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

17.5.28

This site may earn commission on affiliate links.
I haven't had the opportunity to try the latest auto steer yet but I don't see any difference in how the side collision warning functions than the previous version. It basically remains off and on and hasn't yet provided me with an accurate portrayal of what is actually around me. It still needs a lot of work, in my opinion.

I didn't see any difference with the side collision warning either. Can someone who is familiar with AP1 describe what exact behavior we should expect from side collision warning? Is the image supposed to turn RED? Under what conditions?
 
I didn't see any difference with the side collision warning either. Can someone who is familiar with AP1 describe what exact behavior we should expect from side collision warning? Is the image supposed to turn RED? Under what conditions?
On AP1, if you get too close to something beside you, you will hear a metal crunching / scraping sound, followed by increased insurance premiums...

More seriously, on AP1, in addition to the whiskers getting red silently, if an object gets even closer to you, an alarm will sound (the same tone as the collision warning alarm) and your car's image turns red. Furthermore, really severe circumstances may cause your car to jerk the wheel a little to the side to avoid the threat.

But the release notes make it sound like only the audible warning aspect is present. I've never personally had side collision assist/warning kick in, and I've had some cars uncomfortably close to me. Even at orange, the distance between you and the car next to you is pretty frightening.
 
I received my notice to upgrade firmware this morning. Took around 40 minutes. It is 17.5.28. Here is my experience and I hope it is helpful. I will do my best to stick to facts and not "editorialize"..but ...you know ...some opinions or guesses will slip in:rolleyes:!

1. I headed to a freeway that was not too busy where I could drive the "new" 50 mph in the "slow" lane and other drivers would pass me without any problems or in the case of some others... just go the same speed as I -particularly those not comfortable with the weather conditions we encountered.
2. Weather conditions (important I believe): VERY HEAVY rainfall (what a surprise for the Seattle/Puget Sound region!;))
3. Total miles round trip using the 50 mph Autosteer: 38
4. The TACC works flawlessly in all situations I came upon. For my report I have no negative encounters to share.
5. The Autosteer was pretty stable, however there was some "dancing" within the lane from time to time. (see: weather conditions above - I don't know how much that affected the lane movements.) On this 38 miles trip I was able to maintain the 50 mph speed most of the time (I would say 95% of the trip or thereabouts).
6. The most common recurring "problem" was a propensity for the Autosteer to turn on to off ramps (I did not have my signal on-if you were wondering). I had to take control and pull back onto the freeway as I passed every off ramp (with the exception of a few "tests" I describe below). I felt that it was confused by the lane lines. (guessing). On one instance I decided to let it go (there was no one around me) - and autosteer seemed confused as it headed on to the off ramp....then it moved the car back onto the freeway after a few seconds (made me a bit nervous still). So it appeared to correct its mistake. (YES I always kept my hands on the steering wheel) On my 2nd and final off ramp test, I allowed Autosteer to veer onto an off ramp and again it reversed its decision and headed back to the freeway slow lane. On that test I did take control because it was a pretty late decision by Autosteer and if it had been one of those off ramps that had barrels of water to stop out of control drivers.....we may have hit them or come close to hitting them.
7. The other parts of the trip were uneventful and for many of the miles TACC and Autosteer worked well.

Ok. Here is the "editorial" or opinion part I will try and keep separate from what I hope represented facts or factual observation above. I LOVED the few new features 17.5.28 gave us today. I do not expect perfection as we experience this journey (Enhanced Autopilot). Understand that we will be dealt "features" a little bit at a time and it is probably good practice for me as I am NOT a patient person....and can use some practice. I hope we can use Autosteer on other roads soon and/or the 50mph limit is lifted higher. Of course it would also be nice to SOON get lane departure back along with the headlight autodim, rain sensor...and other stuff that was stripped away. I will apologize ahead of time if I incorrectly used any words or phrases (I noticed that has been a hot topic). My intent was to inform or report or just share my perception of today's experience. It was fun to test. I DID NOT try any test of the side collision warning :confused:....that seemed a bit over the top. I hope what I wrote helps.
 
