Can’t believe people here are making excuses about this “It’s not common to have a barricade front of a gate” - like wtf. Don’t run into a barricade, does not matter if there is a gate in front, behind, to the side or nowhere near.
I've had great luck with v12.5 but then again I don't have any gates. Glad you reported this. My biggest disengagement problem are potholes. Don't want to chance having tire and/or suspension damage. This is especially a problem after the winter thaw when potholes are found more frequently.
@@edhill8568 I think it sometimes hard to know which road hazards the car will see and react to it which ones it will ignore. I’m always extra alert when I see something like paper trash blowing across the road.
barricades are an interesting failure point, with them being so small on camera. first time watching I didn't see the barricade on the video before you braked, but going back through the video the barricade is visible for at least 3 seconds before, so should be plenty of time to be spotted. Do you have any other barricades like this one in your area you can test? see if the failure is consistent?
That is similar to parking case when you need to enter and exit with barricades. I don’t believe this version (or any prior versions) does that yet. It likely can’t handle emergency vehicles yet either, so I wouldn’t call these a failure. Not trained to handle such scenario/edge case yet. It is a disengagement, so hopefully you have sent this for future training data.
My 12.5 wanted to go straight in a left only lane. Why are those things not happening to Omar. There is probably very little training data with those types of gates, still I was hoping those new HD cams would pick that up better.
The image for the NN is currently not even SD, we far away from real time HD processing. Even on HW4 cars they currently emulating the camera feed to a lower resolution matching the HW3 resolution. And that image then gets further reduced to something like 360x360. But this issues are not a problem in the long term. If its not fixable with improved architecture of the NN stack we could still just improve hardware processing power to make it work.
Pretty unique setup there. Not usual to have both a barricade and a gate just after. Don't see that anywhere I have lived (over 11 states) or worked (39 states). Not surprised it doesn't know how to handle that yet, very rare situation to train for.
@@tonyvelasco5732 the gate and the barricade are linked together. We have barcodes on the car window when you drive up the gate opens first and once fully open the barricade rises. When you leave there is no authentication required and you just have to wait for the gate to slowly open in 12.3.6 usually would see the gate but not always on time so it’s not exactly trustworthy. No version has ever detected the barricade, and even in this case, given the speed the car was driving, had the barricade not been there. It probably would’ve hit the gate too.
That type of barricade has been a known problem for a while now. From what I've read, the FSD display uses the v11 NN stack, which has a series of NN that find the road, identify objects, street signs, etc.. I've been told (though I don't know for a fact) that FSD 12.x abandons that and takes the image as a whole and feeds that into a single NN, this means weird objects like this, people waving and all kinds of hard to classify things should get recognized and handled properly. I''m curious if anyone else has any insight.
@@joneclegg barricades are not all the same. They can vary in thickness, color, presence or absence of lights or reflectors, etc.. it is not always that important that you know they are a barrier per se, but it is important to know. They are a solid object that you might hit if you don’t stop. Focusing software to identify solid objects in your path would make things safer whether you realize it’s an actual barricade or not.
I'm getting really bad navigational problems with this one, like turning into a residential area 500 feet from the correct turn, or getting into the straight only lane and trying to make a left on a red light from the straight lane, also was suppose to go straight and turned into a parking lot for no reason and started driving around until it eventually pulled toward a parking spot and got stuck.
@@Kev376 interesting. Well, I’m not tested it in 12.5, previous versions have had trouble where lanes are separated by poles that limit lane changes near an intersection. There are a few of these close to where I live and if you are not in the correct left turn lane before these poles start you really cannot turn at that intersection and have to go beyond and then U-turn and come back or in some cases, the Lane you’re in becomes a left turn lane and the polls trap you there such as potentially was going to happen this morning. You can’t really anticipate these until you see them and because they are aligned with your route, they are not visually prominent in a distance. The only real solution is to get into the correct turning lane as early as possible rather than waiting till the last moment.
Excellent edge case! Watch the video in slow motion. @3:19 the speedo says 17mph, @3:21 it has already decelerated to 12mph before the diengagement chime rings. FSD was clearly slowing down before you took over, indicating it saw the gate. I would like to see another attempt
@@ryanslaten oh and I tested on another entrance with barricade only and there was a speed bump before it saw and slowed for and after crossing the barricade opened so no contact. For the barricade in the video with the gate behind, it takes longer for the barricade to open because it doesn’t start to open until the gate is fully opened, and that is slow. If the other entrance was just a barricade, the barricade opens immediately upon the sticker being scanned and the speed bump gets you slowed first so maybe is easier to see or maybe the barricade gets out of the way so fast it doesn’t matter
thank you for your videos! Very informative! As I see fsd is progressing very fast despite the barrier issue. Do you expect Tesla to solve autonomy this or next year?
