Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
I drove 200 miles yesterday (AP2) both I 5 south of Stockton and Pacheco Pass. It struggled to stay in the lane on sharper freeway curves at 75mph but no diving for the off ramp. Lane change very smooth but would not cut in close to the front of a Semi. It delays the lane change until more distance in front of the Semi than if I were in control. Still unable to handle twisty 2 lane roads or up and down roads where the lane lines are invisible beyond a hill crest.

As well as several other posts above:
Folks, I do not think there has been any meaningful work/improvement in the TACC/AP2 over the last several updates except for the AEP. Version 17 42.1 94435ee was installed on 10/25. This is the third update in the five weeks that we have had this MX/AP2.?. We have covered 4,600+ miles in these five weeks.

Over the last three updates, the car has continued to have trouble keeping in its lane especially on curving highways such as I-5 from the Siskiyous to the Sacramento Valley, a route four days after delivery of the car and again today. Most times, overtaking a semi results in a harsh hesitation within about of 25 to 50 yard of semi in the adjacent lane. Sometimes it will not read the up corning curve and just want to continue going straight and cross the lane line. Once in a while it will pick up that it is drifting across a lane line, usually the left lane line, and eventually move slowly back into the center of its lane. Most times I have to take over steering.

Once down in the Valley, the MX is much better behaved. Once in awhile it will still want to wander to the left or pick up a phantom and slow abruptly.

We had an MS with AP1 for 12k miles and drove the above route a couple of times. It also had some issues this I-5 route through the mountains but not nearly as many failures where I had to take over steering. Lane changing ability was pretty similar between the two APs.

A couple of posts above mentioned the lane lines seen in IC and how they lagged what the driver could see. Not surprising, probably due to the the limited clues that the AP has to interpret what is coming up versus the clues that we have such as seeing guard rails and road cuts as well as vehicle actions that tell us there is a turn coming up. Imagine driving in fog where we depend on lane lines only. It takes us longer to detect turns.

Over these 4.6+k miles I have also noticed that often times a truck that is 100 + yards ahead of us and in the adjacent lane is shown on the IC as over lapping the IC's lane line but actually I can see that the truck is a foot or more away from its lane line. Is TACC seeing the lanes different than the AP and is calibration off? I can easily imagine that accuracy of where an object is decreases dramatically with distance when coupled with the calculation for projection of the curve on plane that is shifting and tilting constantly especially with highways through the mountains.

AP2.? is a work in progress. Do not take you hands off the wheel.
 
For the first time I noticed a slowing in a turn while on TACC that wasn't due to following a car. For the first time I saw about 50% ( of usual) regen going downhill from 5000 ft in the Sierras with the battery at 255 mi (X 90 D AP2). The many previous trips down Hwy 4 from 5000 ft with a full charge saw no regen until the battery was down to 240 mi. So there may be some favorable changes in this update.
 
I'm not sure if this is related to the new AP2 firmware or not but I was driving back from Carlsbad CA and I saw a white Model X who tried to use auto lane change to the left but it kept on ping ponging between it's own lane. After a while the X took off straight and manually took control and gave up. This was the first time that I've observed it behind auto lane change from behind.
 
  • Informative
Reactions: AnxietyRanger
Wow, I've never actually read that before. Probably a good thing, because I might be unhappy with the current state of autopilot if I had...

It isn't hard to see why people might have a bit "optimistic" views about Autopilot, Tesla obviously wants them to - if that page (Autopilot) is any guide. Reality be damned. That page really is an eye-opener.

But then, many of us got that eye-opener fater October, 2016 when the AP2 we were then pitched has looked and sounded like nothing the 10-11 months of AP2 have later brought us..
 
It isn't hard to see why people might have a bit "optimistic" views about Autopilot, Tesla obviously wants them to - if that page (Autopilot) is any guide. Reality be damned. That page really is an eye-opener.

But then, many of us got that eye-opener fater October, 2016 when the AP2 we were then pitched has looked and sounded like nothing the 10-11 months of AP2 have later brought us..
It certainly puts things in perspective. If you compare the Tesla description of Autopilot to the actual state of either AP1 or AP2, it makes the difference between AP1 and AP2 comparatively insignificant.

On a somewhat related note, I have an AP1 MX P100D loaner while my AP2 MX P100D is at the SC. First, I didn't realize AP1 P100Ds even existed, especially since my OA told me when I was searching for my inventory car that all P100Ds were AP2. It did give me the opportunity to drive AP1 and AP2 vehicles that were otherwise identical on the same roads.

Personally, I can't tell the difference between AP1 in this loaner and AP2 in my car. I haven't experienced any of the erratic swerving behavior or slowdowns that others have, so the overall experience with lane holding and following traffic has been the same for me.

The data I see on the console is obviously different with AP1, specifically vehicles in adjacent lanes and trucks/motorcycles. I have noticed though in areas where I'd normally be wary of engaging AP, that the AP1 vehicle loses sight of the lane markers more often. My AP2 seems to recognize curbs and unmarked road edges better, showing a lane boundary on the console.
 
