Tesla's autopilot, on the road

[johnlocke]

For those who haven't seen it, the following ad showcasing FSD behaving badly has been running on television here as the main theme of a California Senatorial campaign, via IEVS: https://insideevs.com/news/580991/tesla ... candidate/


One wonders how long the FTC will continue to allow Tesla to call the system "Full Self Driving", as that's never been the case.

No self-interest at all right? The business owner makes software that the OEMs rely on for their own ADAS needs. If the OEMs decided that to develop their own solution, then his business dries up.

If you see the ad, pay close attention to the date/time stamps on the clips. Some of them are quite old and they looked to have been cherrypicked for stupid driver reactions. The guy running the ads is running for senator on this single issue and wants a permeant federal ban on FSD. While I wouldn't trust "full self driving" or even auto pilot at this point, it is inevitably coming. All the major manufacturers are working on some form of FSD.

I personally think that it may take another 10 years to perfect FSD. A serviceable FSD with minor glitches is probably 3-4 years out. It's the corner cases that make this problem so hard. It's hard to predict very low probability activity and even harder to test whether the corrective action actually fixed the problem.

[LeftieBiker]

I largely agree with the above. I also agree that calling these early, beta versions "Full Self Driving" is criminally irresponsible. It's like calling Pro Pilot "Automatic Pilot."

[GRA]

"Full Self Driving (beta)" inspires about as much confidence as "Robot Brain Surgeon (beta)".

[GRA]

For those who haven't seen it, the following ad showcasing FSD behaving badly has been running on television here as the main theme of a California Senatorial campaign, via IEVS: https://insideevs.com/news/580991/tesla ... candidate/


One wonders how long the FTC will continue to allow Tesla to call the system "Full Self Driving", as that's never been the case.

No self-interest at all right? The business owner makes software that the OEMs rely on for their own ADAS needs. If the OEMs decided that to develop their own solution, then his business dries up.

If you see the ad, pay close attention to the date/time stamps on the clips. Some of them are quite old and they looked to have been cherrypicked for stupid driver reactions. The guy running the ads is running for senator on this single issue and wants a permeant federal ban on FSD. While I wouldn't trust "full self driving" or even auto pilot at this point, it is inevitably coming. All the major manufacturers are working on some form of FSD.

I personally think that it may take another 10 years to perfect FSD. A serviceable FSD with minor glitches is probably 3-4 years out. It's the corner cases that make this problem so hard. It's hard to predict very low probability activity and even harder to test whether the corrective action actually fixed the problem.

I'm not aware that he's asking for a permanent federal ban on FSD, he wants a ban on FSD as long as its nothing of the sort, with which I'm in total agreement. That's not saying personal interest may not be a factor.

Stupid driver reactions? That's exactly the problem that has to be solved - autonomous vehicles need to be better than average drivers, not below or on par with the stupid ones.

I agree with you on the likely timeline, at least for ADS that can safely replace humans in some limited situations. We do have some dem/val around the country of such systems now, albeit not for cars owned by the general public (who should be the very last to get them, once the tech has been proved). For example, Cruise has had permission to operate in San Francisco with no driver on-board for a while now (and one of their cars got pulled over by a cop recently - see https://electrek.co/2022/04/10/gm-cruis ... -responds/). Waymo has also been operating in Phoenix and I think a couple of other cities.

[johnlocke]

No self-interest at all right? The business owner makes software that the OEMs rely on for their own ADAS needs. If the OEMs decided that to develop their own solution, then his business dries up.

If you see the ad, pay close attention to the date/time stamps on the clips. Some of them are quite old and they looked to have been cherrypicked for stupid driver reactions. The guy running the ads is running for senator on this single issue and wants a permeant federal ban on FSD. While I wouldn't trust "full self driving" or even auto pilot at this point, it is inevitably coming. All the major manufacturers are working on some form of FSD.

I personally think that it may take another 10 years to perfect FSD. A serviceable FSD with minor glitches is probably 3-4 years out. It's the corner cases that make this problem so hard. It's hard to predict very low probability activity and even harder to test whether the corrective action actually fixed the problem.

I'm not aware that he's asking for a permanent federal ban on FSD, he wants a ban on FSD as long as its nothing of the sort, with which I'm in total agreement. That's not saying personal interest may not be a factor.

Stupid driver reactions? That's exactly the problem that has to be solved - autonomous vehicles need to be better than average drivers, not below or on par with the stupid ones.

