Forums

Will Tesla abandon Autopilot when FSD is released?

Will Tesla abandon Autopilot when FSD is released?

When FSD is released, it will be able to do everything Autopilot can do but more and better. Autopilot will basically be obsolete. Now, I don't think Tesla will turn off Autopilot but will Tesla have any reason to continue developing and supporting Autopilot after they release FSD? I don't see it. I feel like when FSD is released, AP2 will basically become like AP1 now. It might get a few updates here and there but it will basically be abandoned in favor of the new thing, FSD. I think current cars will still be able to use AP of course but Tesla may even drop Autopilot completely from future custom orders. I don't see why Tesla would even continue to offer Autopilot once FSD becomes available. Even if FSD is only L4 to begin with, it will be far better than AP.

Thoughts?

EVRider | May 27, 2018

Even when FSD becomes available, plenty of drivers won’t want to or be willing to use it for various reasons, so it would be a huge mistake for Tesla to stop making AP available without FSD. I love AP (I have AP1), but I have no interest in using FSD at this time. Eventually, fully autonomous cars won’t have the ability to be driven manually, but that’s a long way off.

murphyS90D | May 27, 2018

I expect that a car in FSD mode is not going to exceed the posted speed limit. I predict that all of the people with a lead foot will not use FSD.

TeslaTap.com | May 27, 2018

Considering there are two price points, I expect Tesla to continue with EAP and FSD as separate options. EAP will also continue to improve once FSD is available, although it might not get new features beyond those already stated in EAP. Then again, only Tesla knows for sure what the future marketing of these two features will be.

hoffmannjames | May 27, 2018

@EVRider But why would you need AP when FSD would fill the exact same role? I mean FSD would be true hands free autopilot. Why would you keep the "bad" autopilot where you have to hold the wheel and pay attention when you could get a better autopilot that is hands free? And, FSD could be turned on or off like autopilot now, depending on when the driver wants to use the feature so it would not be a problem for drivers who don't want to use it all the time. Drivers could still disengage it when they want to drive manually. I am just not sure why you would give drivers an option to only use a L2 system that is not self-driving when a better L4 system that is self-driving is available?

I guess Tesla could make autopilot a standard feature so that everybody gets it and just make FSD the extra option for people who want the car to be able to do more? That might be the best way to go.

hoffmannjames | May 27, 2018

@TeslaTap Yes I do expect Autopilot to improve up to the promised features of EAP. I am sure Tesla will be able to port over the FSD code to fix the current limitations of Autopilot where it would handle steering and speed much better and also not hit dividers anymore. :) But I am still not sure, when FSD could do the exact same thing as EAP totally hands free, why you would still give drivers the option to use a L2-3 system that still requires them to pay attention and hold the wheel?

Tropopause | May 27, 2018

I think OP has a good point. All or nothing would be the safest way to operate an FSD vehicle, ie- either full manual driving or full self-driving. This would eliminate the few semi-autonomous issues we are seeing today with people not appropriately maintaining control while EAP is engaged. Joint operational control (responsibility) of a moving vehicle is not as safe as a singular operator, whether it be a human or machine.

johnyi | May 27, 2018

I've seen claims on this forum that FSD is a totally different technology than EAP, rather than being built on top of it, but have never seen any proof. I would expect that FSD would be built on top of EAP, and leverage EAP for the basics and then add additional processing to do the routing and other things to get you from point A to B. Otherwise, why can't you pre-buy FSD without paying for EAP first? It would be like how TACC builds upon basic cruise control. It doesn't manage the throttle differently, but just adds additional intelligence to what's already there. And if you don't want TACC, Tesla still offers basic cruise control.

So to me, yes you will still need EAP functionality, and anyone who doesn't want to pay for L4/5 autonomy would still benefit from EAP. And for FSD to work, EAP (and AEB) would need to be better than it is today. For example, FSD won't magically see stopped firetrucks ahead if EAP and AEB can't - they all use the same sensors. If EAP can't read traffic lights and stop signs, how is FSD supposed to see them?

jjgunn | May 27, 2018

Trying to wrap my head around this....Don't both EAP & FSD work together in conjunction with each other?

carlk | May 27, 2018

EAP is just a subset of FSD. Any improvement in FSD will partially benefit EAP too,

K-D-R | May 27, 2018

I think by the time FSD is viable there will be so much competition for green cars with self driving technology that the distinction will be irrelevant and we’ll all be coveting a new version of a Tesla that doesn’t exist now. Sorry to those who prepayed.

