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

Autopilot impressions

This site may earn commission on affiliate links.
I am 4 days into my trial period. My overall impressions are positive. I have a little bit more than 400 miles on the car total.

I really like the summon feature. It's not necessary but it's helpful. I needed to get a bike off of the wall hanger in front of the Tesla this morning. Instead of getting in, I used Summon to back the car out and got the bike down.

While Autopilot works very well, there are some situations where it has been confused.

When I come across an acceleration/merging lane for an onramp and there is no right side line anymore, it wanders over into the merging lane.

I feel like it brakes too much whereas coasting could work to maintain the set distance. I'm interested in seeing if the effect is to "brake check" people more than what a smooth driver would do in traffic. I need to watch the screen to see if the brake lights come on.

I tried a lane change and there was no one over there and it somewhat abruptly returned to the original lane. Shadows or something? Not sure.

Overall, I am impressed and it has been quite useful in stop and go traffic.

As always, the question is the cost/benefit. While I appreciate the development costs and how it will continue to get better (although I think it's pretty darn good), at this moment, I cannot justify the $5500.

I think it's probably worth $2500 to me as is. There's no science behind this, it's just the cost I'd be willing to pay today for the feature. Which I won't obviously get at that price!

Just my 2 cents.
 
It's not really clear from the Model 3 web page, but one way to look at it is that if actual collision avoidance (as seen in some YT videos, if they're real) is only included in the paid options then it's worth far more than $5500.
 
I've got about the same mileage on mine. While it's pretty nice on the highway, I would agree with your observations. I've noticed in a construction area, it will (justifiably so) disengage for safety. When driving on the interstates in Wisconsin...that would cover many many miles of not being able to really use it, or it getting so annoying for turning off so much that you just switch to the adaptive cruise control.

My commute to work is a real bad road for it as well. Intersections (obviously) lose lane lines so it disengages at almost all of them. I think for the price of it, Tesla needs to address this with other options or train it's camera intelligence to read intersections like humans do. I think the braking is a bit too conservative. When drivers leave my lane (to move to a left turn lane for example) the system will also be a bit too aggressive in accelerating to the next car...and if the light is red..it results in a pretty hard braking since you are essentially speeding up to a car slowing down. Maybe it has to do with my follow distance being too close? 2 cars.
 
  • Disagree
Reactions: smatthew
FYI, if I don't use EAP, I literally forget for a split second that I need to steer. You'd have to tear it from my bare hands at this point, and I'm fully aware of all the nuances. I actually like to see what it does in crazy situations like construction. It fails my "tests" most of the time, but I enjoy being a Test Engineer for Tesla. My favorite test is taking a round-about in Sedona where the inside curb is gradually sloped... that's real fun... it tries so hard, then I literally LOL. I also love watching the improvements from my diligent Engineering work. You're welcome ;)
 
For lane changing you need to keep signaling the change until you are at least half way over the lane line. Otherwise it aborts and tries to move back to the original lane rather abruptly.

You shouldn’t try to kill the signal yourself - this will make it move back to the original lane if it hasn’t moved all the way over. EAP will kill the signal itself once it is done with the lane change.
 
I'm also on a trial and sure others with more experience may disagree, but I did not find EAP very trustworthy on most roads within a few miles of SF. For example, GGB, 101 to San Rafael, anywhere in the city. I found the steering to be to "reactive" with not enough foresight on peripheral conditions. Seemed frequently on the outside of curves that I would have tried to be on the inside. Probably the sensors don't provide enough clarity at that margin and the algorithm/processing is not yet sufficiently complex. For example, it didn't preemptively steer clear of trucks wandering across lane lines (we have 9-10' wide lanes "highway" lanes). There were times that a trailer seemed to get very close to the side of the car. I guess if the data shows that this is not happening, ok. But it certainly made me uncomfortable. If i were driving, I would have cheated the lane to the other side and accelerated past the truck.

It was pretty good on I-80 once you got out to the suburbs and beyond. If I drove out there more than 1-2x per month, I'd consider it. But with infrequent hwy use, I prefer to drive myself. If I had to commute in/out of the city all the time, I'd definitely get it (and try to change my commute).

I wish that the adaptive cruise control and collision avoidance where not tied to EAP steering control. I liked the cruise better that the steering for sure. I imagine some point in the future where they will sell this as a service to capture some revenue from folks like myself that don't use it often enough to justify the purchase. my 2c
 
I have had great experiences with it so far. I mostly only use it on the freeway and stay in the middle lane but it has been driving me around flawlessly so far. The only time it gets in my head a bit is when it accelerates into turns. It hasn't made any mistakes it is just a weird feeling because I am used to human driving slowing down into corners so I have to fight my instinct to brake when it happens.
 
