Blog Musk Touts ‘Quantum Leap” in Full Self-Driving Performance

A “quantum leap” improvement is coming to Tesla’s Autopilot software in six to 10 weeks, Chief Executive Elon Musk said a tweet.

Musk called the new software a “fundamental architectural rewrite, not an incremental tweak.”

https://twitter.com/x/status/1294374864657162240

Musk said his personal car is running a “bleeding edge alpha build” of the software, which he also mentioned during Tesla’s Q2 earnings. 

“So it’s almost getting to the point where I can go from my house to work with no interventions, despite going through construction and widely varying situations,” Musk said on the earnings call. “So this is why I am very confident about full self-driving functionality being complete by the end of this year, is because I’m literally driving it.”

Tesla’s Full Self-Driving software has been slow to roll out against the company’s promises. Musk previously said a Tesla would drive from Los Angeles to New York using the Full Self Driving feature by the end of 2019. The company didn’t meet that goal. So, it will be interesting to see the state of Autopilot at the end of 2020.

 

[diplomat33]

Guess my point is that your debate is purely academic. Nobody wanting to purchase Waymo FSD can do so.

Nobody can buy a Tesla robotaxi yet either. My point is that you have to solve FSD first. That's what Waymo is working on doing.

They might be able to work within tight parameters, but cannot function out on the open highway.

Waymo cars can drive on the open highway.

Have not kept up with Waymo recently. Are their cars still requiring human minders, or are they good enough to be autonomous.

Waymo cars are fully autonomous.

 

[ItsNotAboutTheMoney]

Nobody can buy a Tesla robotaxi yet either. My point is that you have to solve FSD first. That's what Waymo is working on doing.



Waymo cars can drive on the open highway.



Waymo cars are fully autonomous.

They are not fully autonomous if there's intervention every 13k miles.

 

[diplomat33]

They are not fully autonomous if there's intervention every 13k miles.

You can have both. Fully autonomous means that the car is capable of doing all the driving with no human input. The fact that Waymo does some drives with no human in the driver seat supports that. Disengagements can still happen because the safety driver did not like something. The car is still capable of doing the driving. For example, the safety driver disengaged because the car was taking too long to execute an unprotected left turn. The car is capable of doing unprotected left turns and has done many in the past and might have done this one too if the safety driver had waited longer. But the safety driver disengaged because he/she did not like how the car was doing the unprotected left turn (taking too long). So you can have a car that is fully autonomous because the car is capable of doing all the driving and in fact does all the driving and still have disengagements because of how it is driving.

 

[Redbrick]

Since this is a topic on FSD, what exactly were you going to go with, in place of Tesla?

the kona ev...still my preferred to this day or the leaf. I drive this model 3 at sufferance...

 

[nukequazar]

It seems that some ppl might be following this just from a technology point of view.
To me - as I stated in many places, FSD is a safety issue, manual (human) driving should be banned and considered dangerous, but we need a replacement to the manual driving, hence FSD.

My selfish goal is that my kid who is 10 years old, can get into my hand-me-down Model 3 and not have to do the driving by the time she needs to drive.
I only view the FSD solution through the lens of road safety, for those in and outside the car.
Manual driving can only be banned when FSD is ubiquitous and it can only be ubiquitous if it is affordable to the most ppl.

There is zero chance that your Model 3 as it is will “do the driving” for your daughter in six years. I will extend a small chance that with a hardware upgrade costing anywhere from $10,000 to $50,000 to add LIDAR (or some other yet to exist advanced sensor technology) and HD mapping, your car may achieve Level 3-5.

 

[diplomat33]

There is zero chance that your Model 3 as it is will “do the driving” for your daughter in six years. I will extend a small chance that with a hardware upgrade costing anywhere from $10,000 to $50,000 to add LIDAR (or some other yet to exist advanced sensor technology) and HD mapping, your car may achieve Level 3-5.

I agree that our cars will need extra sensors to achieve reliable full autonomous driving (L4-L5). But I would nitpick about the cost. I doubt that in 6 years, the hardware upgrade will cost $10k-$50k. The price of lidar is already coming down now. I am sure in 6 years, the cost of lidar will be even lower.