K-D-R | May 27, 2018

I think by the time FSD is viable there will be so much competition for green cars with self driving technology that the distinction will be irrelevant and we’ll all be coveting a new version of a Tesla that doesn’t exist now. Sorry to those who prepayed.

EVRider | May 27, 2018

@hoffmannjames: There are a couple of reasons why I would want to continue using AP even when FSD becomes available. First, I have no need for FSD and don't want to pay extra for it. Second, and more important, I expect that once FSD becomes available, it will be in beta mode and require an attentive driver, just like AP today, and it will take a while for people to learn to trust FSD; by then I expect EAP to be much more reliable than it is today.

Today I use AP as it was meant to be used: as a driver assistant, not a driver. I keep my hands on the wheel, but I let AP "help" me drive, and that makes driving less stressful and more enjoyable for me. I'm still the one driving the car, whether or not I'm using AP.

sentabo | May 27, 2018

I'm with EVRider. I enjoy the AP on occasion, but have no interest in FSD.

bill | May 27, 2018

AP is another option for less money. So unless Tesla is not interested in making money for those that do not want or cannot afford FSD it makes sense to sell unless it requires significant effort to keep selling it.

hoffmannjames | May 27, 2018

@bill

The issue is that autopilot will be less safe than FSD. Why would Tesla offer customers the choice between a superior autopilot (FSD) or an inferior, less safe, autopilot? It would make more sense, I think, for Tesla to simply replace Autopilot with FSD.

SamO | May 27, 2018

Think of Tesla safety as an envelope that prevents manually driven cars from crashing. So Autopilot and FSD should eventually meet in the middle, with the human having intervention, as wanted.

But if FSD is available at the end of 2019 . . . Autopilot envelope is subsumed in 2022.

SO | May 28, 2018

Anything safety related will probably (hopefully) be done with both packages.

bp | May 28, 2018

When FSD has been validated and is approved for use, this is likely what will happen...

Current EAP is likely an adaptation of AP1 using the AP2 hardware. FSD is likely a different application, designed to operate without driver monitoring. When FSD is operating better than EAP, AP2 cars will use the FSD software in driver assist mode, with only 4 of the 8 cameras engaged. Software on AP1 cars will stick with the original AP implementation (because they lack the additional processing power).

EAP is likely to become a standard feature, included with all cars at no additional cost, like cruise control is offered today as standard on many cars (except Tesla). This is what Tesla has done with other features that were initially offered as "options" and are now included as standard (like the high amperage charger).

FSD is likely to stay as an additional cost option, at least until one of the manufacturers decides to make it a standard option.

EVRider | May 28, 2018

@hoffmannjames: I don’t agree that AP will be less safe than FSD. They use the same technology, and FSD can’t happen until the things that make AP unsafe are eliminated. By then, AP has to be good enough so that you don’t have to hold the wheel.

hoffmannjames | May 28, 2018

@EVRider

If Autopilot becomes so good that it becomes "highway hands free self-driving" whereas FSD is complete hands free self-driving on all roads then yeah, I could see Tesla offering two options for people who don't need or want the full self-driving experience but still want some self-driving assistance. But then we are changing the definition of Autopilot to essentially be "mini FSD". I am looking at what Autopilot is now, a L2 driver assist package where you should keep your hands on the wheel at all times. If Autopilot stays at that level, then I don't see it being needed anymore after FSD. But if Autopilot becomes L3 "mini-FSD" then sure, I could see Tesla keeping it as a cheaper option for people who want some FSD but cannot afford or are not interested in the full FSD package.