@@Maksim-he4ej I believe they can do it if they put enough resources into it. I feel like they are understaffed, though leaving widely known issues unaddressed.
Tesla does lots of training in simulation, and their simulation might not have many randomly placed barricades and gates yet. I'd imagine they have railroad crossings and parking garages but not much else, and those are quite different environments with different signage compared to this one.
Appreciate your testing! Yeah, I hear that those horizontal swinging barricades are an issue for FSD; apparently the camera is unable to separate it from the surrounding environment. Don’t know how it will be resolved, but hopefully Tesla will find a solution. I just received 12.5 this morning on my 2024 Model 3. Need to do more testing, but my initial impression is positive. I especially like the stop sign improvement. It’s much more natural and is quicker, even with the need to stop completely (required by NHTSA). Also, I have not yet noticed any lane dancing, but again need more testing. I have noticed maybe a fraction of a second delay when my vehicle is first at a traffic light and it turns from red to green. Not really an issue at all, just an observation when comparing 12.3.6 to 12.5. One other thing, FSD still does not honor No Turn On Red intersections; no doubt something that will be addressed when it is able to read more signs and respond accordingly. I still think that there’s a long way to go before complete autonomy, but 12.5 appears to be a nice step forward.
@@fedderd given I’ve only had two drives so far I agree that the assertiveness in a good way is a good thing. Except for running into the barricade, it does feel more natural. Of course, we will all know more with more time behind the wheel, but definitely ignore the ridiculous hyperbole from Twitter about how great it is and that it is a solved problem.
@@DrD57 Sir, when I read about your complaints I have a feeling that you are already so used to the product that you lost touch with reality. For me, the person who has no Tesla FSD, all these videos, including this one, look like a miracle. Just from watching FSD videos every half year or so, I can see crazy progress and see that it gets so close to fully automated driving. Hope I didn't insult you, have a great day, and thank you for the video, which was very interesting.
@@RightFiveLeft there has been great progress - to be credible though you have to show the successes and the failures. Too many videos are so full of hyperbole about how great it is that they have no credibility and out viewers at risk of being overly trusting or fsd and that’s dangerous at this point
@@DrD57 you are right. I'm also very interested in failures. Because overall it looks it works fine. That basicly how i found your video. I tried to search for accidents.
To be honest, I like to see the failures. So to me its not clickbait. Im very bullish on FSD, but only seeing the good stuff can blind your judgment. What do you think Tesla is more interested in? It ain't the perfect stuff.
@@cheetahrunout I think they are definitely making a lot of progress and the jump to version 12 was a big step but at the same time I don’t think they should push out software that still has no issues if they affect safety at least. I feel like they need to be putting more resources into thisbecause with more coders these issues probably would’ve already been fixed. As it is, they have to hop around from issue to issue dividing their time between new features and fixing bugs.
@DrD57 issue is that its all neutral nets. Therefore isssues like this need training data instead of more coders. Edit: make sure to accept that your data is shared so Tesla can improve on issues like this. (Perhaps thats mandatory for 12.5, I dont know)
@@celeron55 there are many gated neighborhoods to control access to amenities and to know who is in and out and when they come and go. There are usually paid security people to scan ID’s of visitors and residents have car stickers or sometimes codes to enter.
I'm not trying to be bossy, but as we progress further along the FSD journey, it becomes increasingly important to report those edge cases. Thanks for uploading! Its fun to watch the progress.
Good test. The driving in this and the previous video look good and possibly better than previous versions. Still has problems with gates, but I wasn’t expecting that to be fixed just yet. Thanks for the tests. Which Hardware are you on?
