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

Autopilot a fraud? The linked author bets Elon he'll eat a hat if it changes lanes!

This site may earn commission on affiliate links.
I sold my Model S to buy a new one specifically because of the "auto pilot" feature, and Elon's promise that soon (June or July) software will enable the auto-steering function of the car.

I will not be too upset if the car only keeps its lane and does not automatically change lanes and pass for you. But I didn't know it was impossible - as this guy says.

So, I'm asking my very educated forum fellows - what do you think about this?

Tesla’s game-changing, lane-changing Autopilot? “Currently, there is no radar on the market that can achieve that,â€￾ Toyota engineer says - Daily Kanban

Just for fun let me ask a second question: Do you think Tesla would be smart to make plans to replace sensors on the cars it sells with "auto pilot" hardware as needed to keep up with the industry in the next 5 years? Or should Tesla just say "tough luck" to those of us who buy the car now, and get stuck with very limited auto-pilot ability compared to a car that comes out in a few years than can over-take on its own?

Is customer goodwill worth anything to Tesla?
 
The "change lane" part has already been demonstrated. The pass part can be done with TACC now.

The car will be able to determine that the space to its left is clear. What it will not be able to do, based on current sensors, is determine that the car can change lanes without the driver first verifying that it is safe to do so. (There's not enough rearward visibility with hardware to determine if a car is speeding down the adjacent lane).

Tesla has never claimed that the car will make the determination of whether a lane change is safe to make...only that the space adjacent to the car (and slightly behind) is clear of vehicles, and prevent the automatic lane change if it is not.

So where's the fraud?

Regarding your second question, no--Tesla should not replace sensors. That's why anyone has the right to buy a new car when they want to. It's not about goodwill...it's about not taking on a very low margin--possibily a negative margin--upgrade, when their efforts should be spent on making cars.
 
Last edited:
The automatic lane change feature will require human activation. It will change lanes after a user initiates a lane change but the vehicle won't be overtaking cars by itself. The hardware isn't safe enough to allow for fully automatic lane changes. Think of it as "assisted lane change".
 
I actually took time to read some of the article. Did he really not consider the possibility that the driver must first ascertain whether a lane change is safe? I find it surprising that this thought never crossed his mind.

Then again, maybe not. reading the first few paragraphs, he sounds like a crybaby, upset that (insert car company here) is not getting any attention.
 
Agreed, the current "autopilot" hardware is not sufficient for automatic lane-changing, or even driver-initiated lane-changing.

The problem with the automatic lane-changing, as the author says, is that you need omnidirectional radar; the short-range sonar is not sufficient to detect a car coming up quickly in the target lane.

And the problem with driver-initiated lane-changing is that per the DMV, you're required to signal for several seconds before merging (5 seconds on the highway), so when the driver tips the stalk to merge, the car will actually have to signal for a few seconds and then merge. A lot can happen in those few seconds, and the driver will likely not be looking in the rear view mirror when the lane change begins, so the danger is equivalent to the automatic lane-changing above.

So for these reasons, I expect that the Model S will eventually have rear-facing and likely side-facing radar to be able to accomplish merges safely. Who knows whether these will be retrofittable. For now, I'm perfectly happy with my Signature non-autopilot P85 :)
 
If indeed that's true, how does that affect an autopilot lane change?

Simple solution: The car could initiate the lane change when the turn signal is released, not when pressed. The driver could be required to hold the stalk to the first detent, then the change is initiated at release.

But frankly, if the car requires me to signal for 5 seconds before changing lanes, I'll never use it, and I suspect most others wouldn't either.

Unless the 5 second requirement is required of manufacturers by law, I don't see this 5 second delay being used. It's not realistic.

And if it is required by law, then it seems nobody would waste their time trying to use it.
 
If you're trying to initiate an automatic lane change in congested traffic you may need to have your head checked. TMC states that the driver is still always responsible for the care and control of their vehicle. So an unsafe lane change is still just that. But if you're traveling on open highway and initiate the lane change - the current hardware absolutely is adequate for an automated lane change. Knowing that you don't have long-range radar surrounding the car should be enough to tell the driver to employ caution and common sense when going AP.

Anyone remember the guy in the motorhome that activated cruise control on a highway then proceeded to walk to the back where he sat down with his family? His motorhome drifted off the road and crashed. I hope that type of person isn't in the market for a Tesla.

Maybe I'm missing what the debate is about.
 
If you read the article, he also calls into doubt auto-steering. He says maps are not good enough at this point in time (meaning, I think, the car cannot know where it is supposed to be on the road and where the road is going). I will admit I'll be extremely upset if Tesla decides they can't do the auto-steer even to maintain lanes. But I do wonder about it, because here in Houston there's so much construction I have a hard time keeping up with the lane myself. And on 290 there is a concrete barrier to the left side of the highway so close your mirror will hit it if you drift just the least little bit - due to construction. So must admit I'm a little concerned Tesla will come through.
 
If you read the article, he also calls into doubt auto-steering. He says maps are not good enough at this point in time (meaning, I think, the car cannot know where it is supposed to be on the road and where the road is going). I will admit I'll be extremely upset if Tesla decides they can't do the auto-steer even to maintain lanes. But I do wonder about it, because here in Houston there's so much construction I have a hard time keeping up with the lane myself. And on 290 there is a concrete barrier to the left side of the highway so close your mirror will hit it if you drift just the least little bit - due to construction. So must admit I'm a little concerned Tesla will come through.