To give credit where credit is due, we got 2017.42.a88c8d5 yesterday, and I drove about 100 miles of all sorts of highways in our P90D AP2 vehicle. This is by far the best release ever. It's far superior to any previous release in terms of lane control and silkiness. Still did a couple of nosedives for the left lane, but they are fairly predictable after 10+ months of this. Traversing intersections without a nosedive is a refreshing improvement.
 
To give credit where credit is due, we got 2017.42.a88c8d5 yesterday, and I drove about 100 miles of all sorts of highways in our P90D AP2 vehicle. This is by far the best release ever. It's far superior to any previous release in terms of lane control and silkiness. Still did a couple of nosedives for the left lane, but they are fairly predictable after 10+ months of this. Traversing intersections without a nosedive is a refreshing improvement.

Count me in the nice update camp for the 2017.42 update round. We are almost, but now quite, entirely 'silky smooth', on highways at least. So far the best AP experience yet. Holds center lane steady, no major issues around on/off ramp. Have not yet experienced any phantom braking, and land changes are quick, smooth and not jerky.
 
  • Informative
Reactions: EinSV
Covered about 650 miles of mostly highway driving from Friday through Sunday with 2017.42 on a July AP2 MS100D and it was fabulous. It was able to handle the twisty section of the Henry Hudson Parkway north of NYC (which failed quickly on .34 the last time I was in the area.)

It ran like a champ through the horrendous downpours yesterday where at times even I couldn’t decipher the lane lines.

TACC is smooth, Autosteer had no ping ponging. Auto lane change worked great (except for the heavy rain where it wouldn’t work at all.)

There are definitely still some issues with turns involving hills (crests and valleys) which I can understand being a point of confusion if you were to “see” them in 2D (imagine drawing 3D lane lines on a piece of paper with no depth perception.) Obviously work to be done here, but totally solvable, IMO.

Driver and passenger were very comfortable and relaxed the whole time, definitely the best yet.
 
Have used this now on-and-off through Belgium, Holland, Germany and Denmark and it's really good - no heart-stopping, false moves as of yet.

Only type of situation where I don't yet trust it is when driving at speed 80kmph+ and encountering a pedestrian island splitting the lanes. I get the impression it's going to continue straight and climb the kerb, so I take over immediately in this case. At 50kmph or less it seems to be fine.

So things like this:

Screen Shot 2017-10-31 at 11.13.14.png
 
New X owner here. Just finished road trip from NYC to West Virginia and back. .42 was awesome whole way, didn't have to manually take over once on way down (I78-->I81-->I68), even on some local roads. On way back in very inclement weather AP2 handled great, taking a lot of stress out of a difficult driving environment. Can't comment on past releases (had .24 for first week of ownership and even thought that was good...) but am very satisfied with AP2 (as well as X in general) so far.
 
Has anyone noticed, or it just happened in my case, that the vampire drain has completely gone!
Before the update(was 2017.36), my MX100D was dropping 3.4% per day.
After the update five days ago to this 2017.42, the rate is dramatically decreased to only 1km(rated) per day, which is about 0.2% per day, much less than what Tesla said(1%/day).

I had some infra-red photos before showing that there are many hot spots(namely under the falcon wing doors, the dash panel, and at the front), but these hot spots disappeared after the update.
FLIR0734-741.jpg


I guess these hot areas of the doors are for connecting keyfob, but the auto-door feature still operates normally.

One small change though, is that my android app failed to connect to the car under the 'always connected: off' setting(which I used for the old firmware, and remains off for the new firmware for two days). But after I set to on, the app works fine(there is about 40 seconds delay), and the draining rate is the same. My Energy Saving setting is ON for both old and new firmware.
 
  • Informative
Reactions: GoTslaGo
It isn't hard to see why people might have a bit "optimistic" views about Autopilot, Tesla obviously wants them to - if that page (Autopilot) is any guide. Reality be damned. That page really is an eye-opener.

But then, many of us got that eye-opener fater October, 2016 when the AP2 we were then pitched has looked and sounded like nothing the 10-11 months of AP2 have later brought us..

Keep up the good work, AR. I know you had a lot of buddies who would chime in along with you, but looks like you are getting out-numbered now after the last updates. But don't give up on your whine-fest. You may have to dig harder but I am sure you will find plenty to complain to eternity.
 
Last edited by a moderator:
  • Funny
  • Like
Reactions: Turing and EinSV
For the first time I noticed a slowing in a turn while on TACC that wasn't due to following a car. For the first time I saw about 50% ( of usual) regen going downhill from 5000 ft in the Sierras with the battery at 255 mi (X 90 D AP2). The many previous trips down Hwy 4 from 5000 ft with a full charge saw no regen until the battery was down to 240 mi. So there may be some favorable changes in this update.
As far as I can remember it slowed in turns since the day it was first activated.