Tesla Master Plan: Part Deux

  • Thread starter CodeRedR51
  • 1,518 comments
  • 116,858 views
Tesla puzzles me. The tech is obviously great, but I can't help but feel like they're not doing much to reach broader segments of the population. It should be clear to anyone that millennials, the giant group of people quickly approaching middle aged and increasingly with money to spare, are very interested in adventure/multi purpose vehicles. Things like the Rav4 on the lower end of the spectrum all the way up to kitted out overlanding Sprinters at the top/extreme end. I don't see the Tesla Cybertruck filling this role, it's just too impractical and conspicuous and the Y/X aren't really suitable due to their on road limitations. Maybe not all millennials are into adventuring....but vanishingly few want sedans. Like what is Tesla's long term plan? Are they just going to keep selling sedans and "tall sedans" to the same aging white collar customers in big cities? Its weird to me that they don't seem interested in expanding their customer base. I honestly feel like we are approaching a point where everyone who wants a Tesla has a Tesla. Then what? After all, a big selling point is the cars last a long time / need less maintenance due to the electric drivetrain. Whether the Model S goes from 0-60 in 2.5 seconds or 1.99 seconds is not going to pull people into a Tesla showroom who wouldn't have been interested before. We get it, they are unnecessarily fast. I guess what I'm wondering is....after the Cybertruck and Roadster (both niche products that I don't expect to be huge sellers) what's next for Tesla? Maybe they are happy with their market position and will just keep churning out the S3XY lineup for the foreseeable future...it just seems like they are going to run out of customers.

tl;dr: It's hard for me to understand how Tesla achieved such towering EV market dominance and then just didn't respond to other players developing EVs for pretty consequential segments of the market. That there is so much space for Rivian, Ford, and even GM to get EV trucks and (proper) SUVs to market (and ahead of Tesla, at that!) just seems like such an own goal for Tesla. Musk has always maintained that he wanted electric cars to be the future regardless if they wore a Tesla badge...so maybe this lack of territory defense stems from that...but I somehow doubt it. Certainly Tesla stockholders aren't counting on anything less than Tesla dominating the entire automotive landscape.
 
R1600Turbo
Faster than a McLaren P1 0-100-0 too. We'll see what happens once they start hitting circuits.
Click to expand...
So, it's basically still just really fast in a straight line b/c of instantaneous torque.

If the benchmark is a near-decade old supercar, that's not really something to write home about, either. I think you've far too high of expectations to think these are going to put down any sort of record-setting lap times. You made a note on the last page that they likely need upgraded brakes & fluids for track duty.
 
Last edited:
The Model S weighs 5000 pounds, with little in the way of aero devices and tires smaller than Corvettes from the 90s. At a certain point, "goes fast real good, has low center of gravity" stops being a trump card that overwhelms everything.
 
Last edited:
McLaren
So, it's basically still just really fast in a straight line b/c of instantaneous torque.

If the benchmark is a near-decade old supercar, that's not really something to write home about, either. I think you've far too high of expectations to think these are going to put down any sort of record-setting lap times. You made a note on the last page that they likely need upgraded brakes & fluids for track duty.
Click to expand...
Tornado
The Model S weighs 5000 pounds and has tires smaller than Corvettes from the 90s. At a certain point, "goes fast good, has low center of gravity" stops being a trump card that overwhelms everything.
Click to expand...
I had a conversation with a Tesla PR in the UK about this. Essentially I asked about doing track days in one - relating to range and battery derating - and after a bit of chatter between themselves they couldn't quite decide whether it would be the brakes, the tyres, or the wheels themselves that would fail first, but it would be one of them and it would be well before the battery came into consideration. Of course that was just a regular P85D, so maybe the fast model with the 1980s pop culture meme name has upgraded components to survive the duty.

They also informed me that there's no spare tyre. That's essentially down to the fact that they cannot package a portable jack even close to capable of holding the corner weights of a Model S/X within the price of the car and surpisingly few people ever actually get a puncture, but if you have a puncture you're waiting for roadside assistance. I understand that you can buy one as an optional extra though.
 
McLaren
So, it's basically still just really fast in a straight line b/c of instantaneous torque.

If the benchmark is a near-decade old supercar, that's not really something to write home about, either. I think you've far too high of expectations to think these are going to put down any sort of record-setting lap times. You made a note on the last page that they likely need upgraded brakes & fluids for track duty.
Click to expand...
I'm not aware of any that has tested it on a circuit yet so can't really judge it just yet except for Motor Trends skidpad figures.

