I'm a huge fan of Elon Musk's companies—notably SpaceX, but also Tesla—but Musk himself is in my view becoming an increasingly brash & unlikeable person thanks to his ridiculous antics, bullshit...
I'm a huge fan of Elon Musk's companies—notably SpaceX, but also Tesla—but Musk himself is in my view becoming an increasingly brash & unlikeable person thanks to his ridiculous antics, bullshit Twitter rants, and excessively weasel-worded claims around what his products & companies can do.
This latest gaff continues the trend of his transformation into an angry Howard Hughes-esque character for me, at least.
On the bright side, after the whole underground highway project goes bust, we'll have the opportunity to repurpose those tunnels for what they should have been in the first place: cheap metros.
On the bright side, after the whole underground highway project goes bust, we'll have the opportunity to repurpose those tunnels for what they should have been in the first place: cheap metros.
That article doesn't consider the fact that car utilisation would drop. Of course splitting busses into taxis will increase congestion. But merging individual traffic into smallish busses would...
That article doesn't consider the fact that car utilisation would drop. Of course splitting busses into taxis will increase congestion. But merging individual traffic into smallish busses would cause the inverse. Now just to solve the engineering problem of efficiently finding people to share rides with and dispatch autonomous vehicles.
He got into a spat with the Existential Comics guy a while back, which resulted in this comic that I think is hilarious: http://existentialcomics.com/comic/other/17
This seems like an extremely dangerous claim for Elon to be throwing around. As the article mentions, they've previously had to repeatedly try to emphasize that Autopilot is not for self-driving...
This seems like an extremely dangerous claim for Elon to be throwing around. As the article mentions, they've previously had to repeatedly try to emphasize that Autopilot is not for self-driving (despite a lot of their marketing implying that it is). For example, in their blog post about the crash last March, they blame the driver for having their hands off the wheel for six seconds and expecting that Autopilot wouldn't run into a lane divider (that I remember reading it had previously been able to navigate repeatedly).
This is one of the major issues with self-driving cars—a state where they're "quite good" may actually be worse than not having it at all, because it makes people complacent when they get used to the car being able to drive itself in most situations. They stop paying full attention because they usually don't need to do anything, and then if they actually do need to intervene, they're not ready.
While I agree, to be fair these were the stats offered in the original blog post: 200 Autopilot trips a day, past that same exact spot. Something went very wrong in this one case. I would assume...
While I agree, to be fair these were the stats offered in the original blog post:
Our data shows that Tesla owners have driven this same stretch of highway with Autopilot engaged roughly 85,000 times since Autopilot was first rolled out in 2015 and roughly 20,000 times since just the beginning of the year, and there has never been an accident that we know of. There are over 200 successful Autopilot trips per day on this exact stretch of road.
200 Autopilot trips a day, past that same exact spot. Something went very wrong in this one case. I would assume the engineers have invested a ton of time in figuring exactly how and why.
I won a court case for a very bad traffic accident involving my company truck (I wasn't driving, but own the company). It involved a driver being distracted from the road for less than three seconds—which was determined by traffic light timing during the accident reconstruction. Six seconds is a significant time to ignore the road. I know that's counter to Elon's claim, and he should never have made it, but until it truly is "Autopilot," the driver needs to remain attentive to the road.
This is one of the major issues with self-driving cars—a state where they're "quite good" may actually be worse than not having it at all, because it makes people complacent when they get used to the car being able to drive itself in most situations.
Saw this in a recent youtube video. The guy was trying out Autopilot, and at the beginning of the video he's very worried and scared to let the car steer and brake for him. By the end of the video, he's quite comfortable and talking to the camera in the passenger seat, barely paying attention to the road. It is indeed a major issue while they're in this state of, as you say, "quite good."
Navigate on Autopilot just isn't quite ready for prime time yet, and certainly should not be considered any sort of FSD system. As a Model 3 owner, I actually quite rarely enable Navigate on...
Navigate on Autopilot just isn't quite ready for prime time yet, and certainly should not be considered any sort of FSD system. As a Model 3 owner, I actually quite rarely enable Navigate on Autopilot on my car.
I'm kinda surprised that Tesla is ready to turn off stalk-to-confirm for Navigate on Autopilot, too. It's such shit where I live—it throws on the blinkers at random unnecessary times, wants to merge into slower lanes, wants to start merging for the exit 4 miles before the exit, including into lanes which will soon end. These are all really dumb decisions which leave me never using Navigate on Autopilot unless I know it's not going to frustrate me (which it does).
As a specific example, one of the freeways on my way home drops from 6 lanes to 3 lanes in just about a mile stretch, and Navigate on Autopilot always just wants to sit in the lane that's always has cars merging into it as the lanes end, but the autopilot system is not good enough at handling the zippering without sudden heavy braking—I frequently hit the "early collision avoidance" beeps on routine merges with Autopilot enabled, including once today on my drive to work.
(Important context for non-Tesla owners: the Autopilot system as a whole is quite excellent, and I trust the car to drive more than I trust myself driving it, and probably 90% of my miles in this car are on AP. The newest feature, "Navigate on Autopilot", just isn't up to the same quality levels as the TACC and Autosteer systems, but it will get better with time.)
AP is the combination of two very well refined and tested features: Traffic Aware Cruise Control (TACC) and Autosteer. TACC should be implicit (keeps a set speed, stops and goes in traffic), but...