@@anunexpectedfire4062 I took delivery March 19 and it’s a 2024 model Y picked up in Central Florida so I assume is version 4 but have not looked in the pages to find it
Looks like the cameras cant detect the barricades and thats a major issue. I believe its the "vision only" navigation thats affecting the FSD from getting where it need to be. I've seen issues (mostly disengagements) with driving in heavy rain and fog and unmarked roads as well. May be a radar would have helped to fix it. But then, certain people are very adamant, aren't they ? 😉
@@Investchan the cameras have plenty of resolution to see them but the software just isn’t looking for them. It would be helpful if map data included this information like it does for stop signs but the Tesla vision software still needs to see them, recognize them, and react. Another branch of this problem, our cases where the driver has to punch a code or talk to someone to get the gate open, whether via intercom or in person. But absolutely the car does not need to just drive into these things.
when you say collision I think you should mention that FSD failed to recognize a gate. the latter is far less of a problem than a collision with another car or a pedestrian. tesla probably knows about this issue and has not attributed a lot of resources to fixing it yet because it's not as high a priority as making sure it doesn't hit humans or cars.
@@mkkkai one would hope - they really do need to have more people working on the obvious gaps in performance of FSD. Probably no one dies if I crash through the gate but between repairs to the car and the gate could be into the $1000 range or more and missing something right in front of two cameras does not instill confidence. I knew prior versions had this issue so my foot was on the brake pedal but what other blind spots are there that we may not anticipate. When you have serious fan boys or even Elon touting a version that never made it to general release was 10x better without showing data it doesn’t make them more trustworthy or help Tesla at all. All known issues should be fixed with adequate coding resources before another release comes out. Given how much people are paying and have paid over the years it’s not a great idea to skimp on software development like this.
@@isonlynameleft I waited as long as I could to see if it would notice and felt like it’s slightly slowed down right before I hit the brakes but yes, it was going to hit it in spite of that and I’m not so sure it would not hit the gate next because they are pretty darn closeand we were moving pretty fast
At what cost is the public providing their lives for future rewards of a personal or corporate profit? When did it become okay to take a life for a product under testing? Shouldn't this remain outside of public harm until proven safe?
@DrD57 When did public safety concerns of potential future harm, such as death become of secondary Interest to Public Safety Officials over the corporate and personal interests of a Software Product akin to Nazi Drowning concentration camp inmates in order to learn of revivilation (sp.) products to sell on the market for improving health. It's a gruesome NAZI style business model. Most of us would qualm about being the cause of the death of others to enhance and improve our soulless product. Rather than payout millions; build a fifty-million town of roads. Test the product for its safety BEFORE you release said product because it has been proven to be safe before public saturation.
@DrD57 Previous reply deleted, second go: Public Safety concerns speak per product testing implies the known unknown of zero deaths per testing trials. Why is FSD given a pass on civilians deaths with beta testing a known imperfect automated guidance system?
@@stuarthdoblin what deaths? Do human drivers cause zero deaths? Should all human drivers also be banned? Tesla has been pretty clear from the beginning that the driver is responsible for safety and should always have a hand on the wheel and eyes on the road. People who participate in the beta testing have to accept that.
@DrD57 That's my point. I never consented to provide my life on a silver platter to Elon's development of FSD. I was just violated and told to accept the new normal. Not OK.
@@dluchin1998 parking has been fine the last few versions. Elon said 12.4 was 10x better than 12.3.6 get 12.4 and 12.4.1 and 12.3.2 were never released. The barricade issue is a big deal as it hitting anything right in front of you and they have to be aware of this. I would not release another version until this is taken care of.
@@DrD57 I guess it's your title I take issue with. It sounds like your saying this version is a failure, when really you just had a single disengagement at a place you already knew was an issue.
@@dluchin1998 it was A failure (the behavior). I assume there will be others. If enough people report these issues they will hopefully get more attention. I still don’t understand why speed limit signs that don’t feature black text on a white background or still ignored either is that is certainly a safety issue as well but colliding with something right in front of the car to me would be number one priority. If it can’t see a barricade right in front of you then what else might it miss?
Obviously it was for a clickbait... For me this drive would have been A minus on my score card as I am macro versus micro.... But again I also agree it's technically a failure if you are talking full on robotaxi capability...
@@jh-qb3yb any time a human must take action to prevent an accident. It is a failure for FSD. Many times there are odd behaviors by FSD that are not really a safety issues, such as slowing down unexpectedly, a few miles an hour, and then picking back up speed or unusual lane changes. But anything a human is required to do to prevent an accident is a failure. When fanboy were openly promoting this version and versions past as being nearly perfect this becomes a bigger safety issue as it can encourage unjustified trust and reduced vigilance from drivers. As users who paid a lot for FSD it is not our job to be cheerleaders when we do an evaluation of an update we should show the good and the bad as we see it.
