-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Is EAP worth it?
- Thread starter Fursty Ferret
- Start date
Given the UK is allowing this limited Level 3 FSD (in very limited situations) later this year, I'm expecting to see significant development from Tesla. Given Tesla have told us FSD customers we aren't getting all the features due to our local regulations, when those regs change Tesla will need to adopt.
What's interesting is that this perhaps will be the first time L3 happens in a Tesla, all the FSD in the US is still L2 where drivers need to touch the wheel. We see them getting ready by adding AI to the internal camera to check you aren't asleep or dead. The regulations also seem to recommend that drivers are given things to do while using L3 Autopilot, e.g. watching Netflix, emails etc.
What's interesting is that this perhaps will be the first time L3 happens in a Tesla, all the FSD in the US is still L2 where drivers need to touch the wheel. We see them getting ready by adding AI to the internal camera to check you aren't asleep or dead. The regulations also seem to recommend that drivers are given things to do while using L3 Autopilot, e.g. watching Netflix, emails etc.
I don't really buy the regulation argument. The limits are:
- You can;t use remote summon from a distance unless on private land, but even basic summon isn't reliable when standing next to the car
- You need to confirm a lane change, yet half the time the suggested lane change isn't right, and either way, is that really a hardship?.
- The maximum speed/lateral g in a corner is too low, yet the car aborts mid corner and doesn't predict the turning force which to me is essential for it to set an appropriate speed. Until the car can work out the cornering g ahead of time and moderate the speed, then raising the threshold will just mean cars aborting AP at higher speeds mid corner which would be madness.
The bigger challenge is Tesla aren't training the car for UK conditions, so we get an empty lane next to cones. It is Tesla failing to localise that will hold it back rather than regulations.
And dropping the radar just feels like the launch of Teslas EAP system when it dropped mobileye in 2016. It took a massive step backwards and took arguanly 2 years to even approach parity with the previous system. They started again with 4D last year and started the limited beta trials in the US, and now they're dropping the radar blaming that but the initial launch of the software isn't better than what its replaced so how can they know its an improvement? Seems to me they are searching for a solution and not finding it.
- You can;t use remote summon from a distance unless on private land, but even basic summon isn't reliable when standing next to the car
- You need to confirm a lane change, yet half the time the suggested lane change isn't right, and either way, is that really a hardship?.
- The maximum speed/lateral g in a corner is too low, yet the car aborts mid corner and doesn't predict the turning force which to me is essential for it to set an appropriate speed. Until the car can work out the cornering g ahead of time and moderate the speed, then raising the threshold will just mean cars aborting AP at higher speeds mid corner which would be madness.
The bigger challenge is Tesla aren't training the car for UK conditions, so we get an empty lane next to cones. It is Tesla failing to localise that will hold it back rather than regulations.
And dropping the radar just feels like the launch of Teslas EAP system when it dropped mobileye in 2016. It took a massive step backwards and took arguanly 2 years to even approach parity with the previous system. They started again with 4D last year and started the limited beta trials in the US, and now they're dropping the radar blaming that but the initial launch of the software isn't better than what its replaced so how can they know its an improvement? Seems to me they are searching for a solution and not finding it.
I wouldn't buy FSD again, based on its huge gulf in performance versus where it needs to be. But I would buy EAP.
There is a lot of valid criticism on here, but IMO people expect perfection, and no system can be perfect - human beings are from perfect. I use it on the motorway all the time, and on major A-roads. Never around town or windy B-roads - it is far too jumpy and over-cautious.
You need indicate or confirm the lane change it suggests, and gently nudge the wheel, it then changes for you. I find it works correctly 95% of the time. There is the odd phantom brake (an average of 1 event per 100 miles, say), but foot is already resting on the accelerator and I can override it. It's also very rarely a true 30 mph drop in speed for me, rather it reacts like a nervous teenager on a driving lesson, with a drop of a 5 mph or so. Enough to scare you and other drivers maybe, but not enough to be a danger.
I've had TACC-like systems from Audi, Merc and Ford in the last 5 years, and I'd say that EAP tries to do more than any of those systems, and mostly gets it right. The only thing that really annoys me is that it's so nervous. If it's a really busy motorway, it won't bother or accept a lane change into a space I think is very reasonable. This can make you a bit 'middle-lanery', and I hate that!
There is a lot of valid criticism on here, but IMO people expect perfection, and no system can be perfect - human beings are from perfect. I use it on the motorway all the time, and on major A-roads. Never around town or windy B-roads - it is far too jumpy and over-cautious.
You need indicate or confirm the lane change it suggests, and gently nudge the wheel, it then changes for you. I find it works correctly 95% of the time. There is the odd phantom brake (an average of 1 event per 100 miles, say), but foot is already resting on the accelerator and I can override it. It's also very rarely a true 30 mph drop in speed for me, rather it reacts like a nervous teenager on a driving lesson, with a drop of a 5 mph or so. Enough to scare you and other drivers maybe, but not enough to be a danger.
I've had TACC-like systems from Audi, Merc and Ford in the last 5 years, and I'd say that EAP tries to do more than any of those systems, and mostly gets it right. The only thing that really annoys me is that it's so nervous. If it's a really busy motorway, it won't bother or accept a lane change into a space I think is very reasonable. This can make you a bit 'middle-lanery', and I hate that!
I've had a couple of scary brake-slamming moments when the car absolutely shits it's pants at...a car on the hard shoulder, or a slow moving lorry...or a car with a trailer...or pedestrians in sight of a fast moving road... And when I say scary, I mean full force brakes, alarms wailing and the baby and dog in the back getting very upset, nevermind what the car behind must be thinking. This is just on standard Autopilot (lane keep), by the way, so I have no intention of handing over more money.
