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

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

This site may earn commission on affiliate links.


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.”






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.

 
Last edited by a moderator:
It would be interesting for someone to study every date and calculate the true "Elon Time" conversion factor.
"3 months maybe, 6 months definitely" was Jan 23rd 2017.
Stop light and stop sign recognition was released April 26th 2020.

Scale factor is about 6? So maybe next fall for the rewrite? This forum should implement a gambling feature so we can all bet!
I feel like it will only be a couple months late this time.

Let's just say I'm not rushing over to the service center to have my HW3 installed. Need to wait for the next rev! For all we know those HW3s are going to overheat once they unleash the full power of the glorious quantum-leaped deep convolutional neural net.
 
Last edited:
I really think the only reason he gives timelines is to put pressure on the engineering team. Smart Summon was 6 weeks away on October 31st 2018 and wasn't released until September 2019.
Elon said 'soon' for HW2 autopilot software in 2016 and tweeted near Christmas that they were working weekends. I don't think it was completely done until a year later.
 
  • Like
Reactions: diplomat33
Unless I missed it, I don't think Elon announced anything new about FSD at the shareholder meeting today. He just repeated what we already knew, that the rewrite does labeling in 4D, and said it will be awesome.
 
  • Like
Reactions: emmz0r
Unless I missed it, I don't think Elon announced anything new about FSD at the shareholder meeting today. He just repeated what we already knew, that the rewrite does labeling in 4D, and said it will be awesome.

From the way he said it, I'm pretty sure it's further delayed. 10 days ago, it was 2-4 weeks for the private beta, and today, he said at least another month.

It's pretty crazy that Tesla is reinventing itself from the ground-up constantly, in almost every aspect of the company. I don't even know how they can keep it all together.
 
From the way he said it, I'm pretty sure it's further delayed. 10 days ago, it was 2-4 weeks for the private beta, and today, he said at least another month.

It's pretty crazy that Tesla is reinventing itself from the ground-up constantly, in almost every aspect of the company. I don't even know how they can keep it all together.

At the shareholder meeting, Elon did seem rather candid about the shortcomings of the current AP software. He said that the 2.5D software has reached what he called a "local maximum". Basically, it has hit a plateau of the best it can do. So Tesla, is doing the rewrite in order to overcome that maximum. I do think (hope) that the rewrite will be worth the wait.

I do applaud Tesla for reinventing itself and innovating when things don't work. I was impressed with Battery Day. I think the new battery cell has the potential to truly be the breakthrough that makes electric cars affordable and mainstream.
 
Pretty soon they might even reinvent their sensor suite! It will be revolutionary.

I won’t bet on it, knowing Elon’s hubris. Based on his comments, he’ll take the Current sensor suite to his grave. I personally think Tesla can achieve robotaxi fsd with the current sensors. I don’t believe sensors are the limiting factor, more the neural net and software side.
 
I won’t bet on it, knowing Elon’s hubris. Based on his comments, he’ll take the Current sensor suite to his grave. I personally think Tesla can achieve robotaxi fsd with the current sensors. I don’t believe sensors are the limiting factor, more the neural net and software side.

I would agree that the sensors currently aren't the limiting factor. Though it may be that the sensor suite is contributing to the delays in development, if it is making everything much more difficult to accomplish.
 
From the way he said it, I'm pretty sure it's further delayed. 10 days ago, it was 2-4 weeks for the private beta, and today, he said at least another month.

It's pretty crazy that Tesla is reinventing itself from the ground-up constantly, in almost every aspect of the company. I don't even know how they can keep it all together.

Yeah, I think I mentioned earlier in this thread how the rewrite is always n + 4 weeks out... No surprises there.
 
I think fewer sensors makes it easier, if you ignore the large number of corner cases.

I think I agree with this statement. If you don’t want to solve the problem, it makes a lot of sense to have a very simple sensor suite! Definitely makes it easier to complete the software.

This is a very novel argument for a simple sensor suite, one I have not heard before.
 
From the way he said it, I'm pretty sure it's further delayed. 10 days ago, it was 2-4 weeks for the private beta, and today, he said at least another month.

It's pretty crazy that Tesla is reinventing itself from the ground-up constantly, in almost every aspect of the company. I don't even know how they can keep it all together.

To be fair to Elon, I think when he spoke about the rewrite in July/August he did say October for early access. A month away is still October...

I dint think us peasants are getting it until December though, and that’s me being optimistic.
 
Wow, did Tesla really enable traffic controls in UK and EU? I never thought it'd happen. The traffic controls in the UK are bonkers.

Tesla Traffic Light and Stop Sign Control begins international rollout

I don't think you guys appreciate the utter significance of Tesla releasing the traffic control feature world wide. My mind is beyond blown. I didn't think this was possible, and no other company is anywhere remotely close. I thought Tesla would find some feature-breaking issue, and it would stall their development. What we fail to appreciate constantly is that Tesla is pushing the reliability and accuracy of NNs beyond what is understood to be possible. Just imagine the size and depth of the NNs and data needed to roll out a consumer worthy traffic control feature world wide...

FSD development is accelerating at Tesla, whereas it's been stagnant or constant at others for a long time.

I think you clearly misunderstanding what's going on here. The NN needed for international rollout of traffic light and stop sign control already existed in Mobileye's EyeQ4 that went into production in Q4 2017.

It had NN for traffic signs, lights and road signs, etc for all major countries. It also had NN for relevance of traffic light. For example which traffic light handles which lane/lanes and types of lights, etc.
Tesla is actually late in this regard. All tesla did was add a basic actuation response control that would respond to the output of their NN.

Its basically the same as stopping for a car ahead of you. As Geohot says, "you treat the red traffic light and stop line like another vehicle" Its ACC for red lights/stop sign on the stop line. Its the exact same stopping code.
This is why it stops on the line or stops so back and away from the intersection. There's no intelligence in the current control system.

BMW already experimented with this feature using the 2017 EyeQ4, I can even say that Tesla copied the way they presented the feature.
For example they presented it as a separate feature that you can enable in addition to adaptive Cruise Control
And Lane Keeping. So you can for example just have ACC and Traffic/Stop sign control on.

But because these OEM and their tier1s move at a snail pace, it will be probably another year till you see it in a car, yet it will be using a chip from Q4 2017.. Which makes it appear that Tesla is so ahead, but its more like the traditional OEM are so slow to implement any technology.

 
I think I agree with this statement. If you don’t want to solve the problem, it makes a lot of sense to have a very simple sensor suite! Definitely makes it easier to complete the software.

This is a very novel argument for a simple sensor suite, one I have not heard before.
Tesla is trying to solve the general case before moving on to the more complicated corner cases. Makes sense (haha pun) to me, even though I don't 100% agree with it.
 
  • Funny
Reactions: AlanSubie4Life