FSD Beta 10.69

[old pilot]

You might want to keep your foot near the Rheostat.

Footage Shows Tesla in Eight-Car Pileup After Stopping for No Reason

Anyone feeling overly confident about Tesla’s current driver-assist capabilities should probably sit down and take a look at this footage of a Model S allegedly engaged in Full-Self Driving mode seemingly single-handedly sparking an eight-car pile-up on San Francisco’s Bay Bridge.

currently.att.yahoo.com

 

[X Fan]

You might want to keep your foot near the Rheostat.

Footage Shows Tesla in Eight-Car Pileup After Stopping for No Reason

Anyone feeling overly confident about Tesla’s current driver-assist capabilities should probably sit down and take a look at this footage of a Model S allegedly engaged in Full-Self Driving mode seemingly single-handedly sparking an eight-car pile-up on San Francisco’s Bay Bridge.

currently.att.yahoo.com

yup, there has never been accidents whereby a car is disabled in a lane and the vehicles speeding behind them crash into each other.

FWIW: if it was an FSD phantom issue, why didn't the driver take over and increase speed back to normal? There was plenty of time to do that (looking at the video at least 6 seconds).

 

[old pilot]

yup, there has never been accidents whereby a car is disabled in a lane and the vehicles speeding behind them crash into each other.

FWIW: if it was an FSD phantom issue, why didn't the driver take over and increase speed back to normal? There was plenty of time to do that (looking at the video at least 6 seconds).

I agree, just giving a warning to all to be careful!

 

[Knightshade]

FWIW: if it was an FSD phantom issue, why didn't the driver take over and increase speed back to normal? There was plenty of time to do that (looking at the video at least 6 seconds).

Also numerous folks in the other thread I've seen discussing this have said that's a divided highway type road/tunnel on which FSD does not engage-- so FSD wouldn't have been involved- at best it'd be regular (old code) AP/NoA that is (soon) to be replaced by FSD code once the merge happens.

 

[AlanSubie4Life]

FSD wouldn't have been involved- at best it'd be regular (old code) AP/NoA that is (soon) to be replaced by FSD code once the merge happens.

Yes. Though colloquially using the term FSD is fine, even if it is not accurate.

FSD was potentially in use, roughly speaking. Even though FSD Beta feature for V11 most likely was not. (Though possible, pretty unlikely.)

 

[willow_hiller]

Yes. Though colloquially using the term FSD is fine, even if it is not accurate.

FSD was potentially in use, roughly speaking.

The main issue is that The Intercept (organization that obtained the footage) is trying to imply that FSD Beta is to blame. Right after describing the incident, their article states:

"Just hours before the crash, Tesla CEO Elon Musk had triumphantly announced that Tesla’s “Full Self-Driving” capability was available in North America, congratulating Tesla employees on a “major milestone.” By the end of last year, Tesla had rolled out the feature to over 285,000 people in North America, according to the company."

 

[AlanSubie4Life]

The main issue is that The Intercept (organization that obtained the footage) is trying to imply that FSD Beta is to blame. Right after describing the incident, their article states:

Yes. There are many examples of inaccurate media reporting, in general. This is what Twitter is going to fix.

 

[old pilot]

Yes. There are many examples of inaccurate media reporting, in general. This is what Twitter is going to fix.

I agree with: inaccurate media reporting,. I disagree with: Twitter is going to fix.
The bottom line is that the Tesla driver is not responsible for the accident, the drivers behind are. The same thing has happened many time with ICE cars running out of gas, engine failure, transmission failure and driver failures. I don't know what happened and none of us will until there is a accident report and an investigation.

 

[Dewg]

At the risk of adding fuel to the fire, for some reference to the discussion of the accident which occurred in California:

REAR-END ACCIDENT CAUSED BY SPEEDING OR CHANGING SPEED
Under California Vehicle Code Section 22350, drivers must maintain a reasonable driving speed. Therefore, when a trailing car is driving too fast and collides with a car in front of them, the rear driver will likely be liable for excessive speed. However, speed goes both ways. A front driver that drops their speed dramatically by slamming on their brakes may also potentially be liable for a rear-end collision.

Although the front driver may have a good reason to dramatically decrease their speed by applying their brakes, the act of brake-checking (intentionally applying brakes to promote distance between the front and rear driver) is not protected by law. It can be a dangerous practice causing liability for the brake-checking front driver.

