My silky smooth 320 mile road trip with almost 100% AP control

[CameronB]

Last weekend, I drove from the Bay Area to SoCal (Ventura/Oxnard area) and armed with software version 2017.28.c528869 on my (early June delivery) Model S, I did almost the entire trip with AP2. I marveled at how well it did, how smooth it was, and how little I actually drove. I knew I had barely driven the trip but I almost couldn't believe it myself. Rather than have a "seat of my pants" subjective evaluation, I decided that heading back to the Bay Area (same route - US 101), I would do my best to document the specifics. To do this, I mounted my phone in such a way that I could quickly snap the camera and capture my nav and energy screen. I knew additionally that my iPhone would capture full lat/long coordinates so that I could later review the location for more accuracy.

The diary of my return trip will be shown in the next post. But first some disclaimers...

DISCLAIMERS:

1. If you have read my other posts, you may see that I have had a pretty positive experience with AutoPilot (AP2) even before the "silky smooth" updates. No, it’s not perfect and there have been a handful of head-scratching moments (“What is AutoPilot thinking/doing?!?”) since I’ve owned the car. I know and understand that some people have had very bad experiences. I sympathize and wish there was not as much variability between AP2 users. So if you are reading this and you are one of those persons that thinks “AutoPilot tries to kill me every 3 minutes”…please note that I am not invalidating your experiences. You have your experiences (good, bad, or murderous!) and I have mine. I respect and believe yours. Please do the same for me.
2. For those who have AP1 and read this and say, “AP1 can do just as well or better”, I say “Awesome!”. I am not trying to make this an AP1 vs AP2 thread. I have no experience with AP1. I have only even had my Tesla for about 2 months, so AP in any form is all new to me. I have nothing but love for my AP1 brothers and sisters out there! Like above, I only know what I have used (AP2 on my car alone).
3. I list the disengagements below on my 320 mile trip. I did NOT count disengagements for a) exiting a highway or b) for a problem I encountered sometimes whereby auto lane change would not work. At first I thought this was random, but I later was able to figure out in advance when it would and when it would not auto-lane change for me. So while I typically used auto-lane change to keep AP2 as in charge as much as possible, there were some times auto lane change did not work and in those cases, I momentarily disengaged, changed lanes and then re-established AP. I don’t count those lane changes as disengagements.

 

[CameronB]

Diary of My Trip

Started AP in Ventura, CA.

First disengagement: 31 miles after start of trip. In Santa Barbara. Why? AP disengaged itself seemingly because I passed the indicated offramp to Hwy 154 and remained on 101. Even though 154 was technically a shortcut, it involved a steep mountain pass and I imagined using considerable energy to crest it. Reset disengagement miles.


Second disengagement came after another 36 miles. Lompoc. Just a few miles before Buellton supercharger offramp. Reason: Driving through an approximately 3/4 mile long-ish construction zone (empty - no workers at that time). Car was getting a little too close for comfort to some cones. Maybe would have been OK but didn't want to risk it. Completely aware that AP should not be used in construction zones. But it had been doing so great all trip and there were just some cones constricting slightly the width of the lane so I thought I’d push it a tad bit. Of course I was hyper vigilant and gave AP a short lease when it meandered close to said cones. It made it through about a half mile before I took control. So In reality, this is not really AP’s “fault” as it is known to not handle construction zones.

3rd disengagement came in Santa Margarita after 70 miles of uninterrupted AutoPilot(!). That’s approximately 1 hour of self-driving! Super impressive. And during this time, it was handling quite windy stretches at 70-75 mph. Having the car handle itself in stop and go traffic on a highway is one thing. But having it hurdle into curves at 75mph is a whole other ball game! Was it scary - yes, a bit! Was I hyper, hyper vigilant ready to take control of both wheel and brake in a split second - hell yes! But though the pucker factor was indeed high, it handled the curvy roads at speed very, very well. And many of them were not gentle turns but rather aggressive ones. Always stayed in it’s lane. A few times got close to the botts dots, but I don't think it ever hit them. What can I say…it was damn smooth! After handling those high speed curves, the disengagement came when I crested a large steep hill. AP disengaged itself almost exactly at the summit. I think the sensors momentarily lost track of the road as it came over the top. I took over. Uneventful but did surprise me a little.


4th disengagement came only about a mile after the above. Heading down the steep somewhat windy mountain around Santa Margarita. Reason: car started to swerve a bit towards the adjacent lane. Didn't go over the divider, but I thought it might so I brought it back into my lane manually and then re-engaged.


5th disengagement came just after Paso Robles. Another car swerved towards me quickly from the side. I took control of the car and no problem. Perhaps AP would have handled it, but the erratic driver’s speed and sudden swerving made me not want to take any chances.


6th disengagement came a staggering 106 miles later! That’s right, from Paso Robles to about 10 miles south of Gilroy (San Juan Bautista) AP did all of the driving. Probably a shade under an hour and a half. Unfortunately I don’t recollect what caused this disengagement. At my previous supercharger stops in Buellton and Atascadero I made notes but sadly when I got to my last top-off at Gilroy, I didn’t continue my notes.