Edit: a truck load of Plaid Model S's just went by me going East on I-10.
 
Last edited:
FSD Beta 9.0 is going well.

It cannot see Seattle's monorail columns, they are invisible to the system. It consistently tries to drive straight into them.

(language warning username and video)

 
Last edited:
They said they were getting rid of their radar systems and literally every engineer on the planet except those who work for Tesla said, "That's a bad idea".

It was a bad idea.

I also notice that none of these systems from any company working on autonomous tech ever seem to be tested on snow-covered roads up north. What's that all about?
 
Last edited:
Relying solely on vision is a terrible idea. I took a computer vision class last semester and when I asked my professor he said it's possible. I found multiple papers he has worked on and here are some quotes.

[1812.07179] Pseudo-LiDAR from Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Driving

arxiv.org

our approach achieves impressive improvements over the existing state-of-the-art in image-based performance --- raising the detection accuracy of objects within the 30m range from the previous state-of-the-art of 22% to an unprecedented 74%.
Click to expand...
LiDAR and pseudo-LiDAR lead to highly accurate predictions, especially for the nearby objects. However, pseudo-LiDAR fails to detect far-away objects precisely due to inaccurate depth estimates. On the other hand, the frontal-view-based approach makes extremely inaccurate predictions, even for nearby objects.
Click to expand...

[1906.06310] Pseudo-LiDAR++: Accurate Depth for 3D Object Detection in Autonomous Driving

arxiv.org

Instead one can use commodity stereo cameras for the former and a cheap sparse LiDAR to correct the systematic bias in the resulting depth estimates.
Click to expand...

In both papers, it seems like a vision only system would only work well in conjunction with a LIDAR system. At this time, the algorithm just isn't accurate enough. It definitely is a huge safety issue to allow people to beta test these on public roads, thinking that it's completely autonomous.


https://www.businessinsider.com/tesla-full-self-driving-fsd-beta-v9-update-elon-musk-2021-7

"Full Self-Driving is in early limited access Beta and must be used with additional caution. It may do the wrong thing at the worst time, so you must always keep your hands on the wheel and pay extra attention to the road. Do not become complacent," the beta v9 release notes said, according to one driver's screenshot.
Click to expand...

Really confidence inspiring.


Keef
I also notice that none of these systems from any company working on autonomous tech ever seem to be tested on snow-covered roads up north. What's that all about?
Click to expand...
Because getting it to work in perfect conditions still isn't completely feasible. Waymo and Argo.AI (I applied for a summer internship, but they ghosted me lmao) are still testing in quite restrictive areas, because they're not confident for the systems to work in all conditions. I think for the foreseeable future, all these systems will have quite restrictive operating domains
 
Last edited:
Keef
They said they were getting rid of their radar systems and literally every engineer on the planet except those who work for Tesla said, "That's a bad idea".

It was a bad idea.

I also notice that none of these systems from any company working on autonomous tech ever seem to be tested on snow-covered roads up north. What's that all about?
Click to expand...
Waymo has one of their test fleets just across the street from my office. Zoox is just a block away. They circle around all day, every day. Based on my observations, they have a hard enough time navigating tricky intersections (2-way stops seem to really flummox AVs* from what I see) in perfect weather - I can't imagine how they could function in snow.

(To be fair, the intersection in question flummoxes humans even worse, there are multiple near-misses almost every day)

I wonder what truly was the reason for Tesla dropping radar. Musk spun it in typical Musk fashion (Plaid is so good, the Plaid+ is unnecessary) but the engineering team must have reached some sort of limitation with the original FSD hardware that could only be resolved (while maintaining "backwards compatibility") by going vision only. Maybe the radar and vision systems were disagreeing too often in urban environments.
 
Last edited:
I think the reason Tesla wanted to drop LIDAR is cost. The time it'll take to develop a working system is on the scale decades at this point. Tesla's a business. Adding LIDAR to each car for no reason is an unneeded cost. They can use this to market their special tech while also saving manufacturing costs while still selling each car at the same price
 
Eunos_Cosmo
Waymo has one of their test fleets just across the street from my office. Zoox is just a block away. They circle around all day, every day. Based on my observations, they have a hard enough time navigating tricky intersections (2-way stops seem to really flummox AVs* from what I see) in perfect weather - I can't imagine how they could function in snow.

