Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other • Hacker News Board

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#81

Post by Volundr » Sun, Apr 06, 2025, 3:03 PM UTC

Gud wrote: post #53

energy123 wrote: post #30

Earlier quoted context omitted.

Multiple independent sensor streams is like having multiple pilots fly a plane instead of just one pilot. The chance of a fatal error (false negative identification of an object near the car) decreases from p to p^n. The size of that decrease is not intuitive: if p=0.0001 it now becomes a much smaller number with the introduction of a second pilot or second independent sensor stream (0.0001^2). Now the errors are not…

Not at all. The car should be smart enough to figure out which sensor is faulty. Otherwise, the car should not be driving itself at all. It's more like, a pilot has access to multiple sensors. Which they do.

I think you and GP are in agreement. If you have a sensor with a really bad .5 chance of failure, by having to your chance of failure decreases to 0.5^2=0.25.

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#82

Post by luckylion » Sun, Apr 06, 2025, 3:03 PM UTC

pfannkuchen wrote: post #77

xbmcuser wrote: post #22

The point I soured on Musk was when he ditched radar/lidar and tried to go with camera's alone. This made me realize he is not the genius he is made out to be but instead he is a fraud/charlatan and over the years his statements on different topics have only hardened that belief. Why the fuck would you want ai cars to be the same as humans you should want them to be many times better and radarr/lidar sonarr type tech…

You do want them to be better than humans, but vision quality is not really a major source of human accidents. Accidents are typically caused by driving technique, inattention, or a failure to accurately model the behavior of other drivers. Put another way - would giving humans superhuman vision significantly reduce the accident rate? The issue here is that the vision based system failed to even match human capabilit…

Will limiting auto-pilots to human-level vision not increase their accident rate?

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#83

Post by fallingknife » Sun, Apr 06, 2025, 3:03 PM UTC

Finnucane wrote: post #49

ReptileMan wrote: post #25

Earlier quoted context omitted.

We have more than 100 years of proof that visual detection works quite well. The problem is that we don't have certification process before a tech is deployed on the streets - your AV must pass this and this tests before you are allowed to use it. We don't care how you got there lidar, gps/astrology, camera/ai are all ok as long as you pass

each year there are millions of car accidents and tens of thousands of fatalities. Are we really sure visual detection works that great?

We're talking engineering here. Great isn't the standard. The standard is good enough and within budget.

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#84

Post by NoTeslaThrow » Sun, Apr 06, 2025, 3:03 PM UTC

pfannkuchen wrote: post #77

xbmcuser wrote: post #22

The point I soured on Musk was when he ditched radar/lidar and tried to go with camera's alone. This made me realize he is not the genius he is made out to be but instead he is a fraud/charlatan and over the years his statements on different topics have only hardened that belief. Why the fuck would you want ai cars to be the same as humans you should want them to be many times better and radarr/lidar sonarr type tech…

You do want them to be better than humans, but vision quality is not really a major source of human accidents. Accidents are typically caused by driving technique, inattention, or a failure to accurately model the behavior of other drivers. Put another way - would giving humans superhuman vision significantly reduce the accident rate? The issue here is that the vision based system failed to even match human capabilit…

> but vision quality is not really a major source of human accidents.

Never driven before?

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#85

Post by sitkack » Sun, Apr 06, 2025, 3:03 PM UTC

redox99 wrote: post #9

This article does not adjust for the fact that Teslas drive many more miles on Autopilot + FSD than any other brand. So it's a pretty worthless comparison.

Please stop spamming Tesla support over this thread.

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#86

Post by dagw » Sun, Apr 06, 2025, 3:03 PM UTC

SideburnsOfDoom wrote: post #50

dagw wrote: post #23

Earlier quoted context omitted.

Without knowing how many FSD miles Teslas have done compared to other brands, it's hard to judge. It could just be that Tesla owners are far more likely to trust and use their cars FSD capabilities, and thus end up in more FSD accidents. Other brands might have so bad FSD that people simply not trust it and basically never use it, and thus never get into an FSD accident.

> Other brands might have so bad FSD that people simply not trust it and basically never use it The issue is not quite how good the automation is in absolute terms, it's how good it is vs. how it is sold. Tesla is an outlier here, right down to the use of the term "FSD" i.e. "Full Self-Driving", when it's nothing of the sort.

That's kind of my point. We don't know from the data if the problem is that Tesla has worse FSD than the non-accident brands, or if all brands are equally good/bad and Tesla owners are more likely to use the feature. Tesla could very well have the best FSD software on the market and still lead the accident stats simply because how their users are (ab)using the feature (which I suspect is closest to the truth)

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#87

Post by moralestapia » Sun, Apr 06, 2025, 3:03 PM UTC

xbmcuser wrote: post #22

The point I soured on Musk was when he ditched radar/lidar and tried to go with camera's alone. This made me realize he is not the genius he is made out to be but instead he is a fraud/charlatan and over the years his statements on different topics have only hardened that belief. Why the fuck would you want ai cars to be the same as humans you should want them to be many times better and radarr/lidar sonarr type tech…

>This made me realize he is not the genius he is made out to be but instead he is a fraud/charlatan and over the years his statements on different topics have only hardened that belief.

That was Karpathy's decision [1] and, yes, I also have that perception of him.

I know this is not going to be well received because he's one of HN's pet prodigies but, objectively, it was him.

1: https://www.forbes.com/sites/bradtempleton/2022/10/31/former...

(one of many)

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#88

Post by maxerickson » Sun, Apr 06, 2025, 3:03 PM UTC

pfannkuchen wrote: post #77

xbmcuser wrote: post #22

The point I soured on Musk was when he ditched radar/lidar and tried to go with camera's alone. This made me realize he is not the genius he is made out to be but instead he is a fraud/charlatan and over the years his statements on different topics have only hardened that belief. Why the fuck would you want ai cars to be the same as humans you should want them to be many times better and radarr/lidar sonarr type tech…

You do want them to be better than humans, but vision quality is not really a major source of human accidents. Accidents are typically caused by driving technique, inattention, or a failure to accurately model the behavior of other drivers. Put another way - would giving humans superhuman vision significantly reduce the accident rate? The issue here is that the vision based system failed to even match human capabilit…

The improved sensors improve awareness, improving situational awareness of human drivers would have a huge impact.

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#89

Post by redox99 » Sun, Apr 06, 2025, 3:04 PM UTC

tyree731 wrote: post #64

redox99 wrote: post #24

Earlier quoted context omitted.

The shortcomings of 2 year old software is not relevant, because that's not what the cars run now.

By this argument, since Tesla regularly updates the software, no one can discuss the weaknesses in their self driving software.

I never said that. Just discuss the current version or at least something recent (any v13 version).

Re: Self-Driving Teslas Are Fatally Rear-Ending Motorcyclists More Than Any Other

#90

Post by artursapek » Sun, Apr 06, 2025, 3:04 PM UTC

timschmidt wrote: post #78

energy123 wrote: post #30

Earlier quoted context omitted.

Multiple independent sensor streams is like having multiple pilots fly a plane instead of just one pilot. The chance of a fatal error (false negative identification of an object near the car) decreases from p to p^n. The size of that decrease is not intuitive: if p=0.0001 it now becomes a much smaller number with the introduction of a second pilot or second independent sensor stream (0.0001^2). Now the errors are not…

I thought it was consensus that the move away from lidar was driven by a lack of supply compared to desired EV production numbers, and the things said about making vision only work have all been about coping with that reality.

They didn't let lack of battery supply stop them.