Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Software version 17.34 2448cfc

This site may earn commission on affiliate links.
Heh I cheated. I bought myself a tripod for my android phone and shoved all the legs under the headrest and set the headrest down low to grip the legs of the tripod in place. Cost me the whole of $17 and I didn't have to damage anything in the car to "mount" it.
IMG_20170909_093957.jpg IMG_20170909_094025.jpg
Here you go, ghetto phone cam mounting :)
 
After driving around the last two days, two of my pet peaves are still present:

1) Grey font on night-time map display that can hardly be distinguished from the grey map background;

2) I have to manually reload my driver profile to get my headrest in its "recorded" position. The seat and everything else follows my driver profile and resets when I enter the vehicle.
 
Side vehicles do show up, but only when you put your indicator on.
I have had this update for about a week now and have tested EAP on freeways and divided roads. Definite improvement in lane centering, handling curves, auto lane changes are noticeably smoother, and like @conman I also see that now when you initiate an auto lane change by using the turn signal stalk if there is a vehicle in an adjacent lane and somewhat ahead of you the drivers display will show it. I have never seen that EAP behavior until this latest update.

It is unclear to me if that new behavior -- showing a vehicle in an adjacent lane when initiating an auto lane change -- relies on the forward radar or the ultrasonics or both. I have seen that behavior when a vehicle is in an adjacent lane and it's rear bumper is only about 10 ft from my front bumper, so I'm guessing that would be because the car used the ultrasonic sensors, I'm not sure that the forward radar beam is that wide that close to the car.
 
I really wish Tesla gave more detailed update notes so we actually knew what to look for. Guessing what changed and reporting what feels different seems counter-intuitive.
I think Tesla has made the correct decision not to release the level of detail you are requesting. That is because end users (all of us here) do not do quantitative testing of Auto Pilot. We all just report our subjective seat-of-the-pants impressions which can be widely different responses to the exact same software.

Tesla does quantitative testing to gather hard data about Auto Pilot performance before releasing updates. These updates are then subjectively evaluated by end users in a huge number of unique situations and Tesla cannot possibly test in advance for all of those situations. End users will inevitably experience system shortcomings and failures. That is why end users have to stay alert to surrounding conditions and keep their hands on the wheel, as Tesla strongly recommends.

All that said, it seems clear that currently Tesla has the best driver assist features on the market. To those who say that Auto Pilot is "dangerous" or "unsafe", my response is "Then don't use it and wait for it to improve because the last two years have shown that it does improve over time".
 
I was very happy with the previous 17.28 release but since upgrading to 17.34 not so much. We just got back from a weekend trip 3 hours drive away and during the drive my wife told me to stop using autosteer because it was ping ponging too much and bothering her. It was a windy day though but I don't recall a time it has ping ponged as consistently. I have not travelled that stretch of highway before with the X and autosteer sometimes behaves better on certain highways over others. I am hoping this is an anomaly and it it gets smoother. I did notice it seems better (actually almost usable) on local roads when there is only 1 line even if it is only the right hand shoulder line.
 
I think Tesla has made the correct decision not to release the level of detail you are requesting. That is because end users (all of us here) do not do quantitative testing of Auto Pilot. We all just report our subjective seat-of-the-pants impressions which can be widely different responses to the exact same software.

In my day job we've literally accidentally shipped the wrong set of release notes with an update and still amazingly see Internet commenters commenting about behavior changes that we didn't even make.

I completely agree. What release notes say is basically irrelevant in terms of minor changes. The release notes should brief you on features that affect you (e.g. introducing you to new functionality or calling out changed functionality that could otherwise surprise you while driving). Saying that it has minor bugfixes/improvements is entirely appropriate for everything I see going from 2017.28 to 2017.34.
 
  • Like
Reactions: LosAltosChuck
I was very happy with the previous 17.28 release but since upgrading to 17.34 not so much. We just got back from a weekend trip 3 hours drive away and during the drive my wife told me to stop using autosteer because it was ping ponging too much and bothering her. It was a windy day though but I don't recall a time it has ping ponged as consistently. I have not travelled that stretch of highway before with the X and autosteer sometimes behaves better on certain highways over others. I am hoping this is an anomaly and it it gets smoother. I did notice it seems better (actually almost usable) on local roads when there is only 1 line even if it is only the right hand shoulder line.
I realize you were happy with the very limited 17.28 release and I agree it did it's narrowly focused job better than the current release(s).

