After 5-weeks, I turned FSD Beta off!

After years of waiting I finally got my FSD Beta! After 6-weeks I turned it off.

I found the promise of the FSD Beta to be overblown. Yes it is very cool that the car will navigate on city streets to a destination. But the car is far too timid to be usable. It is worse than driving with a student driver. Also, the car’s hesitation causes a lot of problems with other drivers who get confused when they see the car stop or slow, when it should not. But that is not what made me turn it off.

I turned it off because it made the overall driving experience annoying, and more dangerous.

With the beta turned-on, the car makes speed changes at nearly every speed limit sign, even when only running TACC. Not only is this annoying but these changers often alarm my passengers and sometimes surprise the cars behind. The car does not do this when the FSD is disabled.

And then there is a notice at nearly every traffic light that one is coming, even when running only on TACC. I find I am spending time clearing nuisance alerts instead focusing 100% of my time on driving the car. And the sudden speed changes, if you miss one of the alerts, is just another version of the phantom breaking problem with the same set of issues for the cars behind. Note that I have this feature disabled under AP. This does not occur if FSD is disabled.

Next we have speed based lane changes. I disabled this under AP and the FSD stack just ignores this as well, and there is no way to tell it not to do so.

So overall, the benefit of having FSD is very minimal, while the negative impact to quality and enjoyment of driving is very real. I definably will not be buying FSD on a future car, unless these problems are fixed.

"Tesla Autopilot Engaged in Model X" by Ian Maddox is licensed under CC BY-SA 4.0.
Admin note: Image added for Blog Feed thumbnail

 

[dachshund]

If they can't make it work in the software, that's a lot of $$$ in hardware upgrades, or $5.44B to refund.

I'm curious if anyone has asked for a refund after paying full price for FSDb and deciding they didn't like it. It doesn't seem like an outrageous request. And I'm aware that we probably signed something that gets Tesla out of this. But still. It's a lot of money for something that infuriates some customers.

 

[dachshund]

The solid blue line is shared by NoA and FSD beta. If you dont see red lines marking the edges of roadways, then you are using TASS and/or NoA, not FSD beta at all, and in fact FSD beta is NEVER engaged on freeways (the red lines will vanish). It's NoA on freeways that is changing lanes, and that has its own on/off control that is distinct from FSD beta (if you actually have the beta, which I'm unclear about).

Thanks for the reminder. It does get confusing. It's a good thing the car is so enjoyable to drive without all this stuff. It's also good to know other brands have similar self-driving features. When we have enough superchargers for everyone then Tesla may have to do better.

 

[Alpaca]

Radar

 

[ewoodrick]

At this point, Elon Musk has realized what Mobileye already knows, risking end users entire safety on Visual only is risky, for multiple reasons. Depending on any one system for your entire safety has a long way to go. You need Radar and Video to drive safely.

So Tesla will put back in HD Radar and invest in this technology and it will continue to grow, in conjunction with Video, not one or the other.

I always love these uninformed types of post. "Vision will never cut it, Laser/Radar is required."

How in the crap do people drive every day?

Of course vision is good enough, 99.999999999999999% of the miles driven every day are by vision alone (okay, maybe add some sound)

Tesla really doesn't seem to be having much problem judging distances from video, It's figuring out what the heck to do with all that data is the problem.

 

[ewoodrick]

This past weekend, I was stopped at a McDonalds (No, FSD does not know how to handle a McDonalds drive-thru) and took off heading to a Costco 40 miles away.
It was early Saturday morning, so there really wasn't a lot of traffic, but I went through the heart of a medium sized city, down an Interstate and into a major mall area.
The car did it, by itself, brought me up to the front door of Costco. I didn't crash, I didn't damage the car, no one flipped me birds, no one shot at me. My passenger didn't throw up, no phantom braking in the Interstate. It would have been considered a normal drive by most.

Now, there are indeed times when the car doesn't do that well. As the rules say, I have to pay attention at all times and indeed the car can do some things when you least expect it (although I'm pretty good at knowing when it will have trouble). That's what we all agree to when we turn it on.

But the car CAN do it.

 

[dachshund]

I always love these uninformed types of post. "Vision will never cut it, Laser/Radar is required."

How in the crap do people drive every day?

Of course vision is good enough, 99.999999999999999% of the miles driven every day are by vision alone (okay, maybe add some sound)

Tesla really doesn't seem to be having much problem judging distances from video, It's figuring out what the heck to do with all that data is the problem.

I interpret the announcements of HW4 with radar as admitting that vision-only is an experiment that has fallen short.

And the promises have already started. “Limited release in March.” March… in some year not necessarily this year...

 

[ewoodrick]

I interpret the announcements of HW4 with radar as admitting that vision-only is an experiment that has fallen short.

And the promises have already started. “Limited release in March.” March… in some year not necessarily this year...

