safety is far more than just causality.

But more importantly, their miles traveled very easily exclude small snippets of time where an engineer takes over, but still counts the rest of the journey where the computer did fine. How representative is that? How good are humans when we let jesus take the wheel the moment driving gets slightly complicated? If a human overrides the computer only 1% of the time, that could still be anywhere from 1% to 100% of trips with at least one human override.

It's very easy to catastrophically mislead yourself when working with averages [0]. Statisticians and Economists go through absolutely absurd lengths to make comparisons robust. Even if we had access to 100% of the data on Waymo's self driving cars, I don't know a single statistician that would be comfortable making a comparison to extremely coarsely aggregated data on human driving. Only through extensive testing in situations that are representative of how humans (and not just test engineers) drive will we be able to conclude that one is safer than the other. And that's why Phoenix is important to Waymo.

[0] https://www.google.com/search?client=ubuntu&channel=fs&q=the...

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#268

Post by michael_storm »

smlacy wrote: post #250
jonmc12 wrote: post #146

Am I the only one feeling the press + big tech narrative of "driverless vehicles = safety" is way ahead of itself? OK, "driver error is blamed for 94% of crashes", but what is the validation strategy to show where driverless cars can quantifiably improve? In 2016, Philip Koopman of CMU discusses that there are non-trivial engineering challenges of achieving safety in NHTSA Level 4 vehicle automation: https://users.ec…

Your argument is about the same as "Humans can calculate arithmetic perfectly accurately, why do we need computers to do it for us?" There are a host of obvious reasons why even a "simple" L4 car will beat a human hands down: * Reaction time. * 360 awareness / visibility. * Lack of exaggerated "Human Reflex" to surprising events (i.e. swerving violently to avoid a dog, and hitting other vehicles) * Keeping perfect sp…

There are a host of obvious reasons why even a "simple" human will beat an L4 car hands-down:

* Lack of software bugs due to careless programming.

* Millions of years of brain evolution allows adapting to unforeseen circumstances.

* Vastly decreased likelihood of being tricked by vandals exploiting flaws in recognition algorithms.

* Inability to be hacked en masse and driven into each other/houses/pedestrians.

* Much harder to back-door.

* Can be judged in court if they swerve to avoid a pigeon and hit a crowd of pedestrians.

I could go on, but this should drive the point home pretty clearly, IMHO.

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#269

Post by Ajedi32 »

plopz wrote: post #255
Ajedi32 wrote: post #199

Earlier quoted context omitted.

> Will car ownership be killed with self-driving cars? Probably. It seems way more convenient and efficient to be able to summon a car to any location on demand than to have just one car that you use for everything. No more parking, no more refueling, no more vehicle maintenance; just push a button and a car shows up within minutes (or in some cases, seconds) to pick you up, and step out of the vehicle when you reach…

How do you solve the issue of extreme demand M-F at 8am and 5pm and low demand any other time?

Same way any other business that has brief periods of high demand (electric companies, ISPs, restaurants, etc) does; you anticipate the demand and make sure you have enough capacity available at those times to meet it.

With a self-driving car fleet there are probably lots of other little tricks you can do too, like offer discounts for sharing a vehicle with another passenger, or use machine learning to predict what locations will have the highest demand during particular parts of the day and move portions of the fleet to cover those areas.

Re: Alphabet's Self-Driving Cars to Get Their First Real Riders

#270

Post by aetherson »

smlacy wrote: post #250
jonmc12 wrote: post #146

Am I the only one feeling the press + big tech narrative of "driverless vehicles = safety" is way ahead of itself? OK, "driver error is blamed for 94% of crashes", but what is the validation strategy to show where driverless cars can quantifiably improve? In 2016, Philip Koopman of CMU discusses that there are non-trivial engineering challenges of achieving safety in NHTSA Level 4 vehicle automation: https://users.ec…

Your argument is about the same as "Humans can calculate arithmetic perfectly accurately, why do we need computers to do it for us?" There are a host of obvious reasons why even a "simple" L4 car will beat a human hands down: * Reaction time. * 360 awareness / visibility. * Lack of exaggerated "Human Reflex" to surprising events (i.e. swerving violently to avoid a dog, and hitting other vehicles) * Keeping perfect sp…

Quick, what's the reaction time for Waymo's driverless cars? Support your answer with evidence.

For extra credit: describe what the reaction time is for Waymo's driverless cars are during various points of the garbage collection cycle.

People here have a tendency to imagine an idealized version of driverless cars and assume that actual driverless cars are the idealized versions of those cars, and then compare the idealized cars to very non-idealized humans.

I think that the real world is going to be more complicated than that.

It is easy to write code that takes a non-trivial amount of time to process to the point of output on any given hardware. I imagine that all of us have done that at various times. We can certainly imagine that reaction time could be a virtue of driverless cars -- even 99th percentile reaction time. Given sufficient hardware (and not just hardware on the core CPU/ram chips -- this is hardware out to the sensors and the subprocessors associated with each of them).

Similarly, we could imagine putting enough sensors on the vehicle that it truly has 360 degree awareness and visibility. But we do see real cars in the real world that have surprising blindspots.

There are cost tradeoffs to all of these things, and complexity tradeoffs. Is there a happy medium? I think that there very likely is, even without any fundamental advances in the state of the art. Are we within a year or two of that happy medium? Maybe.

Post reply on HN