I agree with you on the likely timeline, at least for ADS that can safely replace humans in some limited situations. We do have some dem/val around the country of such systems now, albeit not for cars owned by the general public (who should be the very last to get them, once the tech has been proved). For example, Cruise has had permission to operate in San Francisco with no driver on-board for a while now (and one of their cars got pulled over by a cop recently - see https://electrek.co/2022/04/10/gm-cruis ... -responds/). Waymo has also been operating in Phoenix and I think a couple of other cities.

When I said "stupid driver reactions" I meant the on screen reactions of the drivers in the clips. For someone who is supposed to be monitoring the car's path, they seemed awfully surprised. In terms of reliability, the baseline goal ought to be safer operation than 90% of the drivers on the road. Not that you never have an accident but that you have far fewer than the average driver. 30% of all accidents are rear end collisions. Active cruise control and automatic braking could reduce that to nearly zero. Driver inattention is another large factor. 90% of all accidents are caused by human error. Removing the driver from the equation seems like a good idea.

[GRA]

I'm not aware that he's asking for a permanent federal ban on FSD, he wants a ban on FSD as long as its nothing of the sort, with which I'm in total agreement. That's not saying personal interest may not be a factor.

Stupid driver reactions? That's exactly the problem that has to be solved - autonomous vehicles need to be better than average drivers, not below or on par with the stupid ones.

I agree with you on the likely timeline, at least for ADS that can safely replace humans in some limited situations. We do have some dem/val around the country of such systems now, albeit not for cars owned by the general public (who should be the very last to get them, once the tech has been proved). For example, Cruise has had permission to operate in San Francisco with no driver on-board for a while now (and one of their cars got pulled over by a cop recently - see https://electrek.co/2022/04/10/gm-cruis ... -responds/). Waymo has also been operating in Phoenix and I think a couple of other cities.

When I said "stupid driver reactions" I meant the on screen reactions of the drivers in the clips. For someone who is supposed to be monitoring the car's path, they seemed awfully surprised. In terms of reliability, the baseline goal ought to be safer operation than 90% of the drivers on the road. Not that you never have an accident but that you have far fewer than the average driver. 30% of all accidents are rear end collisions. Active cruise control and automatic braking could reduce that to nearly zero. Driver inattention is another large factor. 90% of all accidents are caused by human error. Removing the driver from the equation seems like a good idea.

I agree that driver inattention is a huge problem - after all, I avoid being injured or killed about every ten days on my bike now instead of once a month or so, which was the case when I began riding in heavy street traffic more than a half century ago (sheesh, it's really been that long?). The increase is almost entirely due to the proliferation of cell and even more so smart phones, along with there now being an entire generation of drivers who've grown up believing that interacting with said phones takes priority over every other activity, including watching the road and controlling the car. AFAIC, autonomous vehicles that are safer than most drivers can't get here soon enough.

The problem is that current Driver Assistance Systems (DAS) are nowhere near good enough to meet the threshold you suggest, while at the same time they enable if not actively encourage driver inattention (Tesla finally caved and removed the video game while moving feature from their display, after it was clear the NHTSA was actually going to start doing some of their job again). It goes against more than 80 years of human factors research to believe that if you tell drivers that the car can control itself most of the time, but they need to monitor it and be ready to resume control immediately at any time, that they won't let their attention wander and mentally disengage. A few may be able to do so for short periods of time, but most can't/won't. Pretending otherwise, as Tesla does, is to me criminally irresponsible.

We need Automatic Driving Systems (ADS) that meet something like the standard you describe above, not DAS; we can't call a DAS by a name that implies it's an ADS and then turn it loose on the public. IMO DAS that can steer the car should be banned. I'm also not a fan of ACC, as it's another means of encouraging/enabling driver disengagement. With conventional CC I know if I don't watch what's going on I will hit any car I overtake if I don't take action - ACC will prevent that much of the time, but not in critical situations that are most likely to cause a serious or fatal accident, and which require a driver response that is both instantaneous and correct.

[johnlocke]

ACC and anti- collision braking need to be mandatory features on all new cars Just like side view mirrors and taillights. As safety features they rank up there with seatbelts and airbags. I understand your feelings about making it easier for drivers to be inattentive. But the fact is that drivers are already inattentive and anything you can do to reduce accidents and injuries benefits us all.

FSD will take time but at some point you may need a special license to manually operate a vehicle. We'll all be a little safer and driving will be a lot less fun. God only knows what they'll do about motorcycles and bicycles.

[GRA]

ACC and anti- collision braking need to be mandatory features on all new cars Just like side view mirrors and taillights. As safety features they rank up there with seatbelts and airbags. I understand your feelings about making it easier for drivers to be inattentive. But the fact is that drivers are already inattentive and anything you can do to reduce accidents and injuries benefits us all.

