Tesla FSD drives down railroad tracks, gets hit by train

[pamalabama]

Stop making excuses. We aren't talking about cruise control. You and others and Tesla are arguing that FSD DRIVES better and safer than human drivers. Prove it. Prove that FSD makes fewer mistakes than human drivers.

We never said that. Tesla never claimed that. They claim that FSD + Human driver exceeds the safety of a human driver.

FSD does not hit cars or pedestrians. So the safety of the system is very good. FSD might go too fast in a school zone but that is a relatively easy thing to predict as FSD has very few limitations and they are easily known.

FSD is driving and making split second decisions. If it were crashing all the time we would know about it since a self driving system would be hard to supervise if the fundamental driving were dangerous. Luckily the fundamental driving of tesla is excellent, even in hard conditions.

you can just turn on FSD in heavy rain and it seems to see better than you can. It's almost like it enhances safety

Sponsored

 

[COdogman]

Did you even read what you posted? Tesla doesn't know anything other than accidents.

Tesla collects video data just like rivian for training purposes. They don't record every close call event. They don't record accident avoidance maneuvers because those are things that they have designed to have high reliability.


Rivian doesn't even know that they get into accidents, hence rivian is not reporting any accidents to the NHTSA that happen in user vehicles. That seems alarming to me if anything.

Why does rivian not know when one of their cars gets into an accident while using the adas system? What are they hiding?

That is absolutely NOT the point of that article. Now you are just flat out lying.

Rivian isn't making up BS stats about how safe their ADAS is. Stop trying to change the subject.

 

[pamalabama]

Why you gotta lie so much.

https://rivian.com/support/article/how-do-i-access-data-on-my-event-data-recorder

I'm not lying. Look at the NHTSA document which shows all accidents when using ADAS technology. It's almost only tesla reporting accidents.

Rivian reported 2 or 3 but they are in-house reports from test drivers. There is also a report of a user who manually submitted a report.


Rivian (which has not driven that many miles) somehow has no accidents reported from user vehicles, yet 2 or 3 from in-house vehicles? We know these accidents are happening because some of them are even reported on this forum.

Even the NHTSA acknowledges that while automakers are required to submit telemetry data for every accident that occurs within 5 seconds of an adas being active, most manufacturers lack the capability to do it.

Tesla is hiding nothing. There was a user who disengaged FSD on purpose, steered his car into a tree and flipped it over. They still report it even if Tesla is not at fault. Tesla telematics contol unit is in the top of the car. the car flipped over and the car landed right on it. Tesla still shared the data anyways. They hid nothing

 

[COdogman]

We never said that. Tesla never claimed that. They claim that FSD + Human driver exceeds the safety of a human driver.

FSD does not hit cars or pedestrians. So the safety of the system is very good. FSD might go too fast in a school zone but that is a relatively easy thing to predict as FSD has very few limitations and they are easily known.

FSD is driving and making split second decisions. If it were crashing all the time we would know about it since a self driving system would be hard to supervise if the fundamental driving were dangerous. Luckily the fundamental driving of tesla is excellent, even in hard conditions.

you can just turn on FSD in heavy rain and it seems to see better than you can. It's almost like it enhances safety

Mesa woman killed in Tesla ‘Full Self-Driving’ crash; feds to probe automaker

https://www.azfamily.com/2024/10/18...full-self-driving-crash-feds-probe-automaker/

An agency database says the pedestrian was killed in Rimrock, Arizona, just north of Camp Verde and 100 miles from Phoenix in November 2023 after being hit by a 2021 Tesla Model Y./QUOTE]

 

[SteveInBend]

For me, personally, it's a binary decision: either the car drives itself automatically 100% of the time and the manufacturer assumes all liability for any accidents or it doesn't drive the car at all.

 

[pamalabama]

Mesa woman killed in Tesla ‘Full Self-Driving’ crash; feds to probe automaker

https://www.azfamily.com/2024/10/18...full-self-driving-crash-feds-probe-automaker/

yes. That was one of the two deaths with FSD.

There are only two deaths. One was this pedestrian back in 2023 and the other was a guy where a truck pulled out in front of him.

Both of those instances were not technically FSD because FSD was only on city streets. That was the old navigate on autopilot product

 

[COdogman]

yes. That was one of the two deaths with FSD.

FSD does not hit cars or pedestrians.

 

[pamalabama]

Read my full comment. That was navigate on autopilot.

Basically tesla has a product called autopilot which has very old perception from 2017.

Tesla used FSD for city streets driving only and the autopilot product would switch on when FSD entered the freeway/highway.

These are both freeway/highway accidents.

They did it this way because autopilot on highways was "good enough" compared to the amount of interventions you would have in city streets driving

That's why most companies do not offer city streets driving. The vulnerable road user detection that tesla has is not easy for others to achieve. You can see FSD dodging small rabbits running across the road, families of ducks/birds, etc.

It's not easy to do this

It's worth mentioning that there are quite a few people who have deliberately not updated to an FSD version that enables FSD on highways (instead of autopilot) because autopilot allows you to set your speed to anything you want and you can minimize lane changes. FSD on highways does not have speed control. It drives the speed it wants and changes lanes however much it wants.

 

[COdogman]

Read my full comment. That was navigate on autopilot.

Basically tesla has a product called autopilot which has very old perception from 2017.

Tesla used FSD for city streets driving only and the autopilot product would switch on when FSD entered the freeway/highway.