Ya... Stuff like that is going to force Elon to eventually admit the self driving taxi will be limited to certain areas. I think we're a few years away from vision-only systems being able to handle things like that barricade. I also have seen it fail to detect when some objects move while the car is stationary/parked. Im guessing that would also mean the vision system wont detect new objects that get close to the car while its parked.
I think all these issues are very solvable, but I don’t think they have enough people working on it so they hop around from one thing to another and they feel a lot of pressure to push out new releases, including from me because we are all eager to see the new stuff. In general, it’s better to respond things that are not there than to miss things that are.
@@DrD57 I'm sure we can all put our FSD vehicles in situations it is not trained for. I've yet to hear of FSD stopping for all slat type gates. Have you read that in any of the releases? That is somewhat of a special circumstance IMO. When a release says it is now able to identify and react to slat gates, then to me at least, that would be a failure.
@@bennybob60 How about reacting to "big metal object in front of windshield"? Doesn't matter if it's a gate or a barricade or a child's slide at a playground. If there is an object in front of it, the car should avoid the object whether it specifically identifies the exact type of object or not. Otherwise, you're telling me the car erroneously identified the gate as part of the road or had no idea what the heck it was seeing and decided to proceed anyway. Neither option is anything but a failure.
Can’t believe people here are making excuses about this
“It’s not common to have a barricade front of a gate” - like wtf. Don’t run into a barricade, does not matter if there is a gate in front, behind, to the side or nowhere near.
I've had great luck with v12.5 but then again I don't have any gates. Glad you reported this. My biggest disengagement problem are potholes. Don't want to chance having tire and/or suspension damage. This is especially a problem after the winter thaw when potholes are found more frequently.
@@edhill8568 I think it sometimes hard to know which road hazards the car will see and react to it which ones it will ignore. I’m always extra alert when I see something like paper trash blowing across the road.
barricades are an interesting failure point, with them being so small on camera. first time watching I didn't see the barricade on the video before you braked, but going back through the video the barricade is visible for at least 3 seconds before, so should be plenty of time to be spotted. Do you have any other barricades like this one in your area you can test? see if the failure is consistent?
@@Elias_Ainsworth92 there are 3 more of these at other entrances. I’ll try when when I get a chance.
That is similar to parking case when you need to enter and exit with barricades. I don’t believe this version (or any prior versions) does that yet. It likely can’t handle emergency vehicles yet either, so I wouldn’t call these a failure. Not trained to handle such scenario/edge case yet. It is a disengagement, so hopefully you have sent this for future training data.
My 12.5 wanted to go straight in a left only lane. Why are those things not happening to Omar. There is probably very little training data with those types of gates, still I was hoping those new HD cams would pick that up better.
The image for the NN is currently not even SD, we far away from real time HD processing. Even on HW4 cars they currently emulating the camera feed to a lower resolution matching the HW3 resolution. And that image then gets further reduced to something like 360x360. But this issues are not a problem in the long term. If its not fixable with improved architecture of the NN stack we could still just improve hardware processing power to make it work.
Pretty unique setup there. Not usual to have both a barricade and a gate just after. Don't see that anywhere I have lived (over 11 states) or worked (39 states). Not surprised it doesn't know how to handle that yet, very rare situation to train for.
@@tonyvelasco5732 the gate and the barricade are linked together. We have barcodes on the car window when you drive up the gate opens first and once fully open the barricade rises. When you leave there is no authentication required and you just have to wait for the gate to slowly open in 12.3.6 usually would see the gate but not always on time so it’s not exactly trustworthy. No version has ever detected the barricade, and even in this case, given the speed the car was driving, had the barricade not been there. It probably would’ve hit the gate too.
Don't drive into stuff... Hard?
That type of barricade has been a known problem for a while now. From what I've read, the FSD display uses the v11 NN stack, which has a series of NN that find the road, identify objects, street signs, etc..
I've been told (though I don't know for a fact) that FSD 12.x abandons that and takes the image as a whole and feeds that into a single NN, this means weird objects like this, people waving and all kinds of hard to classify things should get recognized and handled properly.
I''m curious if anyone else has any insight.
@@joneclegg barricades are not all the same. They can vary in thickness, color, presence or absence of lights or reflectors, etc.. it is not always that important that you know they are a barrier per se, but it is important to know. They are a solid object that you might hit if you don’t stop. Focusing software to identify solid objects in your path would make things safer whether you realize it’s an actual barricade or not.