7th disengagement came just after I had stopped for a top-off at the Gilroy Supercharger. About 11 miles after the 6th one. Like above, I didn’t note what caused the disengagement but I have a vague recollection that it was another car doing something concerning. But I can’t fully remember the details.

Continued another 39 miles to Los Altos without any further disengagement.


Realizing that after every one of the 7 disengagements, I almost immediately re-engaged it, the car literally drove itself 320 miles or 99% of the highway portion of the trip! I felt like I was living in the future for real.

And almost all of that time was silky smooth. Probably had maybe 5 or 6 wobbly or ping-pong moments. It was so smooth that the few times it did ping-pong on me, it was shocking! It was the best AP2 experience I have had by far and it was SO GOOD!


Remembering that I did manually handle offramp/onramps and the couple of blocks between the highway and the 3 Superchargers I used…I’d estimate that out of the 5 hours for the trip (ignoring the parked hour of combined supercharging time), I actually drove the car about 15 minutes total - for all needs combined. That means for the entire 320 mile, 5 hour journey, AP2 drove for 95% of the trip and I only drove a total (for any reason!) of 5%. To say I am impressed is an understatement.

 

[CameronB]

Here is a map of the 7 disengagement points. #3 and #4 are separated only by a mile so they are overlapped here in the map.

 

[oktane]

Very informative post, thanks for chronicling your experience.

 

[Tangible]

@CameronB Thanks for this very useful data and analysis. It seems clear that AP2 has evolved into a competent Adaptive Cruise Control /Lane-Keeping Assistance equipped vehicle, able to perform well for long stretches of highway driving. In this, based on your report, it has now reached parity with the ACC/LKA systems found on other high-end cars for the last several years.

By eschewing industry-standard terminology and grandiously calling these useful and now-common capabilities "Autopilot", and by allowing the system to be used on local streets where its behavior is erratic and dangerous, Tesla has created expectations that cannot be met. I think that's the reason for the different perceptions that you so respectfully referenced. I look forward to someday having a self-driving car that can handle construction zones, off-ramps, children darting out from between parked cars, etc. Tesla and other manufacturers are probably ten years away from that, though Tesla lies about both its current and imminent capabilities. In the meantime, we can enjoy the limited but helpful benefits of what we do have for freeway/highway driving.

 

[azred]

Cameron, that was very informative. Although your post was intended to document AP, I think you also gave me one additional, big nudge towards the larger battery on my Model 3. As a former Santa Barbara resident I used to ride my motorcyle much more than drive a car and a favorite route was over 154 to points North. I would never have thought of skipping 154 in favor of boring, but flatter 101. (Skipping 154 also means skipping Cold Spring Tavern on Stagecoach Road just off the 154 -- and that would be a mistake.)

 

[calisnow]

So if you are reading this and you are one of those persons that thinks “AutoPilot tries to kill me every 3 minutes”…please note that I am not invalidating your experiences.

This is a radical concept for some people here. Good luck.

You have your experiences (good, bad, or murderous!) and I have mine. I respect and believe yours. Please do the same for me.

You're quite the optimist.

I am not trying to make this an AP1 vs AP2 thread.

Life is what happens while you're making other plans.

 

[3Victoria]

parity with the ACC/LKA systems found on other high-end cars for the last several years.

Do tell.

 

[calisnow]

...by allowing the system to be used on local streets where its behavior is erratic and dangerous

Data please - you have accident statistics to compare with local road accident rates of human drivers?

 

[mdevp]

Sad OP has to write a half page disclaimer that looks like a lawyer wrote it just for writing a positive AP2 post....

 

[MarcusMaximus]

Sad OP has to write a half page disclaimer that looks like a lawyer wrote it just for writing a positive AP2 post....

In all fairness, that's at least no worse than people having to worry about being called out as fake for reporting hardware issues they have with their cars(e.g. the suspension thread from this last week)

 

[MarcusMaximus]

@CameronB Thanks for this very useful data and analysis. It seems clear that AP2 has evolved into a competent Adaptive Cruise Control /Lane-Keeping Assistance equipped vehicle, able to perform well for long stretches of highway driving. In this, based on your report, it has now reached parity with the ACC/LKA systems found on other high-end cars for the last several years.

With lane changing capability, of course.

 

[jeffro01]

I think it's safe to say that AP2 is improving although not likely fast enough to remotely make up for the massive gap between expectations set by Elon and reality...

Jeff

 

[ecarfan]

First disengagement: 31 miles after start of trip. In Santa Barbara. Why? AP disengaged itself seemingly because I passed the indicated offramp to Hwy 154 and remained on 101. Even though 154 was technically a shortcut, it involved a steep mountain pass and I imagined using considerable energy to crest it.

Cameron thanks for your terrific post. I'm glad you had such a positive experience with AP.

Regarding your thought on why AP disengaged at the 154 offramp: they way I interpret what you wrote it sounds like you believe that AP disengaged because the car nav route recommended using 154 but you elected to stay on the 101. Do I have that right?