AP is the combination of two very well refined and tested features: Traffic Aware Cruise Control (TACC) and Autosteer. TACC should be implicit (keeps a set speed, stops and goes in traffic), but Autosteer is just lane keeping, not a navigational feature—the driver still has to make most decision. This combo works just about flawlessly—you hear occasional the stories of issues, but they're few and far between. I can say with confidence based on my experience driving with AP that, if you're paying attention to the road, the symbiotic relationship of Driver+AP makes you far safer, and AP has more safely reacted to things I felt I would have missed or messed up.
Navigate on Autopilot is an additional complexity layer on top of the AP stack: this gives it the ability to do highway navigation: interchanges, offramps, and lane merges (moving left to pass cars, or shifting to the right for highway exits). As you'd get from my rant above, Navigate on Autopilot doesn't feel....finished—it might be okay for most, but my commute just has a high preponderance of issues for Navigate on Autopilot.
I'm a huge fan of Elon Musk's companies—notably SpaceX, but also Tesla—but Musk himself is in my view becoming an increasingly brash & unlikeable person thanks to his ridiculous antics, bullshit Twitter rants, and excessively weasel-worded claims around what his products & companies can do.
This latest gaff continues the trend of his transformation into an angry Howard Hughes-esque character for me, at least.
On the bright side, after the whole underground highway project goes bust, we'll have the opportunity to repurpose those tunnels for what they should have been in the first place: cheap metros.
They're 14 feet across, which is wider than the deep level tubes in the London Underground, which are only 11'8".
That article doesn't consider the fact that car utilisation would drop. Of course splitting busses into taxis will increase congestion. But merging individual traffic into smallish busses would cause the inverse. Now just to solve the engineering problem of efficiently finding people to share rides with and dispatch autonomous vehicles.
He got into a spat with the Existential Comics guy a while back, which resulted in this comic that I think is hilarious: http://existentialcomics.com/comic/other/17
This seems like an extremely dangerous claim for Elon to be throwing around. As the article mentions, they've previously had to repeatedly try to emphasize that Autopilot is not for self-driving (despite a lot of their marketing implying that it is). For example, in their blog post about the crash last March, they blame the driver for having their hands off the wheel for six seconds and expecting that Autopilot wouldn't run into a lane divider (that I remember reading it had previously been able to navigate repeatedly).
This is one of the major issues with self-driving cars—a state where they're "quite good" may actually be worse than not having it at all, because it makes people complacent when they get used to the car being able to drive itself in most situations. They stop paying full attention because they usually don't need to do anything, and then if they actually do need to intervene, they're not ready.
While I agree, to be fair these were the stats offered in the original blog post:
200 Autopilot trips a day, past that same exact spot. Something went very wrong in this one case. I would assume the engineers have invested a ton of time in figuring exactly how and why.
I won a court case for a very bad traffic accident involving my company truck (I wasn't driving, but own the company). It involved a driver being distracted from the road for less than three seconds—which was determined by traffic light timing during the accident reconstruction. Six seconds is a significant time to ignore the road. I know that's counter to Elon's claim, and he should never have made it, but until it truly is "Autopilot," the driver needs to remain attentive to the road.
Saw this in a recent youtube video. The guy was trying out Autopilot, and at the beginning of the video he's very worried and scared to let the car steer and brake for him. By the end of the video, he's quite comfortable and talking to the camera in the passenger seat, barely paying attention to the road. It is indeed a major issue while they're in this state of, as you say, "quite good."
Navigate on Autopilot just isn't quite ready for prime time yet, and certainly should not be considered any sort of FSD system. As a Model 3 owner, I actually quite rarely enable Navigate on Autopilot on my car.
I'm kinda surprised that Tesla is ready to turn off stalk-to-confirm for Navigate on Autopilot, too. It's such shit where I live—it throws on the blinkers at random unnecessary times, wants to merge into slower lanes, wants to start merging for the exit 4 miles before the exit, including into lanes which will soon end. These are all really dumb decisions which leave me never using Navigate on Autopilot unless I know it's not going to frustrate me (which it does).
As a specific example, one of the freeways on my way home drops from 6 lanes to 3 lanes in just about a mile stretch, and Navigate on Autopilot always just wants to sit in the lane that's always has cars merging into it as the lanes end, but the autopilot system is not good enough at handling the zippering without sudden heavy braking—I frequently hit the "early collision avoidance" beeps on routine merges with Autopilot enabled, including once today on my drive to work.
(Important context for non-Tesla owners: the Autopilot system as a whole is quite excellent, and I trust the car to drive more than I trust myself driving it, and probably 90% of my miles in this car are on AP. The newest feature, "Navigate on Autopilot", just isn't up to the same quality levels as the TACC and Autosteer systems, but it will get better with time.)
If you don't mind explaining, what's the difference between NoAP and AP? I thought AP was able to navigate already?
AP is the combination of two very well refined and tested features: Traffic Aware Cruise Control (TACC) and Autosteer. TACC should be implicit (keeps a set speed, stops and goes in traffic), but Autosteer is just lane keeping, not a navigational feature—the driver still has to make most decision. This combo works just about flawlessly—you hear occasional the stories of issues, but they're few and far between. I can say with confidence based on my experience driving with AP that, if you're paying attention to the road, the symbiotic relationship of Driver+AP makes you far safer, and AP has more safely reacted to things I felt I would have missed or messed up.
Navigate on Autopilot is an additional complexity layer on top of the AP stack: this gives it the ability to do highway navigation: interchanges, offramps, and lane merges (moving left to pass cars, or shifting to the right for highway exits). As you'd get from my rant above, Navigate on Autopilot doesn't feel....finished—it might be okay for most, but my commute just has a high preponderance of issues for Navigate on Autopilot.
Thanks for explaining.