EVRider | May 28, 2018

@hoffmannjames: Maybe AP (without FSD) will never be completely hands free in all situations, but I expect that it should be able to do the things it does today well enough so drivers don't have to intervene because something isn't working as expected. For example, the car should never drift out of its lane, auto lane change should always work when it's safe to change lanes, and so on.

If AP remains available as a separate option (we can only speculate about that now), I can't see Tesla making it perform less capably than FSD would under the same circumstances.

hoffmannjames | May 28, 2018

@EVRider

Thanks. I appreciate your thoughts on this. Considering that when AP2 was revealed, Tesla offered 2 separate options (EAP & FSD) and this was even before FSD was available, seems to suggest that Tesla does not intend to get rid of AP when FSD comes out. Of course, this plan could change in the future. When FSD does come out, if most people skip EAP and just buy FSD instead, Tesla may have little financial incentive to keep EAP.

FREE ENERGY | May 28, 2018

FSD, links pls

dborn @nsw.au | May 28, 2018

At current rates of progress 5 years maybe, 20 years definitely!

EVRider | May 28, 2018

@FREE ENERGY: There's not a lot of information about FSD (Full Self Driving) on Tesla's web site, but you can find some of it here: https://www.tesla.com/models

bill | May 28, 2018

AP could be less functional but completely safe. For example you might need FSD to take and exit and follow the Nav and not hold the wheel. AP might not do those things but it could still be safe.

carlk | May 28, 2018

Giving it more thought I'm starting to see the point op made There is little reason for a separate EAP when FSD is widely adopted. Although I don't think that will happen in the very near future.

Uncle Paul | May 28, 2018

Maybe it would be something like Mercedes does for their 2 cruise control systems.

Basic system simply holds the vehicle at a set speed. Driver does not need to constantly adjust pedal pressure to maintain a desired speed when the vehicle goes up and down hills or around corners.

Advanced system utilizes basic system to maintain constant speed, but adds radar that adjustes speeds to safely maintain a set distance from the car ahead when they speed up or slow down.

Tesla base system wil offer speed control and distance control plus change lanes safely. Advanced system will navigate on and off ramps, and take the vehicle from point to destination without driver assistance. Will also automatically park the vehicle at destination and retrieve itself when called. Will also charge itself and return to home if directed.

removalrapidcar | May 29, 2018

I think Tesla would not get rid of AP when FSD becomes available. They might be considering that the car system will be more efficient if it has both functions. On the other hand,I think a car buyer would not want FSD.

Haggy | May 29, 2018

EAP won't be obsolete by definition if it doesn't fall from general use. Technology of that nature is still growing and is being added to other cars made by other companies.

In a way, FSD and the Model S (especially the P version) have conflicting design goals. People buy a Model S to drive it. They buy it because of the power and acceleration and handling, but nobody wants to be driven that way. FSD should drive about the same whether it's a P100D or a base 75. In essence, having FSD makes it like having two different cars. One is for times you don't want to drive.

AP doesn't decide anything for you, but relieves the tedium of long trips and stop and go traffic. Somebody who enjoys driving might not see the fun in being in control when driving down I5 at a fixed speed for several hundred miles while driving straight on level ground. Likewise, stop and go traffic isn't fun for most people. For those who want to automate the tedious but are fine driving at other times, AP is a good fit.

FSD doesn't build on AP by adding turns and traffic light/stop sign support, and follow a route. It's a different design. It views the world as objects rather than following lines. Thus there's no way to simply turn on pieces of it that currently exist in AP, because there's no reason to develop things in that particular order, perfecting certain parts first because they happen to be in AP. But once FSD is in place, Tesla may be able to use sections of code from FSD to handle AP functions. Or Tesla might decide that it makes no sense to do so, if FSD is using a more capable processor by the time it's released. It's too early to tell, but we can't assume that in the future Tesla won't be able to get current AP functionality to work as reliably as the same aspects of FSD, at least on capable hardware.

As it is, there are those who want things to be even more granular and would be willing to buy TACC alone. So there's room in the market for more than FSD.

brianp6621 | May 29, 2018

Some people are talking about AP vs FSD like it is 2 different sets of hardware. It isn't, all the sensors/inputs are the same (not counting the inactive camera's that AP doesn't rely on right now since Tesla has decided to build all AP tesla's with what they consider all the FSD hardware, whether FSD is capable on the current hardware is another matter). It is basically 2 different code bases that are programmed to do different levels of capabilities with the same hardware/inputs.