When I come across an acceleration/merging lane for an onramp and there is no right side line anymore, it wanders over into the merging lane.

That's one of the first time I learn doing autopilot. It thinks that the 2 lane merging was a very wide lane and will start moving to the center. It is not really that dangerous just annoying. I don't use autopilot on the right lane.

But it is kind of dangerous there is a off ramp on the LEFT lane. You are driving on the left lane and now there is a left and straight, autopilot may try to sway to the middle right where the concrete corner is... gotta watch out for that.
 
  • Like
Reactions: Abstrakt1
I'm also on a trial and sure others with more experience may disagree, but I did not find EAP very trustworthy on most roads within a few miles of SF. For example, GGB, 101 to San Rafael, anywhere in the city. I found the steering to be to "reactive" with not enough foresight on peripheral conditions. Seemed frequently on the outside of curves that I would have tried to be on the inside. Probably the sensors don't provide enough clarity at that margin and the algorithm/processing is not yet sufficiently complex. For example, it didn't preemptively steer clear of trucks wandering across lane lines (we have 9-10' wide lanes "highway" lanes). There were times that a trailer seemed to get very close to the side of the car. I guess if the data shows that this is not happening, ok. But it certainly made me uncomfortable. If i were driving, I would have cheated the lane to the other side and accelerated past the truck.

It was pretty good on I-80 once you got out to the suburbs and beyond. If I drove out there more than 1-2x per month, I'd consider it. But with infrequent hwy use, I prefer to drive myself. If I had to commute in/out of the city all the time, I'd definitely get it (and try to change my commute).

I wish that the adaptive cruise control and collision avoidance where not tied to EAP steering control. I liked the cruise better that the steering for sure. I imagine some point in the future where they will sell this as a service to capture some revenue from folks like myself that don't use it often enough to justify the purchase. my 2c

Totally agree with you regarding performance north of SF on 101. It can generally do a decent job not killing me going over the headlands but it's not the most confidence-inspiring driving until it gets down to Mill Valley. Then there are a few turns later that give it a little trouble, including the curve right after the Central San Rafael exit. You can always tell which Teslas are on AP on that curve - they all jerk (a lot of human drivers do too because it's a curve that looks really gentle then suddenly sharpens, kind of like 80s racing video games).

But on 280N heading out of SF to the South Bay (my wife's commute), it's heaven.
 
I was all set to not get it, but a coworker let me drive his Model 3 with EAP for about 30 minutes on the road I normally commute on, and I came away fairly impressed. I ended up caving and asked my IDA to add it to my final MVPA. It's definitely worth $3k. It remains to be seen whether it's worth $5k. What I do know is that if I buy it with the car, I pay 6.25% state sales tax. If I get it after, for whatever reason, I would have to pay 8.25% state + local sales tax, not to mention the additional premium for getting it after delivery.
 
For limited access road Auto Lane Change you put the stalk all the way down for a turn, not half way to signal a lane change. It works well. We have rough, poorly marked roads in my medium sized Florida town and it does well on them and will eventually do much better. TACC I believe, gives you as much braking authority as Autosteer. EAP is half the reason I bought the car and I'm very satisfied. EAP gives you a lot of other collision protection that you won't normally experience until someone gets very close to you. It steers in the middle of the lane much more closely than you do. It currently is slightly wide in turns but someone pointed out that lets it see farther around the curve. I used APs most of my career in the airlines and Autopilot is the appropriate name for it. It is pretty close to Auto-land where we had no input except to go-around. We don't yet know what redundancy it has. Some say it has two electric steering controllers which would be required for level 5. Buy the EAP at delivery.
 
I've just started playing with it myself (having just got the car 10 days ago, but the first week wasn't driving until I could get it wrapped). Once I've done auto-steering on the interstate. It kept in the lane perfectly, except as you noted, when an exit on the right appeared and I was in the right lane. It veered into the "middle" before coming back over. I didn't have to correct it, but I thought I might.

This morning on backroads, I use TACC about half of the way to work, and used it about 80% of the way back home. It worked very well, though at times when it's following someone into a turn, it will accelerate too much until I get situated with the car right in front. Then it slows down. I did feel that sometimes it waits a little too long to start to slow down, and once at a stop, it takes a lot longer to start moving than I would like. So I wish it was more aggressive at starting, and less aggressive at slowing down.

In all, I like it, and when I was in stop-and-go traffic, I could easily search for some music to listen to instead of having to speed-up/slow-down frequently.