I love the car otherwise, mind - features like this are just a nice to have as far as I am concerned.
I love the car otherwise, mind - features like this are just a nice to have as far as I am concerned.
some good points.
- limited g in corners should be easy to predict. It knows the corner as its coming up and should be able to estimate g vs speed. Either slow down, or if keeping up with traffic it should alert and abort before you hit the corner. Cancelling mid-corner is crazy. It'll cancel in some locations on the motorway in the UK (M23 for example for me) when any motorway should be safe to take low radius corners at higher speeds
can you just ignore the lane change suggestions?
Regulation is the favourite scapegoat for a lot of the failings of AP.. As you say, predicting corners and going around them at a safe speed shouldn't be that hard. If NoA lane changes actually made sense half the time I'd be more worried about having to confirm them.. as it is I'm glad we do, otherwise teslas would be all over the road.
MrBadger
Badger out
Cornering speed seems solved in FSD city streets beta. Its one of the changes in behaviour that I hope filters down to everything. There may also already be some improvement from the newer NN's in .4.15.x and .4.18.4 as actions seem much smoother in that although I have had a lane flare/split disengagement so not all solved. My concern is that Tesla have started making noises that highways is 'done', which it certainly is far from (possibly a reference to it being HW2.x compliant), but hopefully that will not stop benefits from FSD and HW3 enhancements from filtering through.
I've posted a video a couple of times from the FSD beta which does the same trip back to back showing how the car behaves completely differently.
Last edited:
I enjoy driving. The Autopilot works well, but is of limited utility: I mostly use it in 20 and 30 limits in areas where speed limits are enforced.
On A and B roads that don't have well defined white lines it is hardly worth the bother and on dual carriageways and motorways I find monitoring the system more demanding than actually driving.
So I have no intention of upgrading to anything more than the basic autopilot that the car was delivered with.
On A and B roads that don't have well defined white lines it is hardly worth the bother and on dual carriageways and motorways I find monitoring the system more demanding than actually driving.
So I have no intention of upgrading to anything more than the basic autopilot that the car was delivered with.
In my opinion there’s absolutely no point buying FSD in the UK right now. EAP gives you 99% of the functionality, literally the only thing you’d be missing is “traffic light stop and go”.
EAP might be worth it to you, depending on how often you drive and what sorts of roads you travel on. I used NoA on a trip to Wales and it worked really well, even the auto lane change bit. I’ve experienced no phantom braking either. That said, I don’t do an awful lot of mileage and I can’t discount people’s personal experiences.
It is reasonable to think that on a long enough timeline there would be some divergence between EAP and FSD. I find it interesting that - as far as I know - EAP is a Europe only thing, which suggests to me that Tesla know the £6800 they ask for FSD, whilst hamstrung by UNECE regulations, isn’t tenable. Introducing EAP at half the price with 99% of the current functionality is a shrewd way of not discounting FSD whilst acknowledging that it’s overpriced.
As a FSD owner I’d like to think we’d see a FSD Beta over here before the end of the year. Realistically though the wheels of regulation turn very slowly and Tesla’s seem to as well, which doesn’t fill me with a great deal of hope of any meaningful improvements any time soon.
EAP might be worth it to you, depending on how often you drive and what sorts of roads you travel on. I used NoA on a trip to Wales and it worked really well, even the auto lane change bit. I’ve experienced no phantom braking either. That said, I don’t do an awful lot of mileage and I can’t discount people’s personal experiences.
It is reasonable to think that on a long enough timeline there would be some divergence between EAP and FSD. I find it interesting that - as far as I know - EAP is a Europe only thing, which suggests to me that Tesla know the £6800 they ask for FSD, whilst hamstrung by UNECE regulations, isn’t tenable. Introducing EAP at half the price with 99% of the current functionality is a shrewd way of not discounting FSD whilst acknowledging that it’s overpriced.
As a FSD owner I’d like to think we’d see a FSD Beta over here before the end of the year. Realistically though the wheels of regulation turn very slowly and Tesla’s seem to as well, which doesn’t fill me with a great deal of hope of any meaningful improvements any time soon.
Had another drive in a Model 3 today. Autopilot is incredibly basic compared to the Volvo equivalent (hard braking for no reason, doesn't accelerate after a lane change until you've been in the new lane for a few seconds, surprisingly violent steering snatches when driving over arrows, and doesn't slow down for bends) so I think I'll need to stick with the EAP.
It's surprising really given that Pilot Assist in the Volvo range is now 6 years old and considered one of the weaker driver assist systems out there.
It's surprising really given that Pilot Assist in the Volvo range is now 6 years old and considered one of the weaker driver assist systems out there.
As far as I know EAP does not add any additional smooth sophistication to the basic function when compared to plain Autopilot. They are as good or bad as each other. It just adds lane changing, summon and parking into the mix.
ElectricChapel
Member
EAP does fix the annoyance of not being able to change lanes without Autopilot disengaging. Is it worth the price? You decide.
I really liked the Pilot Assist in my XC40, the fact that the car starts accelerating when you indicate to overtake was one of those very small details that made me think Volvo really nailed the user experience.
I really liked the Pilot Assist in my XC40, the fact that the car starts accelerating when you indicate to overtake was one of those very small details that made me think Volvo really nailed the user experience.
I had a three month trial of FSD and I’d never ever pay for it. Summon only works if you are right next to the car and even then mostly fails to work. Also it’s a gimmick and has no real world use.
auto lane change worked 1 in 5 times and that’s all you really get over the basic cruise control. I wouldn’t pay a penny for it.
auto lane change worked 1 in 5 times and that’s all you really get over the basic cruise control. I wouldn’t pay a penny for it.
Similar threads
