Reminder: radar used to be the "future of Tesla's self-driving Autopilot." If you read the afterword to the paperback edition of Ludicrous: The Unvarnished Story of Tesla Motors you know that Tesla even spent years developing radar in-house inverse.com/article/20833-…
If radar is so useless, why do the early Model Xs have brackets for corner radar? ebay.com/p/700484339

The TMC guys even stumbled onto the fact that Model X was supposed to have corner radar: teslamotorsclub.com/tmc/posts/3223…

Even Elon knows sensor diversity/redundancy matters.
The radar story matters because it reveals that Musk has been fumbling for a strategy even after he started taking customer cash for FSD, he was naive enough to think Tesla could beat Bosch/Conti, and that the north star for this safety-critical tech is cost and not safety.
Add up all of these facts, and there's no way not to conclude that "Full Self-Driving" is a scam. The most you can say is that Musk took consumers cash for a product he wasn't close to understanding, which doesn't make it any less of a scam.
I should add here: there is still a ton left to report on Tesla's in-house radar program and my reporting days are over (at least for the moment). If you're a journalist who wants to report this story out properly, hit up my DMs and I will help however I can.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with E.W. Niedermeyer

E.W. Niedermeyer Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @Tweetermeyer

10 Apr
Critical as I am, I generally give Tesla credit for their powertrains and Superchargers... but this 739-page thread about loss of range and/or charging speed and/or increased "vampire drain" makes me wonder. Some (understandably) angry owners. teslamotorsclub.com/tmc/threads/su…
The guy who has done more to help owners understand these problems than anyone, @wk057, is now being pushed out of the community by TMC mods. I saw the owner-investor war for TMC's soul coming years ago, and unsurprisingly the investors have won.
Here's what the conflict comes down to: owners want an open exchange of information about the products they have spent a lot of money on, while investors want to suppress or spin any information they perceive as being negative for the company. This is a big issue for a forum!
Read 4 tweets
21 Feb
In 2015, I stumbled onto my first real Tesla story when I found they'd rather hook up Superchargers to diesel generators than make their battery swap station available. I asked their comms department about their emissions claims and the answer shocked me. bloomberg.com/opinion/articl…
Tesla keeps a running count of its "carbon impact" at tesla.com/carbonimpact. How, I asked, did they calculate this number? The answer: they simply assumed that every vehicle had zero direct or indirect emissions. But did they buy zero-emission power for all Superchargers? No.
Go back and look at the claims Tesla has made over the years and you'll find that it has always said/implied that Superchargers were zero-emission. They have even claimed, repeatedly, that they would be 100% solar, off-grid and "zombie apocalypse-proof."

dailykanban.com/2015/05/27/tes…
Read 7 tweets
21 Feb
These three paragraphs, about IBM's failure to deliver on Watson Health's soaring ambitions, hold several important lessons relevant to AI and AVs. Though very different, health care and driving are both promising but deeply challenging areas for AI. wsj.com/articles/ibms-…
First: lots of data doesn't solve every problem. Particularly when the costs of failure are high, as they are in health care and driving, achieving the necessary level of consistent accuracy is very difficult. As the level of accuracy/reliability rises, the challenge deepens.
Second: The "customization problem" that Ng refers to here is one reason that most serious AV developers are pursuing SAE Level 4 (geofenced) autonomy. Limiting the domain and tailoring systems to it is key to achieving and validating safety-critical levels of AI performance.
Read 6 tweets
4 Feb
Here's a fun mental exercise: if you did want to run a self-driving car scam starting in 2016, how would you have made it different from Tesla's "Full Self-Driving"?

It's genuinely difficult to think of a better way to pull that off than exactly what Tesla has done.
Start with the most important consideration: the marks. Do you target VCs? Sure, they are hype-susceptible and fallible but they also have access to experts and lawyers. No, you would target the public, who know nothing.

Incidentally, only one AV developer takes consumer cash.
What's the pitch? Well, everyone else is pitching Level 4 robotaxis because the tech is pricy, so not that. Sell the dream: SAE Level 5 autonomy, but in a car that you can afford to own yourself. Better yet, it can be a robotaxi that works for you and pays for itself!
Read 12 tweets
3 Feb
This starts in five minutes, so you just have time to register and tune in. Really excited to learn more about the semifinalists in @USDOT's Inclusive Design Challenge!
Kathy Klinich of @UMTRI is explaining their work on a universal docking interface geometry for wheelchairs and an automated seatbelt system, aimed at allowing wheelchair users to use an autonomous vehicle without any human assistance. Such a cool and important project! 🦾🦾🦾
AV developer @May_Mobility's @tara_a_lanigan says accessibility is a core value for the company, which is why they teamed up with @UMTRI to pilot implementation of their wheelchair anchoring system in their autonomous shuttles as part of the USDOT Inclusive Design Challenge.
Read 8 tweets
2 Feb
I needed some comedy today, and boy did this do the trick
Musk doesn't materially address any of the meaningful manufacturing questions/issues and Munro quickly stops even hinting at them and lobs softballs, but apparently this comes across as a deep, substantive and focused talk about manufacturing to lots of folks🤷‍♂️
"The real obstacle to Full Self-Driving is state-to-state variations in road lane markings" sure is a take.
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!