Lane keeping will be utilizing the front-facing camera to determine where the lanes are, not just maps and GPS. This allows for real-time recognition and lane keeping in areas where there have been recent road work, construction, detours, etc.

Automated lane change will not look behind to determine if it's safe to change lanes. That is the driver's responsibility. All it can do is determine if there is a car in the blind spot. It has no way of knowing if there is a car coming up on you in an adjacent lane at 80 MPH that is only two car lengths behind. Something like this could only be determined with a rear or side oblique radar. You can bet that rear and side radars will not be offered for retrofit.
 
That is the driver's responsibility. All it can do is determine if there is a car in the blind spot. It has no way of knowing if there is a car coming up on you in an adjacent lane at 80 MPH that is only two car lengths behind. Something like this could only be determined with a rear or side oblique radar. You can bet that rear and side radars will not be offered for retrofit.

That is the entire point of the article...

However, the author went one step further and said that without rear-facing radar, this feature would be so unsafe that Tesla won't dare release it.

I would not be entirely surprised if he turns out being correct. But unlike him, I wouldn't bet my hat though.
 
That is the entire point of the article...

However, the author went one step further and said that without rear-facing radar, this feature would be so unsafe that Tesla won't dare release it.

I would not be entirely surprised if he turns out being correct. But unlike him, I wouldn't bet my hat though.

The problem with the article is that he assumes (erroneously) that when Tesla says "automatic lane change", that it means that the car decides that it's safe to change lanes. Tesla, to my knowledge, has never made that assertion (other than ensuring the area adjacent to the car is free of obstructions).
 
The automatic lane change feature will require human activation. It will change lanes after a user initiates a lane change but the vehicle won't be overtaking cars by itself. The hardware isn't safe enough to allow for fully automatic lane changes. Think of it as "assisted lane change".

the biggest problem with this feature is there will be people that do not realize this because nobody reads the manual. they will assume the car will only do so if safe and will activate the lane change when there are fast approaching cars and there will be accidents caused by this. even if the owner realizes this there may be others borrowing the car that do not. it is just inevitable.
 
Given that Infiniti, Mercedes, and probably most other luxury car makers already build cars with the same lane holding feature, it seems unlikely that Tesla would decide they can't do it.

I will admit I'll be extremely upset if Tesla decides they can't do the auto-steer even to maintain lanes. But I do wonder about it, because here in Houston there's so much construction I have a hard time keeping up with the lane myself. And on 290 there is a concrete barrier to the left side of the highway so close your mirror will hit it if you drift just the least little bit - due to construction. So must admit I'm a little concerned Tesla will come through.
 
Given that Infiniti, Mercedes, and probably most other luxury car makers already build cars with the same lane holding feature, it seems unlikely that Tesla would decide they can't do it.

I'm sure they will do lane holding. I'll be more than upset if we don't get that.

But lane changing is another ball of wax and I'll never use it even if they end up delivering it.
 
The problem with the article is that he assumes (erroneously) that when Tesla says "automatic lane change", that it means that the car decides that it's safe to change lanes. Tesla, to my knowledge, has never made that assertion (other than ensuring the area adjacent to the car is free of obstructions).

I've always assumed it was in the American sense of automatic in "automatic door", where you press a button to open the door. The car could say "Lane change required in n seconds. Please check for traffic and press turn signal to change lanes"..

As if being able to change lanes with a turn signal isn't cool enough. Some people want the moon on a stick.
 
The problem with the article is that he assumes (erroneously) that when Tesla says "automatic lane change", that it means that the car decides that it's safe to change lanes. Tesla, to my knowledge, has never made that assertion (other than ensuring the area adjacent to the car is free of obstructions).

and when they release the feature he'll say it didn't meet his requirements and the bet is null and void and thus won't eat a hat.
 
Ok, this is a simple logic problem. No matter how good the side and rear sensors are (whether just ultrasonic as they are now, or ultrasonic + camera + radar as some would have liked), the driver is still responsible for the actions of the car. As such, he or she is compelled to shoulder check and perform due diligence before initiating a lane change. That will never go away until we have truly autonomous cars. Therefore any sensors we do have now are nothing more than conveniences and an extra safety feature. To argue otherwise is like saying the automatic emergency braking feature should never have been turned on (or can't be turned on) because the driver has to look forward when driving to avoid front end collisions!
 
Can't wait for lane holding. That will be very cool. However, I'd never use the auto lane changing feature myself.

I guess I don't understand your concerns with the lane changing.

I see the functionality of it as, when driving on a multi-lane highway with TACC and lane-holding active, I'm approaching a slower-moving vehicle in front of me and I check my mirror (or camera) to see that a lane change is safe, tap the turn signal, and the car changes lanes to pass the slower moving vehicle. Once clear of the car I pass, I check that it is safe, then tap the turn signal again and the car returns to the original lane.

Are you saying that you will not use lane-holding, or that you will change lanes manually? I don't know what the car will do if you force it out of the lane. Will it cancel lane holding and you will have to restart it?

I guess I don't understand if you think it is unnecessary, or unsafe, or you just want to be in control.
 
Last edited:
Article writer's PS "If that Autopilot changes lanes automatically at the touch of a blinker stalk by, say, one year from now, in a series Model D, I will publicly eat a Tesla baseball hat on Twitter, Facebook and YouTube. How about it, Elon?"

Maybe some salt and pepper might help with the taste.