What a total and absolute misconstruing of the announcement. The Elon statement is that HW 3 WILL DO FSD. HW 4 will just do things a little faster and better.
In no way, form, shape, or fashion, did Elon indicate that HW3 in any fabrication fell short.
No where have I ever seen mention that Tesla is having any issues determining distance to objects. It's a pretty easy thing to do in vision, after all, you do it every day, don't you?

The biggest thing, I'm quite sure, about HW 4 is that it is just a lot newer. HW 3 is around 3 years old at this point. A 3 year old computer is not something that you would want to go out and buy today.

 

[SalisburySam]

No where have I ever seen mention that Tesla is having any issues determining distance to objects. It's a pretty easy thing to do in vision, after all, you do it every day, don't you?

Really? I do, but my car only does so because it has radar and USS still working. There is certainly evidence that Tesla THINKS determining distance is easy, as well as evidence they haven’t done it yet. That evidence being the functional losses to cars without USS…but yeah, that’s only temporary until your statement about vision being ”…a pretty easy thing to do...” becomes true. I’ve been known to gamble, but I wouldn’t take on that one with real money.

 

[Supcom]

Of course vision is good enough, 99.999999999999999% of the miles driven every day are by vision alone (okay, maybe add some sound)

Sound is optional. Deaf people drive cars.

 

[ewoodrick]

Really? I do, but my car only does so because it has radar and USS still working. There is certainly evidence that Tesla THINKS determining distance is easy, as well as evidence they haven’t done it yet. That evidence being the functional losses to cars without USS…but yeah, that’s only temporary until your statement about vision being ”…a pretty easy thing to do...” becomes true. I’ve been known to gamble, but I wouldn’t take on that one with real money.

And I'm running current software with radar turned off, still doing a great job.

The USS is a known difference and is probably existing because of the delay getting V11 out the door. I expect that this whole USS bitching will become irrelevant soon (as everyone has always said it would)

 

[kabin]

I interpret the announcements of HW4 with radar as admitting that vision-only is an experiment that has fallen short.

And the promises have already started. “Limited release in March.” March… in some year not necessarily this year...

Yep. There's been too many design changes based on faith. The project has seemingly been scientifically rudderless the last couple of years. Sensors add chaos and make production difficult so lets get rid of them and then later sheepishly add more cameras and radar. And we don't need to rehash the hardware shortcomings. FSD hasn't significantly matured and progress has all but stalled. If it continues, next on the hit parade is a loss in FSD customer confidence.

 

[BrendonHolt]

What a total and absolute misconstruing of the announcement. The Elon statement is that HW 3 WILL DO FSD. HW 4 will just do things a little faster and better.
In no way, form, shape, or fashion, did Elon indicate that HW3 in any fabrication fell short.
No where have I ever seen mention that Tesla is having any issues determining distance to objects. It's a pretty easy thing to do in vision, after all, you do it every day, don't you?

The biggest thing, I'm quite sure, about HW 4 is that it is just a lot newer. HW 3 is around 3 years old at this point. A 3 year old computer is not something that you would want to go out and buy today.

To be clear, literally, on a clear day with no vision issues, you are 100% correct. Again I stand with my position that for SAFETY reasons multiple systems will be required. I reverse this, what would be the intent to remove LIDAR or RADAR as a secondary system, I would contend COST only.

No one, including you, a true Tesla Advocate and Musk Supporter can come up with an argument why multiple technology would be cancelled for anything other then cost.

There is no TECHNICAL ADVANTAGE to remove LIDAR or RADAR, and peoples lives are literally on the line IF it fails, for any reason. No need, technology is there, see my other posts in regards to Automotive Sensor Technology.

 

[BrendonHolt]

Yep. There's been too many design changes based on faith. The project has seemingly been scientifically rudderless the last couple of years. Sensors add chaos and make production difficult so lets get rid of them and then later sheepishly add more cameras and radar. And we don't need to rehash the hardware shortcomings. FSD hasn't significantly matured and progress has all but stalled. If it continues, next on the hit parade is a loss in FSD customer confidence.

Many other EV and ICE Manufactures are working on this also. It is my high hope that this competition will remove this debate permanently, when Elon and more importantly our beloved Tesla realizes they are loosing their edge and up their game. These manufactures are holding off release but the technology they are using is already in Commercial Trucking Fleets.

Time will tell, but to be clear, any purchase now, based on faith, really is a leap in faith.

Buy Tesla now for what it is, a really great Awesome Electric Vehicle that has a well developed Charging Network and fast Charging Technology. It is cool for what it is. The FSD works for those willing to take risks and into the concept, but not production ready.

Again, my humble opinion, others are more then welcome to theirs. Time always tells.

 

[BrendonHolt]

I always love these uninformed types of post. "Vision will never cut it, Laser/Radar is required."

How in the crap do people drive every day?