These are both freeway/highway accidents.

They did it this way because autopilot on highways was "good enough" compared to the amount of interventions you would have in city streets driving

That's why most companies do not offer city streets driving. The vulnerable road user detection that tesla has is not easy for others to achieve. You can see FSD dodging small rabbits running across the road, families of ducks/birds, etc.

It's not easy to do this

I'm sure all those excuses sound great to that woman's family and all future pedestrians, bikers, first responders who "weren't/ aren't hit" by FSD or "good enough" Autopilot.

Just admit you are totally ok with people dying while Tesla experiments on all of us to prove camera only ADAS is possible.... How many are acceptable to you?

I've reached my daily patience limit for your nonsense. Have a great evening.

 

[pamalabama]

I'm sure all those excuses sound great to that woman's family and all future pedestrians, bikers, first responders who "weren't/ aren't hit" by FSD or "good enough" Autopilot.

Just admit you are totally ok with people dying while Tesla experiments on all of us to prove camera only ADAS is possible.... How many are acceptable to you?

I've reached my daily patience limit for your nonsense. Have a great evening.

you're just talking shit. 2 people died. In 4 billion miles. Where are those 40 other people who didn't die?

90% of accidents are caused by human error. Computers don't get tired.

https://www.tesladeaths.com/miles


If you look here they take any opportunity (when tesla shares mileage of tesla vehicles) to calculate miles per death. It is below average but not low

 

[KootenayEV]

All I can add is that I hope you have some sense of understanding and can somehow grasp and realize that Tesla FSD is supervised, not autonomous, self-driving, level 5, etc.
By your own conclusion, FSD is therefore unsafe, unreliable because it showed issues in your 30 miles of driving. Fortunately, and like everything else in life, normalization applies to make better sense of events and issues. You know, Statistics 101. Seen within that reasonable and expected and valid context, i.e., when normalized, FSD actually is indeed safer than human driving, much to the chagrin of Tesla haters .

Signed,
Messenger.

You are missing the point. Majority of humans have a hard time following simple directions, and as others have mentioned, our brains are wired in such a way that taking over a task instantaneously is very difficult. My point was that it made a simple serious mistake relatively soon (I didn't even mention all of the other bonehead moves it made) and ergo until there is some ACTUAL RELIABLE DATA that conclusively shows it is safer than a human, it could then go into the general publics hands.

To note, this was only a few weeks ago in a Cybertruck - which really shouldn't matter, as my claim is that this version, and obviously any previous version, shouldn't be legal as a matter of public safety.

 

[COdogman]

you're just talking shit. 2 people died. In 4 billion miles. Where are those 40 other people who didn't die?

90% of accidents are caused by human error. Computers don't get tired.

https://www.tesladeaths.com/miles


If you look here they take any opportunity (when tesla shares mileage of tesla vehicles) to calculate miles per death. It is below average but not low

Thank you for (again) sharing this BS crowdsourced information. That doesn't include any of the data that I previously said would provide insight into the overall safety of Tesla FSD/ Autopilot. Or are there separate, crowdsourced "Tesla injuries", "Tesla FSD errors" websites?

Just stop with this nonsense. Tesla has the data that could prove it's as safe as you claim and they refuse to release it.

 

[carsly]

Tesla has been publishing quarterly safety data since 2018, still waiting on other automakers to follow suit with their adaptive cruise control, lane centering, BlueCruise or whatever comparable ADAS technologies they have in production on public roads. Where are you seeing their data?

https://www.tesla.com/VehicleSafetyReport

Facts only matter if you look at them.

 

[pamalabama]

Thank you for (again) sharing this BS crowdsourced information. That doesn't include any of the data that I previously said would provide insight into the overall safety of Tesla FSD/ Autopilot. Or are there separate, crowdsourced "Tesla injuries", "Tesla FSD errors" websites?

Just stop with this nonsense. Tesla has the data that could prove it's as safe as you claim and they refuse to release it.

Because Tesla shares accident stats and you find they are not reliable enough. Tesla shares the data they have and you won't accept it because tesla looks good.

you won't accept that tesla is driving for billions of miles and accidents happen

Rivian can't accumulate this many miles because they have barely sold any vehicles

Tesla is reporting accidents to the NHTSA through the standing general order. You can see the accidents. It's not that many relative to how many billions of miles tesla vehicles are driving

https://www.nhtsa.gov/laws-regulations/standing-general-order-crash-reporting#level-2-adas

You can see all you need to know here.

You just can't accept that tesla can build a safe system with cameras only. You can't accept because one manufacturer needs 11 cameras, 5 radars (and now lidar with R2) and has the worst ADAS in a modern vehicle, that tesla can't build a safe product

People like to discredit the work that tesla is doing with FSD because it is a large difference that makes other automakers feel obsolete. FSD was unsafe back in version 8 or 9, but back then tesla had stringent safety requirements and very few people had access to the software

 

[COdogman]

Tesla has been publishing quarterly safety data since 2018, still waiting on other automakers to follow suit with their adaptive cruise control, lane centering, BlueCruise or whatever comparable ADAS technologies they have in production on public roads. Where are you seeing their data?

https://www.tesla.com/VehicleSafetyReport

Facts only matter if you look at them.

First, that is not data. It is Tesla's sanitized, completely unverified interpretation of the data.

Second, if you actually read the report you would notice that it doesn't even include FSD. It includes Autopilot and non-Autopilot miles.

Sponsored