If so, I believe it is true that at this time AP does not follow the car nav route. They aren't connected. Yet. When you have AP on, if a turn or on/off ramp is required the driver (human) has to take over and do that. So when you are heading south on 101 past Los Alamos with AP on and the 154 offramp was in view, to take that offramp you would need to disengage AP and make the exit turn manually.

My apologies if I misinterpreted your post.

By the way, the route map image you posted shows you taking 154. But you said you did not use that highway, you stayed on the 101, right?

 

[CameronB]

Cameron thanks for your terrific post. I'm glad you had such a positive experience with AP.

Regarding your thought on why AP disengaged at the 154 offramp: they way I interpret what you wrote it sounds like you believe that AP disengaged because the car nav route recommended using 154 but you elected to stay on the 101. Do I have that right?

Mostly. Let me try to explain it better. I am aware that AP would not exit US 101 and then navigate a few surface blocks and onto 154.

The navigation in the car was telling me to get off 101 and go 154 over the mountain as the route to get me to Buellton (supercharger). I have travelled this area many, many times before. In the moment I felt like although i had enough energy to do as the Nav said, I figured if I stayed on "flat, boring 101" it would use less overall energy and I could supercharge that much faster.

So while AP will not follow the Nav, when I failed to (manually) exit 101 and passed the off ramp the Nav had told me to take, it seemed that AP didn't "like that" and disengaged. As if it noticed that I didn't exit 101. I suppose it could be entirely coincidental that AP disengaged by itself just about 100 ft past the offramp (to 154) that the Nav was telling me to exit...but it seemed that they were related.

 

[CameronB]

By the way, the route map image you posted shows you taking 154. But you said you did not use that highway, you stayed on the 101, right?

Yes the map does show taking 154 (even though I didn't drive it) but that's because that's what both Google Maps and the on board Nav (which uses google maps) both plotted for the route. For purposes of making my homemade disengagement map, I didn't take the time to figure how to force Google Maps to make a route to avoid 154. And also didn't bother to fix it in Photoshop. So yeah the route I took was slightly different than my graphic. 101 all the way north from Ventura until I reached the Bay Area. Good catch!

 

[CameronB]

Sad OP has to write a half page disclaimer that looks like a lawyer wrote it just for writing a positive AP2 post....

Ha! It's ok. I know some people are very opinionated AP1 vs AP2 or have read the reports of how AP can't go more than a few minutes or miles/KMs without doing something dangerous. I have read all those posts. I was just blown away at the performance and wanted to share my experience in my car. I was awestruck that I had just driven in a car that drove 4:45 of a 5 hour trip and wanted to share my feeing of living in the future.

 

[CameronB]

Cameron, that was very informative. Although your post was intended to document AP, I think you also gave me one additional, big nudge towards the larger battery on my Model 3. As a former Santa Barbara resident I used to ride my motorcyle much more than drive a car and a favorite route was over 154 to points North. I would never have thought of skipping 154 in favor of boring, but flatter 101. (Skipping 154 also means skipping Cold Spring Tavern on Stagecoach Road just off the 154 -- and that would be a mistake.)

Glad you liked it. I always enjoy 154 and this was amongst the very few times I opted to skip it. Mileage wise, 154 is shorter path to Buellton from Santa Barbara. I assume that's why the (Google Maps powered) Nav selected that route. But since Nav is integrated along with SC stops and energy usage, I'm curious if the route should have told me to stay on 101 in the interest of energy savings / shorter charge time. I'd kind of think it should have. Even though I had enough power to get over the mountain with a buffer to spare, I was interested in saving power by avoiding the mountain mainly to reduce the charge time at Buellton.

Unfortunately I didn't make notes about how much my staying on 101 vs 154 netted me in terms of charging time. I have a feeling it was net positive but not 100% sure.

 

[Saghost]

What happens when AP "disengages itself"?

With my AP1 X, there are only two types of disengagement I've seen - either I do something to override it, or it gives me red hands and loud beeping (which does happen for reasons other than ignoring hold alerts or max speed, occasionally.)

The force on the wheel for it to decide I'm taking over seems to drop when the car isn't confident, but it's still definitely me steering away from its path.

 

[MarcusMaximus]

Glad you liked it. I always enjoy 154 and this was amongst the very few times I opted to skip it. Mileage wise, 154 is shorter path to Buellton from Santa Barbara. I assume that's why the (Google Maps powered) Nav selected that route. But since Nav is integrated along with SC stops and energy usage, I'm curious if the route should have told me to stay on 101 in the interest of energy savings / shorter charge time. I'd kind of think it should have. Even though I had enough power to get over the mountain with a buffer to spare, I was interested in saving power by avoiding the mountain mainly to reduce the charge time at Buellton.

Unfortunately I didn't make notes about how much my staying on 101 vs 154 netted me in terms of charging time. I have a feeling it was net positive but not 100% sure.

Keep in mind that as your battery level gets lower, the charging speed for those lower levels increases. So it'll take longer to charge back up, but by a less than linear factor.