Should I add EAP option?

[rakda12]

Hi Everyone,

I just ordered a model S 75D on Sunday. I am debating if I should add the EAP option. I have a long commute on highway and there are lots of stop and go traffic on the highway. For folks live in bay area, I am talking about HWY 680 at Easy Bay. If I add the EAP option, I will only use the TACC feature to relieve some of the stress of hitting gas and brake. There are some uphills and downhills.

I read the forums, it seems that the feedback is that the EAP presses the brake too late and too hard. What's your experiences using TACC in a stop and go traffic on HWY?

Thank you,
Kate

 

[DrReid]

I have AP1 and not AP2, but it's incredible in stop and go traffic. It's my favorite thing about the car.

If you have a traffic heavy commute, it's a no brained to activate EAP.

 

[redan]

For me, stop and go traffic is one of the best uses of EAP. And full AP, not just TACC. I think AP is what makes Tesla special.

 

[boaterva]

EAP is worth it. FSD you can definitely wait until there’s some content. In any stop and go, EAP will save your sanity. You still need to pay attention but you aren’t spending an hour staring at a bumper in front of you.

And no idea about the ‘brake too late’ idea. You can adjust the following distance to what you want. Works fine for me.

 

[DriveMe]

You should really try it on a test-drive and decide how you feel about it.
I tried it and I loved it!

If you do want to add the EAP option though, you need to act quickly, or they will want to charge you a $500 fee to change your configuration. I tried to add the FSD to my configuration before the delivery, and there was no way around that fee. Which is ridiculous, IMO, since FSD is not even a configuration change, just a promissory note.

 

[druken]

I live in East bay (Dublin) and commute to Santa Clara. AP2 autosteer and TACC work great on 680 especially in traffic.

 

[Jabardo]

I have AP2, I loved it at first, and would use it all the time. Recently it has become unreliable, hard braking and acceleration, asking to hold the wheel after 2-3mins. It’s a novelty to have, but I enjoy physically driving the car, I regret purchasing it. I’d wait until it gets much better.

 

[futurem3owner]

AP during rush hour traffic is fantastic and a no-brainer to have it on. It's the highway driving where the decision turn it on or not that has me waver back and forth. When you have instant torque and effortless acceleration at your foot, it makes me want to drive my Model S and zip past everyone. But then, I feel like I'm not getting my "money's worth" if I'm using AP only when there's stop and go traffic. So, I decided to drive the car myself for the morning commute and let the car drive itself for the evening commute when I'm a bit tired and there's more traffic.

 

[viper2ko]

Hi Everyone,

I just ordered a model S 75D on Sunday. I am debating if I should add the EAP option. I have a long commute on highway and there are lots of stop and go traffic on the highway. For folks live in bay area, I am talking about HWY 680 at Easy Bay. If I add the EAP option, I will only use the TACC feature to relieve some of the stress of hitting gas and brake. There are some uphills and downhills.

I read the forums, it seems that the feedback is that the EAP presses the brake too late and too hard. What's your experiences using TACC in a stop and go traffic on HWY?

Thank you,
Kate

Do it! When I planned to get my MS I always intended to never get the EAP because of the extra 5k and I love driving myself. I was able to 0.99% APR from Alliant and that drove me to get it. Im awaiting delivery, but without EAP it just wouldn't feel like a Tesla. Keep in mind without EAP you can't take advantage of Summon and auto park as well I believe

 

[brkaus]

EAP is fantastic for traffic. It does well when following someone. It will "brake late" if you come up to already stopped cars, like at a stop light. I will typically flick the lever down a few times to start reducing speed early.

For what is is worth, I believe the reason for the "brake late" problem is that not every stopped car should cause your car to stop. For example, on a curved road, a car stopped on the shoulder would be straight ahead at some part of your drive as you go around the corner. Yes, it sometimes slows for those too.

 

[idleuser]

Hi Kate,

I made a thread asking if EAP is worth it. I personally think 5000 for a half baked software isn't worth it but to each his own.

You can follow it here. EAP question for current owners

 

[PrGrPa]

What comes as standard and what is now available with Enhanced Autopilot (EAP) seems a confusing picture to me. Not helped by folks adding in extra information about differences between AP1(MobilEye) and AP2 (Tesla EAP), let alone further opinion about the hardware changes along the way.

The Tesla site seems to spend more on what will come in EAP in time and how it makes use of some of the self-driving hardware installed on all the cars. Fair enough. Though I would like a boring yet definitive comparative feature list of the different cars and options - much like an iPhone comparison tool.

If I've understood correctly, the current EAP features are:

• adaptive cruise control including stop-start conditions
• auto park (parallel and orthogonal reverse)
• lane-keeping
• change lanes using the left/right signal indicator
• summon (in a straight line)
• increased dashboard information on surrounding conditions

Though if folks have a more definitive list that'd be a great help.