They serve 2 different purposes. One is a chauffeur, the other is a driver's aid.

I think EAP will be a subset of FSD, like it is marketed today to allow people to have choices. I think it will still have a place.

EVRider | May 29, 2018

I agree with @Haggy about the conflicting design goals -- performance won't matter when you're using FSD. Enjoy the performance while you can still drive your own car. :-)

SUN 2 DRV | May 29, 2018

Tesla already has tiers of Included and Optional AP features. The safety features are on all cars and EAP is optional. And even within EAP TACC is a separate feature from AutoSteer. No reason to think the Tesla won't just add a Tier for FSD... actually they already have established the FSD Tier and are charging for it but not delivering it yet.

steven | May 31, 2018

First thing people will do when FSD is released, is take it to the drag strip for a driverless run to shave off another tenth :-D

EVRider | June 1, 2018

Do drag strips have speed limits? :-)

eddyline | June 1, 2018

I don’t think that FSD will be completely safe until all road going vehicles have networked communication with each other. How can any standalone system in a car take into account the vehicle turning left in front of oncoming traffic; rapidly pulling out of a parking space; running a red light; jackrabbiting the green light; etc.? Human drivers are far too unpredictable for it to be as safe as the general public wants FSD to be. We will accept a higher level of risk of injury or death from people than we will from computers.

bp | June 1, 2018

Initially I agreed with eddyline's comment about having vehicle-to-vehicle and vehicle-to-infrastructure communications being critical for FSD.

However, relying on V2V and V2I for FSD may not be a good strategy. Those technologies are not guaranteed to be available in all circumstances, and it's uncertain how well they would be protected from malicious use.

For FSD to really be "safer than a human", it will likely have to rely on the same visual and audible information that human drivers use. With a 360 degree view, FSD has a significant advantage, able to track objects in all directions around the car (not possible with a human).

Though I'm concerned about tracking objects that cannot be seen - such as horns, whistles, sirens, etc. - audible signals human drivers use to alert them of a potential problem before they can see anything. Are the self driving systems also listening for those indicators???

eddyline | June 1, 2018

@hoffmanjames— I suppose it’s possible that the AI system will be able to learn the myriad cues that humans use to interpret possible actions and responses. An example from someone I know:

Training a new UPS driver, the trainer was observing the traffic and the driver. They were following a large delivery truck, when, for no reason the trainer could recall later, he told the driver to switch lanes. When the driver trainee questioned him, the trainer reached over and turned the steering wheel, swerving the UPS van across to the next lane, and just as that happened, the doors of the truck which was in front of them swung open and a forklift rolled out onto the road. The trainer probably saw the doors of the truck in front of them pushing open, but he couldn’t be sure why he did that.

hoffmannjames | June 1, 2018

@eddyline

A FSD car does not need to be able to interpret these cues from humans, it just needs to be able to avoid collisions. So, in your example above, an FSD car does not need to interpret why the doors were pushed open, it just needs to stay at a safe distance from the truck and when it detects the fork lift that had fallen onto the road, brake or change lanes to avoid a collision.

To me, this is a perfect example of actually making FSD unnecessarily difficult. You don't need to teach the car to read body language or micro signs or other cues from humans, just teach it basic defensive driving.

EVRider | June 1, 2018

@eddyline: Even if the government required all automakers to include V2V features in their new cars, it would be many, many years before all cars on the road had those features. FSD, whether it comes from Tesla or someone else, is going to happen long before then, and even if FSD eventually uses V2V in some way, it won't depend on it.

Uncle Paul | June 1, 2018

EVRIDER.

Drag strips do indeed have speed limits.
If you exceed certain trap speeds or quickness you will be required to add additional safety equipment to your car.
Roll cages, safety clothing like fire resistant suits and modern helmets, parachutes, fire supression etc.
Some are required to protect the driver, others the spectators.