 

[mikes_fsd]

the kona ev...still my preferred to this day or the leaf. I drive this model 3 at sufferance...

Again, the question was "Since this is a topic on FSD, what exactly were you going to go with, in place of Tesla?"
How is Kona or a Leaf even in this conversation? BTW my first EV was a Leaf.

 

[mikes_fsd]

This is only true if the software is good enough and the sensors are good enough. For example, if you just have cameras and your camera vision is not very good, then no, it will not be better than humans. Conversely, if you have 360 degree advanced camera vision and radar and lidar

Now I know you are trolling or have completely deluded yourself to the point of being a clown.

You do know that Lidar is not magic, and you have to write software to process all the lidar inputs, right?
You have to have Lidar Vision - in a sense.
Then you also must have Camera Vision (since - for starters - you will not be able to detect traffic light color with lidar)
Then you have to have both of those data streams plan nice - so you are doing a lot more software stitching fu then just simply/only solving for camera vision.

 

[diplomat33]

Now I know you are trolling or have completely deluded yourself to the point of being a clown.

You do know that Lidar is not magic, and you have to write software to process all the lidar inputs, right?
You have to have Lidar Vision - in a sense.
Then you also must have Camera Vision (since - for starters - you will not be able to detect traffic light color with lidar)
Then you have to have both of those data streams plan nice - so you are doing a lot more software stitching fu then just simply/only solving for camera vision.

Yes, you need to do extra work to write lidar vision and fuse the data from both sensors. I know that! And I think that is the better approach! Yes, it is more complex and requires more work but it gives you better perception. And Waymo has done all those things. That's why Waymo has such solid perception because they are collecting data from multiple sources and can form a more complete picture of the world.

Camera-only might be simpler but it will be LESS reliable. Even the best camera vision will never be perfect. It will still miss things from time to time. And if the camera vision misses something critical, you have a risk of an accident. And you have no back up if the vision is wrong. You have no way to verify if the vision is right or wrong since you have no other sensors to check. You have no alternative to correctly detect an object if the camera is not able to detect the object. The only way to solve these problems is by keeping the driver in the seat so that the driver can be the back-up if something is wrong. That's why camera-only cannot give you FSD reliable enough to remove the driver!

Let me make it simple for you: with camera-only, what you do suggest the self-driving car do if the camera vision is wrong? What do you suggest the self-driving car do if the camera vision is not able to detect something? What do you suggest the self-driving car do if the camera vision is wrong but the car does not know that the camera vision is wrong since it has nothing to compare it to?

 

[mikegre]

So we should just trust Elon that Tesla will solve an incredibly difficult problem like FSD just because the stock price is going up? LOL. Stock price has nothing to do with FSD. The stock price is shooting up because investors are speculating and pushing the price up.

Just because FSD may be an "incredibly difficult problem" for you. doesn't mean it's an incredibly difficult problem for Tesla engineers. FSD is not that difficult.

 

[nukequazar]

Now I know you are trolling or have completely deluded yourself to the point of being a clown.

You do know that Lidar is not magic, and you have to write software to process all the lidar inputs, right?
You have to have Lidar Vision - in a sense.
Then you also must have Camera Vision (since - for starters - you will not be able to detect traffic light color with lidar)
Then you have to have both of those data streams plan nice - so you are doing a lot more software stitching fu then just simply/only solving for camera vision.

The only clowning going on here is your Tesla Fanboy nonsense. Waymo has autonomous, Level 4 cars. Tesla has Level 2 TACC. We can start the discussion there otherwise this is a complete waste of time.

You can argue that somehow magically our cars will approach what LIDAR plus HD mapping plus cameras plus sonar plus RADAR can do with only Google maps plus cameras plus sonar plus RADAR then great but I think there is zero chance of our cars with existing sensors and mapping will ever get past Level 2. Even uttering the term "full self driving" with respect to current Tesla cars is disingenuous, i.e. a CEO/marketing lie.

Tesla makes some great products but the dishonesty and greedy behavior of the CEO are too much to handle.