For me, the adaptive cruise is a must-have. As is auto park. Adaptive cruise because my current car has basic cruise with collision-detect (scary beeping if you zoom towards an object). Great features but adjusting the cruise in moving traffic is something for a machine to do and I'd rather have the car avoid the collision than tell me about it. Auto-park because Parktronic on one of our cars is a real face-saver when you can't parallel park without making a candid-camera-like showcase and I'd trust a car to park me over my own abilities.

Lane-keeping and easy lane changing seem nice additions to adaptive cruise. I liked the way they behaved on a test drive in various conditions. It'll be interesting to see how I feel when I get my Tesla early next year with EAP on it.

 

[AnxietyRanger]

EAP is IMO unreliable in heavy traffic. It keeps disengaging and I don't find it trustworthy given the latest updates - it is getting worse in the reliability department, even though its steering inputs themselves are smoother when it works...

For example, earlier this week I enabled EAP in a traffic jam situation and I think it kept going for a minute or two before disengaging for some reason on a straight road. It just didn't like... something. I find it somewhat more usable in sparse traffic, though ghost brakings and ghost steering events remain.

TACC on the other hand, that certainly is usable. It has ghost braking moments during overpasses quite often, but those one can usually live with.

 

[D.E.]

Hi Everyone,

I just ordered a model S 75D on Sunday. I am debating if I should add the EAP option. I have a long commute on highway and there are lots of stop and go traffic on the highway. For folks live in bay area, I am talking about HWY 680 at Easy Bay. If I add the EAP option, I will only use the TACC feature to relieve some of the stress of hitting gas and brake. There are some uphills and downhills.

I read the forums, it seems that the feedback is that the EAP presses the brake too late and too hard. What's your experiences using TACC in a stop and go traffic on HWY?

Thank you,
Kate

I'm glad I got EAP. The autosteering is good for controlled access highways such as interstates. It is not good for secondary roads. It is not good for town/city driving.
The TACC is good. Occasionally it will brake such as a hill with an overpass overhead. I haven't noticed it since my last software update. And that's the thing, behavior changes with software updates. It is beta, expect changes as updates are sent. If you want a polished finished system, this isn't it. Me, I love it. There are others here who don't. Most seem to like it a lot. People I meet like it. Still, understand that it is a system that is changing over time and there will be some quirks. I use it a lot on long interstate drives. It is great in backed up stop and go traffic. I haven't used autopark. Summon is handy sometimes. It impresses people when you can move the car from outside the car with a cell phone.

If you plan interstate driving, I'd get it. If your driving is in town, you probably won't use it much, not as it currently stands. If you like the idea of being on the cutting edge of driving technology, get it. If you aren't sure, you can add it later but it will cost more. So it depends largely on your expectations and people here won't be able to help you much with that.

As far as the car itself goes, it is just awesome.

If you have specific questions, ask.

Best,
Dabid

 

[Tam]

...EAP presses the brake too late and too hard...

I don't doubt those reports because remember, Autopilot is beta so anything worst thing imaginable can happen.

If you are willing to babysit the system, I have found it very helpful.

Please watch my lengthy 40 minute, 45 mile AP2 freeway trip as none of your concerns were recorded (I am sure the system can misbehave when I don't record them). I only had to intervened twice for 2 freeway interchange exits. That despite of stop-and-go traffic, construction zone with cement barriers next to your car that you can practically touch them, even unexpected 2 sharp curves due to construction rerouting and 2 stretches of missing lanes....

 

[rakda12]

Thank you so much for everyone's reply. I did go in for the test drive and decide to go for the EAP.

You guys are awesome. I am sure I will have a lot more question coming

 

[AnxietyRanger]

Looking at @Tam's very helpful video, I started to think: How different is AP2 outside of California vs. California?

Could the maturity of the NN (compared to the very mature EyeQ3 image processing) be to blame?

OP seems to be in California, so Tam's reference can be helpful for them.

I find Tam's video very stable - my AP2 is all over the place many times over in a similar time frame, but I am in Europe, not in California...

Could this explain some of the differences of opinion?

 

[D.E.]

Hi Everyone,

I just ordered a model S 75D on Sunday. I am debating if I should add the EAP option. I have a long commute on highway and there are lots of stop and go traffic on the highway. For folks live in bay area, I am talking about HWY 680 at Easy Bay. If I add the EAP option, I will only use the TACC feature to relieve some of the stress of hitting gas and brake. There are some uphills and downhills.

I read the forums, it seems that the feedback is that the EAP presses the brake too late and too hard. What's your experiences using TACC in a stop and go traffic on HWY?

Thank you,
Kate

I may not have told you everything you probably need to know.

The car and software are in beta. You get the newer software as it is released. The current car is yours to keep. You should assume there will be a superb and very desirable hardware improvement 2 weeks after delivery of your car, free to those who just waited a bit longer to order. It will be a no cost improvement for the lucky procrastinators. They won't be able to add it to yours. There'll be crying, of course. Just expect it. These incremental improvements are not predictable but you will see posts from those who try. It will in no way diminish the car you bought, but you know what they say about the grass being greener.

