7 ms·

I have FSD but not the beta. When cars cut across my lane autopilot will not react for a solid second or so, then proceeds to slam on the brakes after the car i

I have FSD but not the beta. When cars cut across my lane autopilot will not react for a solid second or so, then proceeds to slam on the brakes after the car is clear. It does not instill confidence to say the least. From my POV it feels like someone put a debounce on the reaction to prevent phantom breaking, to the point the car doesn’t brake when it needs to.

ptidhomme 5y ago
IIRC there was also a 1 second delay before reaction on the Uber car that crashed into a pedestrian. There must be quite a lot of false alarms to filter out...
mumblemumble 5y ago
I worry a little bit about things like this. If they were just using debouncing on a band-aid rather than dealing directly with the bad input that led to the phantom breaking, then that would seem to imply that the project is falling into the classic death march project trap. It turns out that the last 10% of the project is going to be 90% of the work. Meaning that the original schedule was off by an order of magnitude. But the business has made made huge promises that mean that the team is not at liberty to adjust the schedule accordingly. And so they're under ever increasing pressure to cut corners and pile on quick fixes.
- zaptrem 5y ago
  Production AP is a completely different software stack than FSD beta. I think the only things that carry over are the car detection, VRU, and traffic light CV NNs.
rootusrootus 5y ago
The part that always bothered me most about AP on the highway was how willing it is to drive full speed into an obvious problem. Obvious for me, that is. Dense merging traffic, for example -- I can see the conflict coming, I can read the way people are driving on that on-ramp, and I take defensive action. My Tesla would just drive right into the center of that mess than then panic when things got crazy. I no longer have my Tesla, but I stopped using AP before I sold the car, partly as a result of this behavior. And partly because phantom braking became pretty common for a number of overpasses we go under on our way to grandma's house, and it really scares the crap out of my wife & kids. Aside from phantom braking, the best use case for AP, in my opinion, is to be a jerk and cruise in the middle lane of the freeway. It deals okay with that most of the time.
- cryptoegorophy 5y ago
  well, you don't just "trust" it, if you are smart enough you quickly learn the good and bad sides of it. When I use it (95% of time) I know what to expect and I have a habit of putting my foot on the accelerator pedal instead of break pedal for "phantom" braking which has gotten a lot better than let say in 2018. You also know the limits - if the lines look strange - no lines at all or if they are not perfect or if there are extra painted lines then you have your hand on your knee with a finger holding the steering wheel ready to quickly grab the wheel and go. You also learn as a pro tip to let the car push its limits and let it do what it wants to do like doing something dangerous then realizing that it would correct itself after your "comfortable" zone. People expect a "human" like behaviour from FSD/AP but it is not, it is far from it. It is a good/ideal assistant but not a full replacement. I wish tesla did some training for everyone on autopilot to show them what to expect and how to handle it.
  - gugagore 5y ago
    > You also learn as a pro tip to let the car push its limits and let it do what it wants to do like doing something dangerous then realizing that it would correct itself after your "comfortable" zone. That's literally "trusting it". As a human, you are attempting to learn the situations under which the car operates safely. You could be very conservative and say "it never works". The big problem is that the kinds of errors that occur in these systems are very different from the kinds of errors that humans expect. Even designers of autonomous systems can be surprised by emergent behavior. You say you can quickly learn the good and bad sides of it, however, that's not true. How much reliability do we expect from a system that, when it fails, leads to sever injury or death? ASIL-D is something like one failure in 10^9 hours [1]. That's 100,000 years. You do not have that much experience with your Tesla, sorry. In principle, you might drive every waking hour for 10 years, never witness an accident, and you still not have an argument that passes even the most basic scrutiny for safety for critical systems. It could make a fatal mistake in the 11th year and that would not be a big surprise. No one has that much experience with the ASIL-D components of any car. That's why I we must rely on regulation and standardization to help keep us safe. Safety isn't just "well I tried it a bunch under these conditions and didn't get hurt". And that's why what Tesla's strategy is reckless. [1] https://en.wikipedia.org/wiki/Automotive_Safety_Integrity_Level https://en.wikipedia.org/wiki/Automotive_Safety_Integrity_Le...
no_butterscotch 5y ago
I don't have FSD, but autosteer + cruise-control in my Tesla does this as well. Additionally I've experienced cases of "phantom braking" (https://www.cbsnews.com/news/automatic-emergency-braking-in-cars-will-be-standard-by-2022-but-false-activations-still-a-concern/ https://www.cbsnews.com/news/automatic-emergency-braking-in-...) which doesn't instill confidence in those features.
LeoPanthera 5y ago
I also have FSD but not the beta and I just wanted to add that my car (a 2020 model X) does not do this. Actually I've always been very impressed with how it smoothly slows to avoid merging cars. Is it possible you have older autopilot hardware that runs different/older software?
- joshribakoff 5y ago
  I have the latest hardware and software. Note I was not talking about merging cars. I’m talking about someone cutting across my lane from a perpendicular street, so they can enter a lane adjacent to mine.

