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

Let’s Talk About Self-Driving Cars

medium.com

Post reply on HN
Print view View original

181–190 of 341 posts

Re: Let’s Talk About Self-Driving Cars

#181

Post by alkonaut »

wrsh07 wrote: post #149
alkonaut wrote: post #118

Let's talk about how we can't make nearly anything "smart" in the world of "IoT" and "smart things". We don't even have a decent robotic lawnmower in 2017 but we want driverless taxis within 10 years? I'm sceptical.

In your opinion, how many orders of magnitude more difficult is driving than mowing the lawn? I honestly don't have a strong sense of the scale of ability here.

level 3-4 driving maybe just a couple of magnitudes harder (e.g in terms of computing power and the amount of code required).

Full autonomous Uber's - so many orders of magnitude that I don't even want to speculate on magnitudes.

I'm pretty sceptic so I put "mostly autonomous vehicles" within reach, but "driverless taxi" several decades away. We could make "autonomous safe regions" on the map where the mostly autonomous cars could work well. But we just can't make them work well in the normal world for humans because of the edge cases.

The last fraction of a percent of scenarios require humans or something very nearly human. You need to be able to improvise, learn, read half-hidden text on road signs and deduce from context what it means etc.

It's not acceptable to just handle 99.9 % of the situations for a driverless car, then you end up with driverless cars stranded everywhere.

I think the last step from "mostly autonomous" to "fully autonomous" is comparable to going from an ai recognizing all the objects in the famous funny Obama-scale photo, to having an au explain why it's funny. We don't even fully understand that problem, much less have any idea of how to solve it.

http://karpathy.github.io/2012/10/22/state-of-computer-visio...

Re: Let’s Talk About Self-Driving Cars

#182

Post by JamesBarney »

sebastos wrote: post #120
JamesBarney wrote: post #64

Earlier quoted context omitted.

Yes. Lets assume a. 1 fatality per 100 million miles driven as an average driver[0] b. .5 fatalities per 100 million miles driven as robo driver c. .25 fatalities per 100 million miles driven as 20-70 sober well rested driver I'm 31yr old male so I have an average life expectancy of 47 more years. That means I lose about .5/100 million * 47 yrs * 365days * 24hrs * 60mins * 60s = 7.4 seconds life lost per mile in robo…

Okay, I've designed a new pill. 1 out of 4 people have adverse reactions to its contents and die instantly, but people who don't live long virile lives until, on average, the age of 100! Your expected gain would be nearly 5 additional years of life! Should I sign you up?

These years aren't comparable because I'm trading years in the immediate future for years in the distant future. Not many people would trade next year for a year and 2 months when they're 80.

Re: Let’s Talk About Self-Driving Cars

#183

Post by random_upvoter »

stale2002 wrote: post #163
random_upvoter wrote: post #66

Earlier quoted context omitted.

When a piece of software crashes out of its model space, it becomes infinitely stupid. A drunk driver, while impaired, retains a level of understanding of the situation but a program has nothing to fall back on. I suspect that accidents with automated driving, while rare, will tend to be more severe. Think train wrecks.

Not really. If something goes wrong, 99.9% of the time, the solution is to slam on the brakes, and maybe swerve a little to avoid an immediate obstacle. If a self driving car only does this, it will be a way better driver than a human. Talking about weird edge cases misses the point. Edge case = slam on the brakes.

You assume that the program will recognize the edge cases. But will it recognize the edge cases of recognizing the edge cases?

Re: Let’s Talk About Self-Driving Cars

#184

Post by potatolicious »

Tiktaalik wrote: post #125

No mention of pedestrians, cyclists, or how cars (self driving or otherwise) affect the urban realm. So much of the self driving discussion seems to be wholly about the automobile technology in a bubble, with near zero thought put into the impacts on people, cities and society. The fact that the idea of signal less intersections, a concept that ignores the existence of pedestrians and cyclists entirely, is even a poi…

I agree with your point - though I'd also say that people who actually working in the field generally are considering these sorts of issues and are not blind to them.

The "OMG only cars will exist and everything will be amazing because pedestrians are a fictional concept" crowd overwhelmingly seem to be wide-eyed futurists who don't actually have professional first-hand exposure to the field.

Re: Let’s Talk About Self-Driving Cars

#185

Post by _archon_ »

alkonaut wrote: post #115
MR4D wrote: post #99

Earlier quoted context omitted.

Imagine some drunk guy hops into his car in bad weather. He taps the auto-cruise button and then passes out. Snow, rain, construction, - something - hinders the ride somewhere along the way home. Car cannot process the issue correctly, and slows to a stop, alerting the driver. Driver, being sound asleep, cannot hear the annoying alert tone. Alert continues, and possibly after some time, alerts authorities of it's dis…

Until cars are reliably at the highest level of replacing humans along the entire way of any drive (I.e cars don't have drivers seats) then obviously it must be a criminal offense to be the "driver" while drunk, even if the driver was hoping to sleep all the way home. What has happened here is really better: the drunk guy is asleep in a stopped car, rather than asleep in a car that just hit another car.