The software is in beta. It has always been in beta. It always will be in beta. OK that last bit is a stretch but so far so good. They've been making the S since about 2011 and there's been beta software all the way. It is always improving and every now and again you find a real nugget, your car will do something it wouldn't do before.

Musk is an optimist. We all love him but he's flawed. His timetable doesn't run on the same clocks we use. When he says a feature is coming, it means he thinks it is coming but the kinks haven't been ironed out yet. So it is probably coming but his year may be closer to a decade. If he says it's coming in the next update, it means it is coming but it may be several updates away. He doesn't mean to mislead us, he's just a little optimistic. Well, a lot optimistic. If someone puts in a suggestion and gets word that it will be included, figure it was written on a napkin and it's dinner time.

Our CEO has distractions. He's making batteries, that's good. He's into solar roofs. He wants to go to Mars. I'm an amateur astronomer and I don't have the slightest idea why we want to put people on Mars. There is nothing to breathe there and there isn't much to do. I think it might also be a one way trip. Anyway it distracts him from making my Tesla better. There is a screen Easter egg, the car can pretend it is on Mars. I guess that's something.

You'll read scary things. Such as the autopilot programmers left and went to work for Google/Apple/YouNameIt. I'm sure he hires new ones but we don't hear about that. You'll read posts about some horrible misfortune experienced by a Tesla owner. Tesla makes over 2000 cars a week. You know something awful is going to happen to a few of them. Some vent here. Then there are people who see a complaint here as some sort of planned anti-Tesla plot foisted on the public, in an owner's blog. It usually involves suggestions of manipulation of stock prices. I'd read the horrors but keep in mind 91% of us are happy enough with ours to plan to buy another one. That is higher than any other car brand. I suggest you follow the bigger trends, and if everyone else's mirror is falling off, check yours and see if it is loose.

The other car manufacturers are on the self drive bandwagon as well. I don't think we'll be left behind. I just don't think we'll be as far ahead as we'd like. But still the auto steering on the interstate we have now is really neat.

All that said, the car you get will be absolutely awesome. If you keep expectations in line, you are likely to be a very happy camper. It doesn't matter if you get enhanced autopilot now, you can always add it later if you want. It costs more but it isn't as if you'll need to trade cars. If you do get it now, it's another neat thing to play with.

Best,
David

 

[Tam]

...I am in Europe, not in California...

Now, that you raised the issue of different countries, that makes me thinking!

As far as I understand, Tesla is using nVidia method of artificial intelligence which does not need a human programmer to code every scenario.

That means, once your car learns how to avoid accidents in California, it should transfer that knowledge when it drives in Europe as well.

However, Tesla does not allow your car to learn and correct its behavior locally on its own without a validation team.

All data and theoretical reaction (shadowing) must be validated by the team first before that theoretical reaction will be enforced fleet wide.

Let's do a scenario without Tesla's validating team protocol: your car would try to run over a traffic cone over and over again but eventually it learns locally that when it sees this kind of shape, it should avoid it (without Tesla validating team). It would then transmit that knowledge to the fleet so globally, all cars will avoid running over traffic cones, and all this in real time.

However, Tesla put that learned behavior in shadow as a theoretical reaction. It would then validate that traffic cone avoidance as desirable and make a decision to enable that behavior to the fleet worldwide.

So, I think there's a bottleneck in Tesla validation process because it involves a human team. It may have a different validating team for Europe and learning are not immediately transmitted worldwide.

What I mean to say is, hang on! I am sorry about inferior Autopilot in Europe right now, but eventually, it will catch up!

 

[AnxietyRanger]

Now, that you raised the issue of different countries, that makes me thinking!

As far as I understand, Tesla is using nVidia method of artificial intelligence which does not need a human programmer to code every scenario.

That means, once your car learns how to avoid accidents in California, it should transfer that knowledge when it drives in Europe as well.

However, Tesla does not allow your car to learn and correct its behavior locally on its own without a validation team.

All data and theoretical reaction (shadowing) must be validated by the team first before that theoretical reaction will be enforced fleet wide.

Let's do a scenario without Tesla's validating team protocol: your car would try to run over a traffic cone over and over again but eventually it learns locally that when it sees this kind of shape, it should avoid it (without Tesla validating team). It would then transmit that knowledge to the fleet so globally, all cars will avoid running over a traffic cones, and all this in real time.

However, Tesla put that learned behavior in shadow as a theoretical reaction. It would then validate that traffic cone avoidance as desirable and make a decision to enable that behavior to the fleet worldwide.

So, I think there's a bottleneck in Tesla validation process because it involves a human team. It may have a different validating team for Europe and learning are not immediately transmitted worldwide.

What I mean to say is, hang on! I am sorry about inferior Autopilot in Europe right now, but eventually, it will catch up!

There has also been talks of using maps on the Autonomous Vehicles forum... If Tesla has mapped areas near it well, that could make a big difference... Seeing that there are U.S. posters who find AP2 unreliable like I do, maps could be the difference...

Or it could be just the neural networks not being taught enough subtle differences in different areas and is thus faltering elsewhere...

@verygreen @lunitiks @wk057 theories?