SAFE DISTANCE-RELATED COLLISIONS (“TAILGATING”)
Drivers found liable for tailgating are virtually always the rear driver. Under California Vehicle Code Section 21703, drivers must maintain a safe distance from other automobiles in front of them. In addition, California law requires drivers to maintain a “reasonable and prudent” distance from other vehicles. Unfortunately, this provides little guidance as it is a subjective standard.

Many traffic safety experts and attorneys suggest maintaining at least a two-car length distance away from the front car whenever possible. So long as maintaining this distance does not place other automobiles in the rear at risk of collision (dramatically decreasing speed to achieve this distance), a driver will likely comply with the law.

Nearly all rear-end traffic collision cases revolve around a theory that one or more drivers involved in the accident were negligent. However, proving negligence is much more complicated than presenting evidence of a crash.

PRESUMPTION OF NEGLIGENCE
In most rear-end collisions, the rear driver is presumed to be the negligent party. Whether the rear driver was speeding, tailgating, driving while texting, or operating a faulty vehicle (for example, poor brakes), the rear driver is likely the negligent driver. However, the rear driver may rebut the presumption of negligent by presenting evidence that the front driver was negligent. Examples of the front driver’s negligence may include:

Unsafe lane changes
Brake checking
Failure to employ a proper turning signal
Malfunctioning taillights

 

[old pilot]

At the risk of adding fuel to the fire, for some reference to the discussion of the accident which occurred in California:

REAR-END ACCIDENT CAUSED BY SPEEDING OR CHANGING SPEED
Under California Vehicle Code Section 22350, drivers must maintain a reasonable driving speed. Therefore, when a trailing car is driving too fast and collides with a car in front of them, the rear driver will likely be liable for excessive speed. However, speed goes both ways. A front driver that drops their speed dramatically by slamming on their brakes may also potentially be liable for a rear-end collision.

Although the front driver may have a good reason to dramatically decrease their speed by applying their brakes, the act of brake-checking (intentionally applying brakes to promote distance between the front and rear driver) is not protected by law. It can be a dangerous practice causing liability for the brake-checking front driver.

SAFE DISTANCE-RELATED COLLISIONS (“TAILGATING”)
Drivers found liable for tailgating are virtually always the rear driver. Under California Vehicle Code Section 21703, drivers must maintain a safe distance from other automobiles in front of them. In addition, California law requires drivers to maintain a “reasonable and prudent” distance from other vehicles. Unfortunately, this provides little guidance as it is a subjective standard.

Many traffic safety experts and attorneys suggest maintaining at least a two-car length distance away from the front car whenever possible. So long as maintaining this distance does not place other automobiles in the rear at risk of collision (dramatically decreasing speed to achieve this distance), a driver will likely comply with the law.

Nearly all rear-end traffic collision cases revolve around a theory that one or more drivers involved in the accident were negligent. However, proving negligence is much more complicated than presenting evidence of a crash.

PRESUMPTION OF NEGLIGENCE
In most rear-end collisions, the rear driver is presumed to be the negligent party. Whether the rear driver was speeding, tailgating, driving while texting, or operating a faulty vehicle (for example, poor brakes), the rear driver is likely the negligent driver. However, the rear driver may rebut the presumption of negligent by presenting evidence that the front driver was negligent. Examples of the front driver’s negligence may include:

Unsafe lane changes
Brake checking
Failure to employ a proper turning signal
Malfunctioning taillights

(intentionally applying brakes to promote distance between the front and rear driver) I agree that this is not correct behavior and should result in penalties.
I presumed the driver of the Tesla to be a safe and responsible driver. Thanks for clarifying.

 

[2101Guy]

(intentionally applying brakes to promote distance between the front and rear driver) I agree that this is not correct behavior and should result in penalties.
I presumed the driver of the Tesla to be a safe and responsible driver. Thanks for clarifying.

Define “driver of the Tesla” in this instance..

 

[old pilot]

Define “driver of the Tesla” in this instance..

Human

 

[Dewg]

Human

To be more specific in the definition: Although some scientists equate the term "human" with all members of the genus Homo, in common usage, it generally refers to Homo sapiens, the only extant member.

So, to clarify, Homo Sapien.

 

[old pilot]

To be more specific in the definition: Although some scientists equate the term "human" with all members of the genus Homo, in common usage, it generally refers to Homo sapiens, the only extant member.