However, I think what is happening is AutoPilot is being given more responsibilities on it's slow way to FSD. Therefore, it needs to be fed more data in order to be able to handle more circumstances than before and hence it lost the narrow focus.

Sure, right now it's driving like a first-time teenager who is going through sensory overload, but it is necessary and it will get better.
 
  • Like
Reactions: Duckjybe
FWIW, my experience with 2017.34 has been fantastic so far. Did an 80 mile trip on the highways this morning with literally zero interventions other than entering/exiting the highway.

Lane changes are definitely smoother, lane holding seems more correct on curves, and most importantly to me, 2-car-ahead radar TACC is finally working again, which avoids frighteningly late braking when your lead car is a tailgater in stop and go traffic.
 
.34 2448cfc feels like a step back from .32 ...especially in the carpool lane on the 405 in traffic....AP2.
Having driven quite a while in it now I'm inclined to agree with you. I trust it less and feel obliged to hold the steering wheel very firmly while I have autosteer on now; at least I no longer get nagged about holding the wheel...
 
  • Funny
Reactions: alcibiades
Has anyone noticed a problem with Summon since this upgrade? I use it a lot backing the car out of my garage, and twice since the upgrade it has either not responded at all, or told me I was too far away from the car. In the past I could activate Summon from my second story bedroom.
 
Has anyone noticed a problem with Summon since this upgrade? I use it a lot backing the car out of my garage, and twice since the upgrade it has either not responded at all, or told me I was too far away from the car. In the past I could activate Summon from my second story bedroom.

I also use it everyday to backout, but while close to the car, but not close enough for it to see me and stop early.
Yes, I see some slight difference in behavior now, but cannot quantify yet. The long fob press is not offering the yellow light signal to proceed so readily as before maybe. But I'm not confident in a conclusion yet, but range or distance from car is not my case.
old |v8.1(17.28 c528869)| AP2 |
new |v8.1(17.34 2448cfc)| AP2 |
 
I think Tesla has made the correct decision not to release the level of detail you are requesting. That is because end users (all of us here) do not do quantitative testing of Auto Pilot. We all just report our subjective seat-of-the-pants impressions which can be widely different responses to the exact same software.

Tesla does quantitative testing to gather hard data about Auto Pilot performance before releasing updates. These updates are then subjectively evaluated by end users in a huge number of unique situations and Tesla cannot possibly test in advance for all of those situations. End users will inevitably experience system shortcomings and failures. That is why end users have to stay alert to surrounding conditions and keep their hands on the wheel, as Tesla strongly recommends.

All that said, it seems clear that currently Tesla has the best driver assist features on the market. To those who say that Auto Pilot is "dangerous" or "unsafe", my response is "Then don't use it and wait for it to improve because the last two years have shown that it does improve over time".

And then don't sell it and pretend it is ready for sale and refuse to give the money back.

I had a couple of random reboots on this firmware. Strange.
 
Just had an event today with AP2 on I-5 in which my vehicle drifted too far to the left in a sweeping right freeway turn and nearly collided with a truck in the next lane to the left. I took control of the steering. The truck was driving a bit far right in that lane, but it appeared my vehicle was clueless about the truck.
 
for me, with model X AP2
17.17 -> very bad (ping-pong in the lane and false positive brake)
17.24 -> best a this time less ping-pong and less false positive
17.32 and 17.34 changed the behaviour, but not really improve reliability of Autopilot.
There is always ping-pong, less in 34 than in 32, but false positive brake are increesed.

in .34, the car some times, slow down for a while (more than 10s) when there is an outgoing lane, but without any reasons.

In sumary, since .24 there is no improovement, only added bad behaviour in some special cases.
 
I think I finally figured out what causes the AP ping ponging.

The AP gets confused when there is a concrete paving joint within the lane. It often ping ponging trying to adjust itself. At times, the screen will read the concrete paving joint as the lane and try to center itself.

I tried to snap a photo today, but the AP was too dangerous to use.