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

New video of Tesla crash demonstrates the problem of semi-automated driving

theautopian.com

Post reply on HN
Print view View original

71–80 of 714 posts

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#71

Post by ehsankia »

threeseed wrote: post #33
kiratp wrote: post #25

This looks like the vehicle tried and failed to get the driver to engage and pulled over best it could. There is no shoulder there so… Let’s wait to get more info as to what the driver was doing and if they were incapacitated. The FSD beta is pretty aggressive on making sure the driver is paying attention via both steering sensors, the in-cabin camera and the touch screen.

> pulled over best it could If that is its best then FSD has no business being on the road. It was erratic and didn't give clear intent to other cars what it was trying to do.

Right, if someone is incapacitated, putting the life of that person and others on the road in danger is still not the right move. It doesn't matter what the driver is doing, any driving system that puts other people on the road in danger does not belong on the road.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#72

Post by borissk »

Sophistifunk wrote: post #11

This is a bug in the meat, and is not going away. The problems are well understood, thanks to decades of examples in semi-automated trains, boats, airplanes, and industrial monitoring systems. Until "self driving" is good enough that you're legally allowed to ride it drunk, it should not be allowed to do anything more complicated than radar cruise and lane-keeping.

It depends on what the priorities are. If self-driving is allowed now, it may lead to a few accidents, but the tech will progress faster and we will have very safe FSD cars sooner.

On the other hand the sooner we get FSD cars, the sooner we will also get ASI and with it potentially the end of human kind.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#73

Post by wpietri »

thaumasiotes wrote: post #6
loopdoend wrote: post #5

Earlier quoted context omitted.

Don’t you just tap the brake to disengage adaptive cruise?

The problem isn't that disengaging the system is difficult. The problem is that deciding whether to disengage is difficult. If you're catching up to a slower car, maybe you should disengage the cruise control. Or maybe you shouldn't. You can't tell. This means that cruise control goes from being something that relieves stress on you to something that causes additional stress.

I'm not getting why you think you need to override the cruise control. I had that concern when starting out with ACC, but I set its follow distance to something generous enough that it would trigger a speed change before I got anxious about it failing to trigger the change. That let me learn to trust it.

Although now that I think about it, I was using the Subaru ACC, so the dash has a visible indicator if it's tracking a car and maintaining speed relative to that. Perhaps yours doesn't have that?

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#74

Post by epivosism »

remarkEon wrote: post #50
epivosism wrote: post #35

The question that is never answered in this kind of story is: 1. How many fatalities per mile does tesla FSD have, compared to its best alternative if we ban it? 2. Focusing on specific failure cases is not relevant since the political decision to allow it is all or nothing. 3. It'd be perfectly logical to accept road testing FSD, even if it's significantly worse in some areas, as long as this is not exploitable, AND…

Other people have said it before, but the issue is that the Tesla crashes we are seeing are ones that a human would trivially avoid. Most people, especially non technical people, aren’t going to care about numbers that say it’s safer or better on net when the car still occasionally does things that humans view as crazy or stupid.

I'm interested in understanding why you would want every subdivision of "type of crash" to strictly improve, rather than being content that the combined crash rate goes down?

i.e. if Tesla causes 100 new crashes of type A but saves 200 of type B, isn't that a gain? Even if type A crashes are one which humans don't normally make.

A way in which I agree with you is if type B crashes are more avoidable by conscientious drivers, and you feel that your personal base rate of B is therefore already low, and that introducing new type A risk is actually NOT an all-in improvement for you.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#75

Post by BoorishBears »

calvinmorrison wrote: post #62
BoorishBears wrote: post #30

Earlier quoted context omitted.

If it wasn't a Tesla the driver would probably be sitting in jail right now for intentionally causing a wreck. But because Tesla's pretend attempt at self-driving engaged in the reckless driving, the driver is protected despite clearly not reacting in anything even slightly resembling a timely manner. Disclaimer: I work in the AV space... but no one in the space should consider Tesla a competitor.

[flagged]

Wouldn't it be easier to stay quiet since you don't have a meaningful rebuttal?

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#76

Post by nanidin »

judge2020 wrote: post #59
nanidin wrote: post #38

Earlier quoted context omitted.

> if they were incapacitated. You can see them get out of the vehicle and walk around a bit in the video[0], so I guess they were not incapacitated - see around the 46s mark in the video. The driver also claimed they were using FSD at the time the car performed the maneuver. [0] https://www.reddit.com/r/teslamotors/comments/108kpgo/footag...

another angle: https://twitter.com/kenklippenstein/status/16128488720611287... The driver was on I-80. Unless this is a Tesla employee, Tesla FSD is not enabled on interstate highways at this time; when you're on a major no-stops no-traffic-light highway, it defaults to Autosteer+TACC, with his subscription likely only enabling the auto-lane-change feature. All this means is that the likely culprits are either TACC s…

FSD vs navigate on autopilot is a pretty small technical detail on the highway. Once you turn on FSD beta, from the driver's perspective FSD and autopilot are the same thing - turned on with the same input (double tap on the right stalk), with the same stuff shown on the display.