Interestingly, he can still be arrested for DWI. This would be an interesting case, though, because the mens rea is clearly not present; the human intended to be conveyed by a capable transportation system, and didn't intend or attempt to control a vehicle while in motion. I look forward to the result of this case for legal precedent reasons.

Re: Let’s Talk About Self-Driving Cars

#186

Post by alkonaut »

Joeri wrote: post #156
alkonaut wrote: post #102

Earlier quoted context omitted.

I think we will soon be in a situation where self driving cars are better than humans for 10% of driving scenarios(highway, gridlock commute, some parking, ...) but nearly useless in others. I think the acceptance for half decent AI will be much lower than it is for poor human drivers, and the hardest parts of autonomous driving is still a very long way out. That's probably fine though, if the automation can reliably…

Ubers and transport trucks can do fixed routes, so that should be relatively easy to automate.

Driving a known "autonomous safe" route, or map region, and simply not driving if the conditions aren't ok (it's snowing, there is construction in the area so maps or lanes are poor etc) works. But that's basically taking the "mostly autonomous"-level driving, and just adapting either the environment and/or the restrictions/schedule/routes to that.

Uber could probably also just have defensive systems that simply park the car and wait for a human driver to arrive, if there is a problem. It's good business for them to have 3 cars per driver. It still doesn't approach the interesting problem though, which is autonomous driving on a human level (Autonomous Uber without backup drivers, in an entire city). That I think is far away.

Re: Let’s Talk About Self-Driving Cars

#187

Post by endentru »

knz wrote: post #29
lg wrote: post #11

Earlier quoted context omitted.

if self-driving cars will be as safe as people say, manual driving should have the same legal status as drunk driving due to the unnecessary danger it creates.

It's a false comparison though. 1) Society tolerates a small amount of alcohol for most drivers. 2) Driving while drunk impairs your physical and mental ability to drive safely and only gets worse as you drink more. Obviously a manual driver has slower reaction times than an autonomous system but they are within a level that society tolerates and far better than someone who is drunk. There are also risks with autonom…

> easy to become reliant on the technology and have a serious accident when a sensor fails (particularly for blind spot detection)

Recently realized this myself when the blind spot detector light went out in our SUV. It's gotten me thinking about how manufacturers would approach a situation like that without forcing the user to undergo (or more accurately ignore) safety tests like a walk-around prior to driving or something similar. Or perhaps the intended use of these sensors should be stated as supplementary - ie "do not rely on them without verifying it's safe to merge yourself"? People would ignore that too. It's an issue I haven't really seen a solution to anywhere.

Re: Let’s Talk About Self-Driving Cars

#188

Post by subpixel »

pif wrote: post #25
francoisdevlin wrote: post #13

Earlier quoted context omitted.

What are you concerns? What should engineers in this space be considering?

Copying form an old comment of mine: - Yielding to an emergency vehicle with sirens on. - Moving backwards to a safe and large enough spot when the route is too narrow to fit self-driving car and oncoming huge lorry (and there is no line marking the limit between road and ravine). - Upon instructions from authority, recognize that the highway is closed due to an accident and, no matter what the driving code says, you…

I'd add:

- disincentivizing investment and innovation in public transit

Re: Let’s Talk About Self-Driving Cars

#189

Post by CPLX »

_archon_ wrote: post #185
alkonaut wrote: post #115

Earlier quoted context omitted.

Until cars are reliably at the highest level of replacing humans along the entire way of any drive (I.e cars don't have drivers seats) then obviously it must be a criminal offense to be the "driver" while drunk, even if the driver was hoping to sleep all the way home. What has happened here is really better: the drunk guy is asleep in a stopped car, rather than asleep in a car that just hit another car.

Interestingly, he can still be arrested for DWI. This would be an interesting case, though, because the mens rea is clearly not present; the human intended to be conveyed by a capable transportation system, and didn't intend or attempt to control a vehicle while in motion. I look forward to the result of this case for legal precedent reasons.

For what it's worth DWI and its variations are usually strict liability statutes.

Re: Let’s Talk About Self-Driving Cars

#190

Post by shuntress »

MR4D wrote: post #99
CPLX wrote: post #19

I find it odd the way people talk about moving through the levels sequentially without really acknowledging there's a massive phase change that happens somewhere between level two and four, where a self-driving car could easily be way more dangerous than the average human driver. That comes as soon as you have a situation where the automation has to be able to handle all foreseeable scenarios without failing because…

Imagine some drunk guy hops into his car in bad weather. He taps the auto-cruise button and then passes out. Snow, rain, construction, - something - hinders the ride somewhere along the way home. Car cannot process the issue correctly, and slows to a stop, alerting the driver. Driver, being sound asleep, cannot hear the annoying alert tone. Alert continues, and possibly after some time, alerts authorities of it's dis…

Imagine a world where theres no self-driving cars and that same drunk guy on that some night hops in to the car and tries to drive home.

Edit: I would gladly accept more failure in uncommon edge cases if it means reduced failure in normal operation (ie: drunk guy slams in to a house at 1 AM on a clear calm night)

Post reply on HN