I'm getting really bad navigational problems with this one, like turning into a residential area 500 feet from the correct turn, or getting into the straight only lane and trying to make a left on a red light from the straight lane, also was suppose to go straight and turned into a parking lot for no reason and started driving around until it eventually pulled toward a parking spot and got stuck.
@@Kev376 interesting. Well, I’m not tested it in 12.5, previous versions have had trouble where lanes are separated by poles that limit lane changes near an intersection. There are a few of these close to where I live and if you are not in the correct left turn lane before these poles start you really cannot turn at that intersection and have to go beyond and then U-turn and come back or in some cases, the Lane you’re in becomes a left turn lane and the polls trap you there such as potentially was going to happen this morning. You can’t really anticipate these until you see them and because they are aligned with your route, they are not visually prominent in a distance. The only real solution is to get into the correct turning lane as early as possible rather than waiting till the last moment.
Excellent edge case! Watch the video in slow motion. @3:19 the speedo says 17mph, @3:21 it has already decelerated to 12mph before the diengagement chime rings. FSD was clearly slowing down before you took over, indicating it saw the gate. I would like to see another attempt
@@ryanslaten I think it either saw the barricade or the gate just too late
@@ryanslaten oh and I tested on another entrance with barricade only and there was a speed bump before it saw and slowed for and after crossing the barricade opened so no contact. For the barricade in the video with the gate behind, it takes longer for the barricade to open because it doesn’t start to open until the gate is fully opened, and that is slow. If the other entrance was just a barricade, the barricade opens immediately upon the sticker being scanned and the speed bump gets you slowed first so maybe is easier to see or maybe the barricade gets out of the way so fast it doesn’t matter
@@DrD57 I see. Thank you for explaining
thank you for your videos! Very informative! As I see fsd is progressing very fast despite the barrier issue. Do you expect Tesla to solve autonomy this or next year?
@@Maksim-he4ej I believe they can do it if they put enough resources into it. I feel like they are understaffed, though leaving widely known issues unaddressed.
Those barriers are very interesting,another thing you don’t normally think of.
I’m guessing it has not been trained on them yet.
Thanks for the video
How many places in USA have a barricade and then a gate - its useful to send the video to Tesla as they can then use more barricade video data.
Tesla does lots of training in simulation, and their simulation might not have many randomly placed barricades and gates yet. I'd imagine they have railroad crossings and parking garages but not much else, and those are quite different environments with different signage compared to this one.
Appreciate your testing! Yeah, I hear that those horizontal swinging barricades are an issue for FSD; apparently the camera is unable to separate it from the surrounding environment. Don’t know how it will be resolved, but hopefully Tesla will find a solution.
I just received 12.5 this morning on my 2024 Model 3. Need to do more testing, but my initial impression is positive. I especially like the stop sign improvement. It’s much more natural and is quicker, even with the need to stop completely (required by NHTSA). Also, I have not yet noticed any lane dancing, but again need more testing. I have noticed maybe a fraction of a second delay when my vehicle is first at a traffic light and it turns from red to green. Not really an issue at all, just an observation when comparing 12.3.6 to 12.5. One other thing, FSD still does not honor No Turn On Red intersections; no doubt something that will be addressed when it is able to read more signs and respond accordingly. I still think that there’s a long way to go before complete autonomy, but 12.5 appears to be a nice step forward.
@@fedderd given I’ve only had two drives so far I agree that the assertiveness in a good way is a good thing. Except for running into the barricade, it does feel more natural. Of course, we will all know more with more time behind the wheel, but definitely ignore the ridiculous hyperbole from Twitter about how great it is and that it is a solved problem.
@@DrD57 Sir, when I read about your complaints I have a feeling that you are already so used to the product that you lost touch with reality. For me, the person who has no Tesla FSD, all these videos, including this one, look like a miracle. Just from watching FSD videos every half year or so, I can see crazy progress and see that it gets so close to fully automated driving.
Hope I didn't insult you, have a great day, and thank you for the video, which was very interesting.
@@RightFiveLeft there has been great progress - to be credible though you have to show the successes and the failures. Too many videos are so full of hyperbole about how great it is that they have no credibility and out viewers at risk of being overly trusting or fsd and that’s dangerous at this point
@@DrD57 you are right. I'm also very interested in failures. Because overall it looks it works fine. That basicly how i found your video. I tried to search for accidents.