FSD will take time but at some point you may need a special license to manually operate a vehicle. We'll all be a little safer and driving will be a lot less fun. God only knows what they'll do about motorcycles and bicycles.

We agree on AEB, but not on ACC. The former is an additional layer of safety on top of the human driver, while the latter, especially when combined with autosteer essentially replaces the human driver. Until such time as an ACC and AEB-equipped car can deal with the fairly common situation of the car it's trailing changing lanes to avoid a stopped vehicle, and the ACC system then identifies that stopped vehicle as such and safely brakes to a halt or changes lanes, IMO it's not safer than a human driver. We have numerous accident reports of DAS-equipped Teslas with both AEB and ACC, rear-ending stopped vehicles on freeways, including stopped emergency vehicles with their lights flashing. Absent statistical data to show a decrease in such accidents, mandating ACC at this time is unjustified IMO. OTOH, AEB has clearly been shown to reduce typical rear-end accidents, and even though those aren't the accidents with the highest probability of severe or fatal outcomes, it's still justifiable to mandate AEB with its current capability.

I expect that the time ADS become common will be about the same point my reaction time has so increased that I'm no longer able to drive safely. As we have a large, aging population. we will definitely need ADS cars, lest seniors suffer major reductions in their ability to socialize. I saw this with my dad, who was very social until he lost his license in his late '80s, and was then dependent on me to drive him around. As I was often unavailable when he wanted to do something, his life shrank. Uber etc. have made things better since then, but are probably too expensive for many seniors for now; removing the most expensive item, the driver, will be a big help, and will also allow people to use their own (ADS) cars, if they wish.

[johnlocke]

ACC and anti- collision braking need to be mandatory features on all new cars Just like side view mirrors and taillights. As safety features they rank up there with seatbelts and airbags. I understand your feelings about making it easier for drivers to be inattentive. But the fact is that drivers are already inattentive and anything you can do to reduce accidents and injuries benefits us all.

FSD will take time but at some point you may need a special license to manually operate a vehicle. We'll all be a little safer and driving will be a lot less fun. God only knows what they'll do about motorcycles and bicycles.

We agree on AEB, but not on ACC. The former is an additional layer of safety on top of the human driver, while the latter, especially when combined with autosteer essentially replaces the human driver. Until such time as an ACC and AEB-equipped car can deal with the fairly common situation of the car it's trailing changing lanes to avoid a stopped vehicle, and the ACC system then identifies that stopped vehicle as such and safely brakes to a halt or changes lanes, IMO it's not safer than a human driver. We have numerous accident reports of DAS-equipped Teslas with both AEB and ACC, rear-ending stopped vehicles on freeways, including stopped emergency vehicles with their lights flashing. Absent statistical data to show a decrease in such accidents, mandating ACC at this time is unjustified IMO. OTOH, AEB has clearly been shown to reduce typical rear-end accidents, and even though those aren't the accidents with the highest probability of severe or fatal outcomes, it's still justifiable to mandate AEB with its current capability.

I expect that the time ADS become common will be about the same point my reaction time has so increased that I'm no longer able to drive safely. As we have a large, aging population. we will definitely need ADS cars, lest seniors suffer major reductions in their ability to socialize. I saw this with my dad, who was very social until he lost his license in his late '80s, and was then dependent on me to drive him around. As I was often unavailable when he wanted to do something, his life shrank. Uber etc. have made things better since then, but are probably too expensive for many seniors for now; removing the most expensive item, the driver, will be a big help, and will also allow people to use their own (ADS) cars, if they wish.

While I agree that ACC when combined with lane keeping isn't a good idea, that wasn't what i was suggesting. Just ACC or even just anti-collision braking. While avoiding a stopped vehicle after the car in front swerves out of the way is a problem, at least ACC would reduce the impact if not avoiding it entirely. I suspect that ACC would have better reaction times than a human driver. A far more likely scenario is stop and go freeway driving where ACC is a real boon and even anti-collision braking would help.

While DAS equipped cars have hit stationary objects, it's also true for ordinary drivers. Whenever a Tesla hits something it seems to be a big deal. Police, firemen, and ambulance drivers regularly get sideswiped by ordinary drivers who ought to watching out for them but don't. While it has been said that Good is the enemy of Perfection, it's also true that good is better than nothing.

[LeftieBiker]

When anyone plows into a stopped car in the road at full or near full speed, it's a big deal. Teslas are slightly more newsworthy because there is usually someone literally sleeping at the wheel at the time.