ptidhomme 5y ago

IIRC there was also a 1 second delay before reaction on the Uber car that crashed into a pedestrian. There must be quite a lot of false alarms to filter out...

mumblemumble 5y ago

I worry a little bit about things like this. If they were just using debouncing on a band-aid rather than dealing directly with the bad input that led to the phantom breaking, then that would seem to imply that the project is falling into the classic death march project trap. It turns out that the last 10% of the project is going to be 90% of the work. Meaning that the original schedule was off by an order of magnitude. But the business has made made huge promises that mean that the team is not at liberty to adjust the schedule accordingly. And so they're under ever increasing pressure to cut corners and pile on quick fixes.

zaptrem 5y ago
Production AP is a completely different software stack than FSD beta. I think the only things that carry over are the car detection, VRU, and traffic light CV NNs.

rootusrootus 5y ago

The part that always bothered me most about AP on the highway was how willing it is to drive full speed into an obvious problem. Obvious for me, that is. Dense merging traffic, for example -- I can see the conflict coming, I can read the way people are driving on that on-ramp, and I take defensive action. My Tesla would just drive right into the center of that mess than then panic when things got crazy. I no longer have my Tesla, but I stopped using AP before I sold the car, partly as a result of this behavior. And partly because phantom braking became pretty common for a number of overpasses we go under on our way to grandma's house, and it really scares the crap out of my wife & kids. Aside from phantom braking, the best use case for AP, in my opinion, is to be a jerk and cruise in the middle lane of the freeway. It deals okay with that most of the time.

cryptoegorophy 5y ago
well, you don't just "trust" it, if you are smart enough you quickly learn the good and bad sides of it. When I use it (95% of time) I know what to expect and I have a habit of putting my foot on the accelerator pedal instead of break pedal for "phantom" braking which has gotten a lot better than let say in 2018. You also know the limits - if the lines look strange - no lines at all or if they are not perfect or if there are extra painted lines then you have your hand on your knee with a finger holding the steering wheel ready to quickly grab the wheel and go. You also learn as a pro tip to let the car push its limits and let it do what it wants to do like doing something dangerous then realizing that it would correct itself after your "comfortable" zone. People expect a "human" like behaviour from FSD/AP but it is not, it is far from it. It is a good/ideal assistant but not a full replacement. I wish tesla did some training for everyone on autopilot to show them what to expect and how to handle it.
- gugagore 5y ago
  > You also learn as a pro tip to let the car push its limits and let it do what it wants to do like doing something dangerous then realizing that it would correct itself after your "comfortable" zone. That's literally "trusting it". As a human, you are attempting to learn the situations under which the car operates safely. You could be very conservative and say "it never works". The big problem is that the kinds of errors that occur in these systems are very different from the kinds of errors that humans expect. Even designers of autonomous systems can be surprised by emergent behavior. You say you can quickly learn the good and bad sides of it, however, that's not true. How much reliability do we expect from a system that, when it fails, leads to sever injury or death? ASIL-D is something like one failure in 10^9 hours [1]. That's 100,000 years. You do not have that much experience with your Tesla, sorry. In principle, you might drive every waking hour for 10 years, never witness an accident, and you still not have an argument that passes even the most basic scrutiny for safety for critical systems. It could make a fatal mistake in the 11th year and that would not be a big surprise. No one has that much experience with the ASIL-D components of any car. That's why I we must rely on regulation and standardization to help keep us safe. Safety isn't just "well I tried it a bunch under these conditions and didn't get hurt". And that's why what Tesla's strategy is reckless. [1] https://en.wikipedia.org/wiki/Automotive_Safety_Integrity_Level https://en.wikipedia.org/wiki/Automotive_Safety_Integrity_Le...

no_butterscotch 5y ago

I don't have FSD, but autosteer + cruise-control in my Tesla does this as well. Additionally I've experienced cases of "phantom braking" (https://www.cbsnews.com/news/automatic-emergency-braking-in-cars-will-be-standard-by-2022-but-false-activations-still-a-concern/ https://www.cbsnews.com/news/automatic-emergency-braking-in-...) which doesn't instill confidence in those features.

LeoPanthera 5y ago

I also have FSD but not the beta and I just wanted to add that my car (a 2020 model X) does not do this. Actually I've always been very impressed with how it smoothly slows to avoid merging cars. Is it possible you have older autopilot hardware that runs different/older software?

joshribakoff 5y ago
I have the latest hardware and software. Note I was not talking about merging cars. I’m talking about someone cutting across my lane from a perpendicular street, so they can enter a lane adjacent to mine.