So, to clarify, Homo Sapien.

Yes

 

[honoraryorange]

I don't see much change in unmarked roads or distance from parked cars in 69.25 compared to 69.3. But I think you are talking about narrow streets with a center line ... parked cars on the sides ? We don't have much of that in Eastern suburbs I drive most.

Late reply here, but almost the opposite - Although I do agree the problem can be visible on narrow streets too.

Here we have very wide streets, with no markings except the center lines (or center turn lane). There's unmarking parking along the curb, and the streets are still wide enough to offer tons of space between those and the regular driving area. On these streets, FSD will basically drive in the center of the ultrawide lane, looking very out of place. And when cars are parked there, it'll drive uncomfortable close to them when the streets are so wide a parked car could normally fling their door open fearlessly without fear of being taken out. Unless a Tesla is around lol

Examples - in both of these scenarios FSD tries to keep itself centered when it really should instead try to keep itself within a standard distance of the center line. When parked cars are around it tends to weave back and forth, coming closer to the center, (while still hugging those cars) and then taking up more space again. And at lights, of course, sitting dead in the middle. I know it has always been like this, but it just feels worse to me with the last update or two.

 

[AlanSubie4Life]

Late reply here, but almost the opposite - Although I do agree the problem can be visible on narrow streets too.

Here we have very wide streets, with no markings except the center lines (or center turn lane). There's unmarking parking along the curb, and the streets are still wide enough to offer tons of space between those and the regular driving area. On these streets, FSD will basically drive in the center of the ultrawide lane, looking very out of place. And when cars are parked there, it'll drive uncomfortable close to them when the streets are so wide a parked car could normally fling their door open fearlessly without fear of being taken out. Unless a Tesla is around lol

Examples - in both of these scenarios FSD tries to keep itself centered when it really should instead try to keep itself within a standard distance of the center line. When parked cars are around it tends to weave back and forth, coming closer to the center, (while still hugging those cars) and then taking up more space again. And at lights, of course, sitting dead in the middle. I know it has always been like this, but it just feels worse to me with the last update or two.

View attachment 894616View attachment 894617

Yep, this is a big (and old) problem. It has changed over time in severity and behavior with different releases. Seem to be at a local maximum for crappiness at the current time.

But in general the car has a hard time driving in the naturally correct position in the road, unless the markings are very clear and consistent. If not, it will veer all over the road with variable lane positioning.

It also turns prematurely and angles the vehicle incorrectly at turns, etc. - a somewhat related problem since it is made worse by lack of markings.

 

[kabin]

And for some left or right turn lanes it frequently doesn't acknowledge or know how to process road lines and instead carries excess speed and then brakes hard when the vehicle is about 50ft from the intersection before driving over the solid left or right turn lane line and into the turn lane.

 

[AlanSubie4Life]

And for some left or right turn lanes it frequently doesn't acknowledge or know how to process road lines and instead carries excess speed and then brakes hard when the vehicle is about 50ft from the intersection before driving over the solid left or right turn lane line and into the turn lane.

Yes, there are a lot of corner cases like this.

 

[JHCCAZ]

To be more specific in the definition: Although some scientists equate the term "human" with all members of the genus Homo, in common usage, it generally refers to Homo sapiens, the only extant member.

So, to clarify, Homo Sapien.

Yes

My wife, the unchallenged authority in all subjects, would say that my Tesla is piloted by a true Neanderthal.

So to clarify, a genus but not a genius.

 

[AlanSubie4Life]

Glad to see we are trying to keep the discussion tightly focused on 10.69 and document current behaviors before the imminent transition to v11. This should act as excellent documentation for future reference - effectively an information-dense snapshot of current state of the art for driver assistance.

It really continues to amaze me how poor the performance is in scenarios with downhill grades, and with higher speeds. I suppose it is really no wonder at all we do not have V11 if they are struggling at 35mph-40mph. (This is the transition zone to doggie-doo.)

For the record, I am linking here to my other video on intervention scenarios, since it is related to 10.69:

What are your most common intervention scenarios?

I've been in the beta program since 10.2 and - while there have certainly been regressions - FSDb has gone from being erratic and scary to quite predictable, almost boring! Zero intervention trips are now common, and even my long daily commute only has a couple of them, usually at predictable...

teslamotorsclub.com

And the direct video link if you don’t need the discussion in the link above: