Wow! It’s been a long time since the last post. I’ve been living with my 2018 Model S with AWD, AP, FSD for quite a while and wanted to share some thoughts as I wait for the FSD beta program.
Sorry in advance for the length but thanks for hanging in there with me.
My 2018 Model S has over 51K miles on it now. I’d say easily over 95% with Autopilot engaged. I pretty much engage it on any road it lets me every single time. The exceptions are when I have passengers in the car as non-highway driving can be scary with AP (see below), and roads without a middle line.
I’ve had a few smallish issues:
- The driver window wouldn’t go up and stay up (they adjusted the sensitivity)
- The cabin was starting to smell musky. They changed the cabin air filter and flushed the system with cleaner. That cost about $240.
- Right now the A/C unit is making a “groaning” noise and is cutting in and out. Noticeable sound inside and outside the car. Thats going into the shop soon. They’re estimating a free repair.
- I did get the free FSD computer upgrade back at the end of 2019 when it became available.
Other than that, just the regular tire rotations/swaps and washer fluid.
The car has been reliable. I haven’t put nearly the miles on it I’d have expected thanks to COVID but at 50K miles, I think I can say I have plenty of experience with it.
As I mentioned a while ago, I got the car for All Wheel Drive (AWD) and Autopilot (AP). I also pre-purchased Full Self Driving (FSD) as it was still relatively inexpensive ($3,000. It’s currently a $10,000 feature) and I wanted to experience it as it came out and evolved (more on that later).
AWD and AP have been pretty much what I expected and wanted and the reason for the upgrade. I have a few gripes on these:
- Auto windshield wipers are machine learning-based and generally not as good as o her automatic wipers on less expensive cars.
- Traffic Aware Cruise Control (TACC) (what others call Adaptive cruise control) disables itself when weather conditions don’t permit it to do its thing. The problem is you get no cruise control at all when that happens.
Overall I’ve used AWD (it’s on all the time, but I mean I’ve driven a lot in snow and ice too) and AP a lot.
The fit and finish of the car have held up and I still love the looks of the car. One small downside is I’m driving more elderly people around these days and entry/exit from the S with its low ride is a bit harder for them. I’m going to play with the ground clearance to see if I can help that.
Full Self Driving
I didn’t buy the 2018 S for FSD but as an engineer, I’m interested in it and seeing what can be done. People get confused between autopilot and full self-driving, so here’s how Tesla defines those right now (it’s been a bit of a moving target over the last 3 years):
Autopilot is very similar to the adaptive cruise and lane-keep assist features from other vendors. There are positive and negative differences. It is much better (but not perfect) at staying in the lane, even around tight turns.
When you look at the list of FSD features, my opinion on them is they’re either of limited use to me or not functional enough to use on a regular basis except for experimentation.
Just to level set, I’m on a 2018 Model S with all software features, the latest production software, no betas (yet).
Here’s my blow-by-blow on the FSD features:
- Navigate on Autopilot (NOA) – if you’re doing pure large highway to large highway changes this mostly works fine. Sometimes it gets to the lanes too late, sometimes its in the wrong lane (i.e 84W to 91S from Boston). If the ramp is tight, then its ramp behavior is bad. Its all over the place on the single lane, doesnt pick the right side for a split etc. Some people may be ok with the behavior as it does get you there, but I believe the behavior could lead to accidents as they’re not what a person would do.
- Auto Lane Change – I really want this to work and have tried different settings. It’s either way too aggressive in lane changes or way to passive in lane changes. While its almost always safe, it lacks the situational awareness of a good time to pick to change lanes with people coming up on your left. Yes, you want to get over there eventually but pick a more friendly time. So I use auto lane change 100% of the time, but by pressing the turn signal myself vs letting it pick when to move over.
- Autopark – I avoid street parking and cities and i’m vary particular about where my wheels are next to curbs, how far I can get from other cars etc. I’ll go in tight on a cub on one side to put more distance between myself and another car to avoid door dings. No curb rash, no door dings, 50K miles. Perhaps this works, dunno, never tried it, so no value to me.
- Summon – For moving your car forward or back a bit, this sort of works and can be helpful. Yesterday I washed my car. When I was done, rather than get in the car with wet shoes, I used summon to back it in on its own. It started fine and ended up at a crazy diagonal. That’s a good case. In the work parking lot, there’s a nice spot that fills with water when it rains so nobody parks in it to avoid stepping in a puddle. I pull up, get out and then use summon to put the car into the puddle and then back it out when I return. Reserved spot for the win. Those are the kind of cases it’s useful for.
- Smart Summon – Ok, I’m going to get bold here and call this a dangerous feature. Go to an empty parking lot and call the car to where you are. Most of the time the summon will fail and it will stop somewhere mid-process. Sometimes, if you’re patient enough and watch it do some really dumb things, it may get near you at an odd angle, etc. When poeple are nearby its not safe. Nobody maneuvers like it does, it confuses people, it’s vision is limited. It’s not labeled as beta probably as its pre-beta. I wouldn’t even call it an alpha at this point. I’d recommend against using this except in very controlled conditions.
- Traffic and stop sign control (beta) – I currently have this on but it’s annoying. With nobody in front of you and approaching a green light the car will stop. With nobody in front of you and approaching a red light the car will stop and then will not start back up when it turns green. If someone is in front of you for both events it does the right thing. The red light and stop sign behavior are ok, but stopping for green lights is asking for an accident. I drive on a 50mph road that has lights on it and when the car comes over the hill and aggressively decelerates for a green light i’m lucky I dont get rear ended (yes, I avoid this case). Also, blinking yellow lights, the car has no idea what that is. Yellow tells it to slow down so it slows aggressively then speeds up, then slows, and you feel it. Stop signs on streets that intersect at less than 90 degrees it thinks are in your lane and it aggressively stops for them. Brake lights sometimes are considered stop lights when they’re up an incline, etc. It’s marginal. I’d call this more alpha than beta.
I got FSD as I wanted to play with it. It has evolved since I got it, but at this point, and while I don’t mean to minimize the work put in by the engineering team, as a customer I’d call it a $3,000 toy of a software feature set that does more harm than good. If I’d have paid $10,000 and expecting to actually use it for the claims I’d probably be angry. The whole thing is a beta that shouldn’t be getting sold. My expectations going in were low and I’m watching it evolve so I’m getting what I wanted and $3,000 was my price of entry.
Tesla finally gave us “the button” to request access to FSD beta which allows FSD on city streets. This seems to have a lot of potential as it stops at lights, then goes on its own. It stops and then makes turns. It handles odd streets, obstacles, etc.
I pressed “the button” and it’s scoring my driving safety. I’ve driven here in MA for over 20 years without any tickets or accidents, 7 of those years in Teslas. I’m currently at a 99% safety score so we’ll see how that works for getting into the beta.
As a long-time owner, an early buyer of FSD, and an engineer with a passion for quality and testing I was surprised I got no response to the asks to get into the FSD beta program before this. The people I saw get in were social media influencers. You may say that many others got in and we only see the social media influencers. That may partially be right but the percentages would be off if true as very few have been in the program and the percentage of people with larger audiences on social media doesn’t match. Yes, those influencers are passionate about Tesla, but the program didn’t seem to be about finding the best testers to shake out the system. I believe FSD beta was more of a marketing ploy than it was a program to get real quality feedback on new software. While we got a lot of videos about the system, nearly all of them were overwhelmingly positive. Given my experience with the production FSD, I have a lot of questions about what this beta will be like. I think we’re shortly going to see what people not building followers think about the FSD beta.
Is FSD beta another marketing stunt?
Tesla is moving into the insurance selling business. They need data, they’ve tied their way of doing insurance to the new safety score they just rolled out that is also tied to FSD beta. Pretty soon the whole world will be hearing about this safety score and Tesla insurance. But how many people will actually get into the FSD beta program? Getting into the program is still vague. Here are Tesla’s words:
Vehicle driving data and my safety score. Is this a random judgment call? Will they check my followers on Twitter first to make sure it will get a good buzz? Forgive me if I’m jaded but the FSD beta to this point was marketing and I’ve just (voluntarily) signed up to give them all my data to help make their insurance program better.
I’m a $TSLA investor (and quite happy with my returns since 2013). I also owned $SCTY until it got folded into Tesla. I’ve bought 2 Model S’s, have Tesla panels on my house, and blog and tweet about Tesla. I love the car and a lot about the company and Elon, but this “not marketing” marketing stuff has to go. They’re very backlogged on demand anyway.
I hope FSD beta is all that it’s been hyped up to be, but I suspect it won’t be. When/if I get into the FSD beta program I will be writing about it. Meanwhile, I have good and reasonable insurance on my S, so that, like the Powerwall, is one Tesla product I will not be getting.
I also have a 2018 Model S with FSD. I agree with most of your comments. Hoping I can get into the FSD beta. I find AutoPark to be very hit-or-miss. My biggest complaint is that is rarely identifies parking spaces correctly. When it does, it seems to work well, both in perpendicular parking and parallel parking. But I would say 95% of the time, it never identifies a spot correctly and I just drive by. Navigate on Autopilot seems to work much better for me than you. Probably a function of differences in our highways / streets. I’m expecting once FSD is released, NOA will pick up improvements from it. (I also only use autolane change by using the turn signal for the same reason as you – the car picks dumb times to change lanes.)
From what I’ve been reading the NOA/FSD on the highway part is unchanged. At least until they “get to a single stack”. Not sure if that will be in the beta or not when they roll it out.
I do use autopilot all the time, on highways and off highways. I just don’t use the FSD components – auto (unprompted) lane changes, auto-exiting highways, etc.