From what I am reading, leaving mobileye was a big mistake.......
There's definitely a huge learning curve for Tesla, but I think I'm seeing about the same huge learning curve I was expecting them to have. Their steady progress seems to indicate they are getting their feet thoroughly wet and should be treading water pretty soon. To me, it seems like a temporary setback, with an expected outcome in 3-5 years of swimming success that is actually better than if they had continued with their ill-fated inter-hemisphere time zone impossible relationship.

Unfortunately, during some of my recent "experience giving" drives for the new shadow mode, I realized I have 10,000 pretty unique highly mature highly trained and highly experienced thorough considerations in all of my driving movements per typical trip (granted, my typical trip and commute is hilly, but that's not unheard of in the world). No AI could learn all that from guessing, unless it learned the full experience of human beings over the course of decades (in aggregate, of course). What's worse, sometimes in less interesting roads, I just couldn't care as much. How is the AI supposed to even know that while it's learning?
 
Last edited:
Here's a breakdown of what I've noticed as compared to 17.3.15.

1. TACC braking is much smoother. It starts braking further out and uses regen to continuously and smoothly brake until the last 5 mph where it applies brakes. This is exactly what I do and I really appreciated that it didn't wait until way closer to stopped cars to decelerate. Big kudos to Tesla.

2. Autosteer working at 50mph is not much different from before. I was able to test it out at O'Hare Airport where it stayed engaged until I disengaged it before the departure terminal dropoffs. It worked fairly well. During sharp turns it worked great. Gradual loping turns made the car zig zag slightly. It was behavior I hadn't seen before. I ran it for a 4 mile stretch including through an I-Pass toll booth where the car did admirably well (but I was very nervous).

3. Side collision alert happened once while at O'Hare and it was a mostly false alarm. Traffic constantly shifts lanes as people adjust to needing different terminals or arrivals vs. departures. A particularly wide and inattentive Enterprise bus got close to my car while autosteer was engaged it threw up a bunch of warnings, including what I suspect was the side collision alert but I was too busy looking around and taking control to look at the IC.

Overall I think Tesla made definite improvements but a lot more work remains to make Autosteer as useful as we expected it.
 
Has anyone actually verified, over a network connection, that the cars are sending "learning" data back to the mothership? It seems like it could be a convenient excuse to delay features. Tell everyone that it's ready, but in shadow mode to learn, while in reality it's not learning anything but you've given your team another 6 months to figure things out.
If it's really learning, you should see relatively large amounts of non-gps data streaming over the wire...
 
Got the update this morning. Drove about 10-12 miles. I noticed zero difference in auto steer (other than the 50mph increase). However I did notice two things on TACC:

1) it feels like the system follows the speed of the car in front a little bit better on sharp curves. Previously it seemed that if the car in front turned and you followed, your car would accelerate during the turn and then it would suddenly break when it came to a straight line with the front car again. Now it feels like it follows the speed of the car in front during the turn as well. At least it is doing it a little bit better.

2) I noticed that the car is using regenerate breaking more in order to slow down and either match the front car's speed or break when the front car is stopped. With the previous version I could feel and hear the breaks applied. Now the car is using regenerate breaking more before it uses the breaks.

At least these are my observations.
 
I got the new update which seems a bit more stable. My biggest gripe is that different than AP1 it only lets you use autosteer on white marked freeways. I would sure like to use it on divided highways with 2 lanes going same direction. Seems this would be much safer than a freeway with a 20 mph differential b/w actual speed and autosteer speed.
 
I don't have an AP2 car (mine is AP1, April 2015 build/delivery), but I've been following this thread to read the opinions of AP2 owners. FYI, I received the 7.0 update in October of 2015 which was the first release of autosteer for AP1 cars. It's characteristics at the time were:
  • Wobbling/slow response on curves, with hunting in the lane.
  • Wanted to take freeway exists when travelling in the right lane.
  • Hunting back and forth in the lane even on straightaways, especially when the car in front wasn't centered.
Sound familiar?

Be patient, AP2 software will eventually come around and exceed the capabilities of AP1.
 
8637722255413272.jpg