Navigate on Autopilot is Useless (2018.42.3)

[scottf200]

This guy has started a YouTube channel 'FSD Feature Complete Challenge' and has a couple great routes planned out for 'consistent testing. Consider subscribing/watching to encourage him more. I appreciated his efforts. (Channel: FSD Feature Complete Challenge )

In the second half of this video, he goes thru video examples of these measurements below:

 

[clydeiii]

It's not doing that. It's assisting you and you're driving. Be sure to know the difference or you'll end up like so many other people, angry that "autopilot crashed". Most of us started out like you, pretty psyched about what seems like an impressive system at first. But over time and experience, you start to see the man behind the curtain more and more.

lol ok I mean I use it every day and I have only become more and more impressed

Today was my first morning commute on 40.x that wasn't super rainy or snowy, so I let it do as much of the commute as possible without any interventions. It was by far the best commute I've had so far. Normally during my exit from DC's Beltway onto I-95, the car gets over too soon. Today, for whatever reason, it got over at the exactly perfect spot (meaning, it did exactly what I would have done in that situation). So the question is, are they tweaking the "dumb robot" if-then-else logic, or is the neural net using image recognition to figure out the best time to get over?

 

[dhanson865]

This guy has started a YouTube channel 'FSD Feature Complete Challenge' and has a couple great routes planned out for 'consistent testing. Consider subscribing/watching to encourage him more. I appreciated his efforts. (Channel: FSD Feature Complete Challenge )

In the second half of this video, he goes thru video examples of these measurements below:

score summary from the end of the video to compare against later

 

[arghx7]

The software he was using does not have the new lane change updates (he had 2019.32.x , latest is 2019.40.x) . The current software is noticeably better than the older releases.

 

[emmz0r]

score summary from the end of the video to compare against later

View attachment 489500

There's no FSD yet, so why grade it?

 

[scottf200]

score summary from the end of the video to compare against later

The software he was using does not have the new lane change updates (he had 2019.32.x , latest is 2019.40.x) . The current software is noticeably better than the older releases.

There's no FSD yet, so why grade it?

The KEY POINT is that someone is going to a huge time and effort to do various test cases & measurements ON AN ONGOING test basis. This is not a one and done test. Seriously people.

Obviously he is testing an older release before 40.x came out but the point is to set up test cases and measure various OTAs.

Obviously he knows there is only NoA and not an official functionally complete FSD.

 

[diplomat33]

This guy has started a YouTube channel 'FSD Feature Complete Challenge' and has a couple great routes planned out for 'consistent testing. Consider subscribing/watching to encourage him more. I appreciated his efforts. (Channel: FSD Feature Complete Challenge )

In the second half of this video, he goes thru video examples of these measurements below:

At one point, he talks about a bridge where the speed limit is 30 mph but everyone goes faster. For one, you can adjust your TACC speed to go +5 mph over the speed limit so he could adjust the speed to go faster than 30 to better stay with the flow of traffic. But also, I don't think it is a big negative that AP follows the speed limit. After all, following the speed limit is the law.

I would also point that a lot of his categories are "N/A" basically since those features have not implemented yet. A lot of his missing features will only be added with "automatic city driving". It is also not surprising that AP gets a bad grade so far since it is not true FSD yet. Heck, it is not even "feature complete" yet hence all the missing features in his test .

Bu I am very happy that he is doing this test because tests are a concrete way of measuring progress. It should be fascinating to see how AP improves in this test over time as more FSD is released.

 

[diplomat33]

The youtube channel Black Tesla has a nice updated video of NOA on the latest 2019.40.2.1 update and HW3:

 

[scottf200]

The youtube channel Black Tesla has a nice updated video of NOA on the latest 2019.40.2.1 update and HW3:

Well I thought that was pretty basic highway driving. Skip first 12 minutes. It did show a reasonable NoA feature of moving over 4 lanes from 4 miles to 2 miles from the exit.

 

[DrDabbles]

I think he is genuinely aware that it is 'assisting' him. Like him that assistance it gives me is pretty significant and really makes the cognitive 'work' of driving easier and much much less mundane. I do some traveling to play sports (or other activities) and go to places that are a little farther because AP makes it 'easier'. My son has a long commute and the 'assistance' make a huge difference.

To be honest, tho, your point about getting too comfortable and having a potential 'autopilot crash' is valid. I have to snap back sometimes if I realize I'm not paying enough attention and look down at the map or music search or whatever for 'too long'. Strange things happen out there and by paying attention I've avoided those situations where 'autopilot' definitely would have crashed!

Just in time for the holidays, and my point...