 

[diplomat33]

Just because FSD may be an "incredibly difficult problem" for you. doesn't mean it's an incredibly difficult problem for Tesla engineers. FSD is not that difficult.

This is the most ignorant thing I've read on this forum in a long time. It has nothing to do with me not understanding FSD. The whole industry says that it is a difficult problem. And if it is so easy to Tesla engineers, why hasn't Tesla solved it then? Why is Tesla still at L2? Why hasn't Tesla even solved phantom braking yet? Why is Tesla lagging behind everybody else in the race to FSD? FSD is easy yet Tesla still has no FSD after 4 years.

 

[mikegre]

This is the most ignorant thing I've read on this forum in a long time. It has nothing to do with me not understanding FSD. The whole industry says that it is a difficult problem. And if it is so easy to Tesla engineers, why hasn't Tesla solved it then? Why is Tesla still at L2? Why hasn't Tesla even solved phantom braking yet? Why is Tesla lagging behind everybody else in the race to FSD?

Because of Tesla, I am now able to go anywhere in perfect safety without ever having to touch the steering wheel, the brake or the electron accelerator pedal (the "gas" pedal).

I've simply taken a tiny bit of the profits I've garnered from Tesla stock and hired myself a chauffeur. Have a nice weekend.

 

[diplomat33]

Because of Tesla, I am now able to go anywhere in perfect safety without ever having to touch the steering wheel, the brake or the electron accelerator pedal (the "gas" pedal).

So you've never had to disengage AP even once? Can your Tesla make unprotected left turns without touching the steering wheel? We can do portions of a drive without having to steer or brake, yes. But we cannot go anywhere in perfect safety without ever touching the wheel or brakes. That is not accurate.

 

[nukequazar]

So you've never had to disengage AP even once? Can your Tesla make unprotected left turns without touching the steering wheel? We can do portions of a drive without having to steer or brake, yes. But we cannot go anywhere in perfect safety without ever touching the wheel or brakes. That is not accurate.

He's not talking about his Tesla. He's being an arrogant SOB, saying he profited so much off the overinflated TSLA stock that he hired a driver. This is why people hate Tesla drivers even more than they hate BMW drivers.

 

[diplomat33]

He's not talking about his Tesla. He's being an arrogant SOB, saying he profited so much off the overinflated TSLA stock that he hired a driver. This is why people hate Tesla drivers even more than they hate BMW drivers.

Thanks. I get it now. I thought he was saying his Tesla could drive him around.

 

[nukequazar]

Just because FSD may be an "incredibly difficult problem" for you. doesn't mean it's an incredibly difficult problem for Tesla engineers. FSD is not that difficult.

This is one of the dumbest things I've seen written on the internet. Autonomous driving cars is probably the most difficult problem that's ever been attempted to be solved for a consumer product. For sure it's much more difficult than flying to the moon or mars. We flew to the moon off of slide rules and some brave men strapped to silos full of rocket fuel. Done. Autonomous driving cars requires advanced sensors, advanced mapping, supercomputing, and years of testing.

 

[DanCar]

Autonomous driving cars requires advanced sensors, advanced mapping, supercomputing, and years of testing.

And dealing with:

1. angry motorists
2. motorists who aren't paying attention
3. Hazards such as pot holes and mattresses
4. People who are suicidal, jumping in front of vehicle
5. John Conner wanting to save us from skynet
6. etc...

 

[avesraggiana]

Yawn...what would be a “Quantum Leap” is if Elon Musk hired somebody to pick up the phone.

 

[Redbrick]

Again, the question was "Since this is a topic on FSD, what exactly were you going to go with, in place of Tesla?"
How is Kona or a Leaf even in this conversation? BTW my first EV was a Leaf.

Kona ev ultimate because of it's life time battery warranty, build/material quality (based on my experience with Hyundai cars) , and simple driver assist...nothing pretentious.. leaf sL plus, because of it's price. So that was and still is my thinking in order of priority. I'd done research on tesla cars, and remain very unimpressed and bitter with the quality, and the service. Hope to rid this pos car as soon as my wife is over it. .(soon enough there will be a 2020 LR+fsd for sale)..I don't see the hype...sorry.