(To be fair, the intersection in question flummoxes humans even worse, there are multiple near-misses almost every day)

I wonder what truly was the reason for Tesla dropping radar. Musk spun it in typical Musk fashion (Plaid is so good, the Plaid+ is unnecessary) but the engineering team must have reached some sort of limitation with the original FSD hardware that could only be resolved (while maintaining "backwards compatibility") by going vision only. Maybe the radar and vision systems were disagreeing too often in urban environments.
Click to expand...

ProjectWHaT
I think the reason Tesla wanted to drop LIDAR is cost. The time it'll take to develop a working system is on the scale decades at this point. Tesla's a business. Adding LIDAR to each car for no reason is an unneeded cost. They can use this to market their special tech while also saving manufacturing costs while still selling each car at the same price
Click to expand...
I'd like to clarify that Tesla was using lidar and not radar? Regardless, reliance on fewer systems, particularly when those systems are heavily computerized and/or not very reliable to begin with is universally a bad idea. This opinion is coming from my aviation experience, from a person flies everything GPS-direct but is still upset that the FAA shut down NDBs because they're so simple you could learn to use one with a few minutes of fiddling in the cockpit. Regardless, airplanes rely on a ton of layers of crap to figure out where they are and where everybody else is - GPS, ground-based ADS-B data, ground-based radar controllers, TCAS onboard traffic detection, constantly updated nav data, and ultimately the pilots. And precision landings require another layer of GPS, effectively, called WAAS. And yet another layer to that is called GBAS which isn't in use by the FAA yet.

Anyways the point is that autonomous cars relying on only cameras is no better than a driver relying on only eyeballs. That's not what I call autonomous. I am admittedly unfamiliar with what Waymo and whatnot are doing but I know they've got big giant sensors and whatnot on their roofs. I really hope they're trying to incorporate not only redundancy but also multiple different interlinked systems to map the environment and make decisions.

But my ultimate belief is that we should simply train drivers to not suck. That would solve most of our problems.
 
Quote from Elon Musk during their Monday earning's call

www.businessinsider.com

Tesla FSD is still a work in progress amid tech challenges, lawsuits

Tesla FSD, or Full Self-Driving, has been hit by lawsuits and investigations questioning its safety, including a fatal crash tied to the tech.
www.businessinsider.com www.businessinsider.com



EDIT:
Transcript of the earning's call via Jalopnik article

TSLA Stock Tesla Inc Q2 2021 Earnings Call

Speaker 1: Thank you for standing by. Welcome to the Tesla, Second Quarter, 2021, Financial Results and Q&A Webcast. At this time, all participants are in listen only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you...
www.rev.com

Yeah. It's like any given the price is going to be wrong. So we'll just adjust it over time, as we see if the value proposition makes sense to people. I'm not thinking about this a lot right now. We need to make full self-driving work in order for it to be a compelling value proposition.

Otherwise people are kind of... Depending on the future. Like right now, if it doesn't make sense for somebody to do FSD subscription, I think it's debatable. But once we have full self-driving widely deployed, then the value proposition will be clear. And at that point, I think basically everyone will go use it. Or a rare individual who doesn't.
Click to expand...
 
Last edited:
Well if they want to make it work they have to make the car know what the hell is happening around it and visual cameras are coming up short. Lidar does things.

It would also be nice if our road system was even remotely standardized but whatever.
 
Tesla should just concentrate on their motors and battery tech. They really are bad at building an actual car. It reminds me of modern GM, some excellent power trains hiding inside average at best vehicles. Sorry, it's just my take on Tesla.
 
LM7325
Tesla should just concentrate on their motors and battery tech. They really are bad at building an actual car. It reminds me of modern GM, some excellent power trains hiding inside average at best vehicles. Sorry, it's just my take on Tesla.
Click to expand...
They should ignore their weaknesses? That doesn't sound like a good idea. If anything, Tesla should spend a great deal more of their time and money on platform design and manufacturing, as unglamorous as that may be. In Tesla's defense, building cars is really, really hard...especially if you are starting from scratch. It took the Koreans several decades to get the quality of their cars on par (now I'd suggest they are above par) with the ROW. The biggest thing I think that would help Tesla is spending more time developing and properly detailing their designs - I suspect there is a lack of development between concept design and manufacturing design.
 
LM7325
Tesla should just concentrate on their motors and battery tech.
Click to expand...
Doesn't Panasonic supply the cells for the battery and build them into batteries in collaboration with Tesla?
 
Speaking of motors, I remember I saw this tweet from a while and forgot to share it



I had no idea Mercedes and Toyota were using Tesla powertrains at one point
 

Latest Posts

Back