Of course vision is good enough, 99.999999999999999% of the miles driven every day are by vision alone (okay, maybe add some sound)

Tesla really doesn't seem to be having much problem judging distances from video, It's figuring out what the heck to do with all that data is the problem.

Appreciate your brilliance, frankly I could be offended by your insult of calling me uninformed, but in reality, it is your OPINION. Frankly yes Video is fine, but has it's limitations, so yes, there is no argument against adding LIDAR or RADAR except COST. On a Foggy Day in the mountains in Pennsylvania, or Northern Pacific, how does that visual work, I can tell you RADAR and LIDAR is 100% accurate without vision.
Removal of Radar was another of Elon Musk's DUMB IDEAS. Do not worry he will return it and get back on board with two Sensor Technologies.

From one uninformed idiot to another, it is our Opinion, only time will tell what we end up with, as Tesla is all over the map, kind of nuts right now.

 

[dachshund]

What a total and absolute misconstruing of the announcement. The Elon statement is that HW 3 WILL DO FSD. HW 4 will just do things a little faster and better.
In no way, form, shape, or fashion, did Elon indicate that HW3 in any fabrication fell short.
No where have I ever seen mention that Tesla is having any issues determining distance to objects. It's a pretty easy thing to do in vision, after all, you do it every day, don't you?

The biggest thing, I'm quite sure, about HW 4 is that it is just a lot newer. HW 3 is around 3 years old at this point. A 3 year old computer is not something that you would want to go out and buy today.

Sorry I wasn't clear. I have one of the new MYs without any sensors, so I'm referring to the fact that HW4 is coming with new radar.

I don't think the set of EAP features will be restored for my car. Nor do I believe I'll have parking warnings. And this is specifically because Tesla is already announcing that HW4 is coming soon (sorry). Sure, it may not be out until next year. But doesn't this announcement tell you that they can't do those features without sensors of some kind?

Sure, V11 is coming and I'm always Suzy Sunshine. I can hope. But I feel like, "oh sure, HW4, fabulous... never mind your customers with HW3 trying to park."

 

[drtimhill]

I interpret the announcements of HW4 with radar as admitting that vision-only is an experiment that has fallen short.

And the promises have already started. “Limited release in March.” March… in some year not necessarily this year...

I think it really comes down to functionality. The earlier radar was pretty primitive, and generated data that wasn't really useful ("something might be moving fast somewhere over there .. maybe"), requiring vision to verify pretty much everything it got from the radar anyway. Better radar was of course available, but then you get into cost issues given Tesla's goals (Waymo, remember, aren't really cost constrained in their sensor suite). But like most tech, radar is improving very fast atm and has now reached a price/performance point where its practical and useful.

 

[Supcom]

I can tell you RADAR and LIDAR is 100% accurate without vision.

Until you come to a traffic light.

And no sensor is 100%.

 

[BrendonHolt]

Until you come to a traffic light.

And no sensor is 100%.

Agree, LIDAR and RADAR are near or equal to 100% on front end traffic. The fact none is 100% makes my case that you have to have at least two.

 

[Supcom]

Agree, LIDAR and RADAR are near or equal to 100% on front end traffic. The fact none is 100% makes my case that you have to have at least two.

Tesla has two forward cameras, so if one fails, the other should suffice long enough to stop the vehicle or for the driver to take over.

 

[drtimhill]

Appreciate your brilliance, frankly I could be offended by your insult of calling me uninformed, but in reality, it is your OPINION. Frankly yes Video is fine, but has it's limitations, so yes, there is no argument against adding LIDAR or RADAR except COST. On a Foggy Day in the mountains in Pennsylvania, or Northern Pacific, how does that visual work, I can tell you RADAR and LIDAR is 100% accurate without vision.
Removal of Radar was another of Elon Musk's DUMB IDEAS. Do not worry he will return it and get back on board with two Sensor Technologies.

From one uninformed idiot to another, it is our Opinion, only time will tell what we end up with, as Tesla is all over the map, kind of nuts right now.

ALL sensors have their limitations, and none, of course are 100%. For example, since both RADAR and LIDAR are active systems they are less sensitive to ambient conditions (light and dark), but are subject to interference, something that will get a lot worse as the technologies are more widely adopted in cars. Cameras cannot see in the dark, but as passive systems are not subject to interference from other cameras.

It is, of course, only my opinion, but I think the decision to stop using radar was related to cost. Tesla's strategy was to equip ALL cars with the hardware necessary for FSD (in their judgement), but this placed significant cost constraints on the sensor array, since this sunk cost could only be recovered based on the uptake of FSD upgrades. The result was a radar that was not adequate to contribute much beyond "the cameras should look over there" and basic "distance to the car in front", and hence the shift in focus to cameras only. Fast forward and the price/performance of radar has improved to the point where it is now economically feasible (and possibly advantageous) to utilize a more advanced (and thus useful) radar.