Let us know when an update fixes that gate! Thanks!
To be honest, I like to see the failures. So to me its not clickbait. Im very bullish on FSD, but only seeing the good stuff can blind your judgment. What do you think Tesla is more interested in? It ain't the perfect stuff.
@@cheetahrunout I think they are definitely making a lot of progress and the jump to version 12 was a big step but at the same time I don’t think they should push out software that still has no issues if they affect safety at least. I feel like they need to be putting more resources into thisbecause with more coders these issues probably would’ve already been fixed. As it is, they have to hop around from issue to issue dividing their time between new features and fixing bugs.
@DrD57 issue is that its all neutral nets. Therefore isssues like this need training data instead of more coders.
Edit: make sure to accept that your data is shared so Tesla can improve on issues like this. (Perhaps thats mandatory for 12.5, I dont know)
What the heck is a neighborhood gate? Is it to keep the poor people out? This must be the most American thing I've seen today!
@@celeron55 there are many gated neighborhoods to control access to amenities and to know who is in and out and when they come and go. There are usually paid security people to scan ID’s of visitors and residents have car stickers or sometimes codes to enter.
I'm not trying to be bossy, but as we progress further along the FSD journey, it becomes increasingly important to report those edge cases. Thanks for uploading! Its fun to watch the progress.
Barricade before a gate is definitely a fringe case, give it a while. Interested to see how it handles our just gate in community - seems may be ok.
@@travwill I am uploading a test using a second barricade without a gate right now. Should be available in the next 20 minutes or so
Good test. The driving in this and the previous video look good and possibly better than previous versions. Still has problems with gates, but I wasn’t expecting that to be fixed just yet. Thanks for the tests. Which Hardware are you on?
@@anunexpectedfire4062 I took delivery March 19 and it’s a 2024 model Y picked up in Central Florida so I assume is version 4 but have not looked in the pages to find it
All recipients of v12.5 are HW4 Model Ys.
@@tommornini2470 interesting 🤔
Unfortunate to see. But thank you for reporting
Looks like the cameras cant detect the barricades and thats a major issue. I believe its the "vision only" navigation thats affecting the FSD from getting where it need to be. I've seen issues (mostly disengagements) with driving in heavy rain and fog and unmarked roads as well. May be a radar would have helped to fix it. But then, certain people are very adamant, aren't they ? 😉
@@Investchan the cameras have plenty of resolution to see them but the software just isn’t looking for them. It would be helpful if map data included this information like it does for stop signs but the Tesla vision software still needs to see them, recognize them, and react. Another branch of this problem, our cases where the driver has to punch a code or talk to someone to get the gate open, whether via intercom or in person. But absolutely the car does not need to just drive into these things.
Did you report it to Tesla?
@@heromiIes many times yes and I’m sure others have too
when you say collision I think you should mention that FSD failed to recognize a gate. the latter is far less of a problem than a collision with another car or a pedestrian. tesla probably knows about this issue and has not attributed a lot of resources to fixing it yet because it's not as high a priority as making sure it doesn't hit humans or cars.
@@mkkkai one would hope - they really do need to have more people working on the obvious gaps in performance of FSD. Probably no one dies if I crash through the gate but between repairs to the car and the gate could be into the $1000 range or more and missing something right in front of two cameras does not instill confidence. I knew prior versions had this issue so my foot was on the brake pedal but what other blind spots are there that we may not anticipate. When you have serious fan boys or even Elon touting a version that never made it to general release was 10x better without showing data it doesn’t make them more trustworthy or help Tesla at all. All known issues should be fixed with adequate coding resources before another release comes out. Given how much people are paying and have paid over the years it’s not a great idea to skimp on software development like this.
It definitely was going to hit that swinging arm!
@@isonlynameleft I waited as long as I could to see if it would notice and felt like it’s slightly slowed down right before I hit the brakes but yes, it was going to hit it in spite of that and I’m not so sure it would not hit the gate next because they are pretty darn closeand we were moving pretty fast
Thank you for sharing !
At what cost is the public providing their lives for future rewards of a personal or corporate profit? When did it become okay to take a life for a product under testing? Shouldn't this remain outside of public harm until proven safe?
@@stuarthdoblin no company has more public safety scrutiny than Tesla.