I think @clydeiii should also probably read this, because they're very obviously in the earliest phase.

Understanding the Impact of Technology: Do Advanced Driver Assistance and Semi-Automated Vehicle Systems Lead to Improper Driving Behavior? - AAA Foundation

 

[DrDabbles]

This guy has started a YouTube channel 'FSD Feature Complete Challenge' and has a couple great routes planned out for 'consistent testing. Consider subscribing/watching to encourage him more. I appreciated his efforts. (Channel: FSD Feature Complete Challenge )

In the second half of this video, he goes thru video examples of these measurements below:

I actually have a pretty similar route that I test against, it's about 100 miles, and it's a mix of highway with tolls, city streets, suburban streets, and rural roads. I should probably start recording it rather than just using it myself to gauge changes.

 

[diplomat33]

I actually have a pretty similar route that I test against, it's about 100 miles, and it's a mix of highway with tolls, city streets, suburban streets, and rural roads. I should probably start recording it rather than just using it myself to gauge changes.

Do it.

 

[dhanson865]

The KEY POINT is that someone is going to a huge time and effort to do various test cases & measurements ON AN ONGOING test basis. This is not a one and done test. Seriously people.

Obviously he is testing an older release before 40.x came out but the point is to set up test cases and measure various OTAs.

Obviously he knows there is only NoA and not an official functionally complete FSD.

exactly why I screen shotted the summary score for further comparison. I'm not sure why you thought my post was negative.

 

[alcibiades]

exactly why I screen shotted the summary score for further comparison. I'm not sure why you thought my post was negative.

Pretty sure @scottf200 wasn't saying you were negative. Rather he was highlighting your post to help the Brothers of the Cruciform Sword understand that the logging is ongoing and detailed, setting up info to be compared to "later" (as you pointed out) when the driver does get updates. It's all good!

I mean, except for EAP/FSD. That *sugar* is whack. I use it less and less.

 

[scottf200]

exactly why I screen shotted the summary score for further comparison. I'm not sure why you thought my post was negative.

Pretty sure @scottf200 wasn't saying you were negative. Rather he was highlighting your post to help the Brothers of the Cruciform Sword understand that the logging is ongoing and detailed, setting up info to be compared to "later" (as you pointed out) when the driver does get updates. It's all good!
I mean, except for EAP/FSD. That *sugar* is whack. I use it less and less.

@alcibiades, is correct. Sorry for that confusion and grouping you in. I just appreciate the effort others put in and looking to setup a comparison. I would like to see @DrDabbles, for a comparison to understand what folks look for.

 

[S4WRXTTCS]

This guy has started a YouTube channel 'FSD Feature Complete Challenge' and has a couple great routes planned out for 'consistent testing. Consider subscribing/watching to encourage him more. I appreciated his efforts. (Channel: FSD Feature Complete Challenge )

In the second half of this video, he goes thru video examples of these measurements below:

I really like this, but I think its easier to simply have pass/fail.

If/when I do similar testing I'll have much of the same categories, but I plan on simply doing a pass/fail grade.

Like Lane Splits I'd combine with lane merging (when another lane is merging in) I'd give an F rating on that firmware version because it would recenter on the merge which isn't what it should do. It should stay to the left, and ignore the lane getting bigger.

This guy is being awful friendly to Tesla by assigning D's and C's to stuff clearly not working right.

 

[scottf200]

Like Lane Splits I'd combine with lane merging (when another lane is merging in) I'd give an F rating on that firmware version because it would recenter on the merge which isn't what it should do. It should stay to the left, and ignore the lane getting bigger.

Mine got a LOT better with staying left vs centering in
2019.36.2.1 (vs my prev of
2019.32.12.7).
2019.40.2.1 is even better.

 

[clydeiii]

Just in time for the holidays, and my point...

I think @clydeiii should also probably read this, because they're very obviously in the earliest phase.

Understanding the Impact of Technology: Do Advanced Driver Assistance and Semi-Automated Vehicle Systems Lead to Improper Driving Behavior? - AAA Foundation

I've been using NoA since it came out. When do I get to exit the earliest phase?

 

[MaryAnning3]

I'd love to see a biobib for wk057.

 

[rhumbliner]

While I have the utmost respect for @wk057, I have to admit I’ve experienced none of those problems. The attached map shows the route in Las Vegas that I routinely travel which, I admit, contains only 1 on-ramp, 1 off-ramp, and 1 freeway interchange. My S handles the ramps at a reasonable speed, changes lanes when expected, merges with traffic well (but very timidly) and notifies me when I need to take the wheel. I’ve gotten to the point where, for this route anyway, I trust it implicitly.