Navigate on autopilot can change lanes on the highway and follow highway exits without user intervention.

IMO when someone says they're using "FSD" they mean they're subscribed/paid for FSD, have the FSD beta turned on, and are letting the car drive itself (whether that's FSD on city streets, or navigate on autopilot on the highway.)

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#77

Post by btilly »

Johnny555 wrote: post #37
marktolson wrote: post #29

Earlier quoted context omitted.

My friend was using adaptive cruise control while towing a caravan. A car merged in front of him, his car applied heavy braking to maintain the desired distance (not necessarily to avoid a crash) but didn't compensate for the weight of the caravan. Almost lost traction, lead to heavy swerving. He's vowed to never use it again.

Doesn't his car have ABS? Even under heavy braking, the car's wheels shouldn't lock up and lead to swerving. Did the trailer have its own brakes?

As everyone is pointing out to you, ABS solves the problem of your car's wheels skidding, leading to loss of control.

It doesn't help with the trailer that you're towing failing to stop as fast as you do, trying to run into you, and trying to turn to the side when it can't.

The solution to the latter problem is to be aware that you have the trailer, and to not brake faster than it can handle. The fact that the trailer makes it harder for you to avoid an accident is exactly why many jurisdictions put extra restrictions on cars towing trailers (for example in CA they aren't supposed to go over 55 mph - even on the freeway).

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#78

Post by borissk »

threeseed wrote: post #60
Sophistifunk wrote: post #11

This is a bug in the meat, and is not going away. The problems are well understood, thanks to decades of examples in semi-automated trains, boats, airplanes, and industrial monitoring systems. Until "self driving" is good enough that you're legally allowed to ride it drunk, it should not be allowed to do anything more complicated than radar cruise and lane-keeping.

This is a bug in Tesla's specific meat. Other self driving implementations e.g. Cruise, Waymo have LiDAR which is an additional and in my opinion necessary sensor for detecting objects. Because in this case it sure looks like a phantom breaking issue.

Yeah, it looks like one of those cases where Elon is being stupid - self-driving cars with LiDAR are a lot safer.

Anyone know how much a LiDAR system cost nowadays?

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#79

Post by sokoloff »

denimnerd42 wrote: post #23
herodotus wrote: post #2

I have a VW Golf and it has what it calls "adaptive cruise control". What makes it different from my older (simple) cruise control is that it will slow down and speed up again as necessary without my intervention. For example, if I set the speed to 70 mph and it gets close to a car ahead that is only going 65 mph, it will slow down and maintain what the software believes is a safe distance. Similarly, if a car in an…

I do love the ACC on my car. One of the worst parts of road trips is following cars you don't want to or can't pass. The ACC totally solves it. I don't see how you were really able to use cruise to follow another car at all before? Even 1mph too slow is not great and if the car in front isn't using any cruise control at all it could be 5mph too slow and then 5mph too fast.

I just did a 1500+ mile road trip over the holidays. During the drive, I was trying to figure out why old-tech (2005) cruise control worked so well.

Even a 1 mph differential in speed is a car length change in following distance every 10 seconds. A 1/2 mph difference is 3 car lengths of change every minute. That seems like enough that would be annoying, but it obviously wasn’t when two cars are following each other both on cruise control.

I still don’t have a fully satisfactory explanation for something that I can easily observe (that it’s usually not annoying).

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#80

Post by threeseed »

ASalazarMX wrote: post #69
threeseed wrote: post #33

Earlier quoted context omitted.

> pulled over best it could If that is its best then FSD has no business being on the road. It was erratic and didn't give clear intent to other cars what it was trying to do.

IMO the car signaled, changed lane and stopped without slamming the brakes. It didn't collide with anything, the pileup was formed by human drivers driving too close and too fast for their reaction time. If it had been a human driver that saw a kitten or a rock in the road, it would probably slam the brakes. I'm not defending Tesla's FSD, which I think is inferior for Tesla's reluctance to use LIDAR in addition to ca…

> IMO the car signaled, changed lane and stopped without slamming the brakes

Which is erratic and reckless.

If it was an emergency situation it should have put on hazard lights, waited until all surrounding cars passed, then changed lanes and very slowly decelerated until it stopped.

All while aggressively alerting the driver what is happening.

Post reply on HN