@DrD57 When did public safety concerns of potential future harm, such as death become of secondary
Interest to Public Safety Officials over the corporate and personal interests of a Software Product akin to Nazi Drowning concentration camp inmates in order to learn of revivilation (sp.) products to sell on the market for improving health. It's a gruesome NAZI style business model. Most of us would qualm about being the cause of the death of others to enhance and improve our soulless product. Rather than payout millions; build a fifty-million town of roads. Test the product for its safety BEFORE you release said product because it has been proven to be safe before public saturation.
@DrD57 Previous reply deleted, second go: Public Safety concerns speak per product testing implies the known unknown of zero deaths per testing trials. Why is FSD given a pass on civilians deaths with beta testing a known imperfect automated guidance system?
@@stuarthdoblin what deaths? Do human drivers cause zero deaths? Should all human drivers also be banned? Tesla has been pretty clear from the beginning that the driver is responsible for safety and should always have a hand on the wheel and eyes on the road. People who participate in the beta testing have to accept that.
@DrD57 That's my point. I never consented to provide my life on a silver platter to Elon's development of FSD. I was just violated and told to accept the new normal. Not OK.
Calling that a failure is a bit of a stretch. I would have expected to have to disengage for that at this stage of development, just like parking.
@@dluchin1998 parking has been fine the last few versions. Elon said 12.4 was 10x better than 12.3.6 get 12.4 and 12.4.1 and 12.3.2 were never released. The barricade issue is a big deal as it hitting anything right in front of you and they have to be aware of this. I would not release another version until this is taken care of.
@@DrD57 I guess it's your title I take issue with. It sounds like your saying this version is a failure, when really you just had a single disengagement at a place you already knew was an issue.
@@dluchin1998 it was A failure (the behavior). I assume there will be others. If enough people report these issues they will hopefully get more attention. I still don’t understand why speed limit signs that don’t feature black text on a white background or still ignored either is that is certainly a safety issue as well but colliding with something right in front of the car to me would be number one priority. If it can’t see a barricade right in front of you then what else might it miss?
Obviously it was for a clickbait... For me this drive would have been A minus on my score card as I am macro versus micro.... But again I also agree it's technically a failure if you are talking full on robotaxi capability...
@@jh-qb3yb any time a human must take action to prevent an accident. It is a failure for FSD. Many times there are odd behaviors by FSD that are not really a safety issues, such as slowing down unexpectedly, a few miles an hour, and then picking back up speed or unusual lane changes. But anything a human is required to do to prevent an accident is a failure. When fanboy were openly promoting this version and versions past as being nearly perfect this becomes a bigger safety issue as it can encourage unjustified trust and reduced vigilance from drivers. As users who paid a lot for FSD it is not our job to be cheerleaders when we do an evaluation of an update we should show the good and the bad as we see it.
Cool 👍😎360video nice
Elon said this one is specifically to test hardware 4.
Doubt much changed in terms of performance for HW3 or lower.
Your video title is very misleading.
@@deltaechomusicnh555 really? What’s misleading about it? Having to hit the brakes to avoid a collision is not exactly a success is it?
Ya... Stuff like that is going to force Elon to eventually admit the self driving taxi will be limited to certain areas. I think we're a few years away from vision-only systems being able to handle things like that barricade. I also have seen it fail to detect when some objects move while the car is stationary/parked. Im guessing that would also mean the vision system wont detect new objects that get close to the car while its parked.
I think all these issues are very solvable, but I don’t think they have enough people working on it so they hop around from one thing to another and they feel a lot of pressure to push out new releases, including from me because we are all eager to see the new stuff. In general, it’s better to respond things that are not there than to miss things that are.
That is a convoluted idea of a "failure" doc. 🙄
Hitting a gate isn’t a failure?
@@bennybob60 convoluted? Any unplanned deactivation to avoid hitting something is a failure - nothing convoluted about that
@@DrD57 I'm sure we can all put our FSD vehicles in situations it is not trained for. I've yet to hear of FSD stopping for all slat type gates. Have you read that in any of the releases? That is somewhat of a special circumstance IMO. When a release says it is now able to identify and react to slat gates, then to me at least, that would be a failure.
@@bennybob60 How about reacting to "big metal object in front of windshield"? Doesn't matter if it's a gate or a barricade or a child's slide at a playground. If there is an object in front of it, the car should avoid the object whether it specifically identifies the exact type of object or not. Otherwise, you're telling me the car erroneously identified the gate as part of the road or had no idea what the heck it was seeing and decided to proceed anyway. Neither option is anything but a failure.