Hacker News Reader

Comment by CyanLite2 | Hacker News Reader

LIDAR cost is coming down, but Tesla has almost no margins on cars. So adding a $500 part per car for LIDAR is out of the question. Even the cheap radar parts were removed from production going forward.

However, it's good to see other automakers like Volvo going the LIDAR route on their 2022 electric models and are offering Level3 autonomous driving.

View on HN

They currently charge $10k for FSD. Even if they don't drop any of their cameras and processing equipment, $500 LIDAR is a 5% increase. Is it just me or are we politely ignoring Musk's reluctance to admit his bold proclamations against LIDAR are the biggest factor in their design decisions.

Andrej Karpathy already explained they were getting too much noise from Radar. Adding LIDAR would only complicated it. Unless you have better knowledge than Andrej Karpathy, I’ll go with his word on this.

Edit: nitrogen I can’t comment directly on yours, so I’ll ass it here. That’s a good goal to have and they tried that for well over two years. They found too much noise and confusion. If Radar says one thing and Vision says another over and over who do you trust? They decided they could train their NN with just Vision. They use LIDAR on test cars to confirm their Vision can recognize distances correctly. In this case I believe cutting out a sensor instead of trying to neuter it is the right decision.

Isn't the basic goal of sensory fusion to be able to use each sensor to the extent that it's good, and ignore it to the extent that it's bad?

Karpathy is the guy responsible for the results we see in these videos. Why would you trust his word about any of this?

Andrej works for Tesla, so yes, I don't believe him.

If he contradicted his boss in public, Musk would just fire him and hire someone who will repeat the party line. I won't believe anything he says about lidar until he's no longer at Tesla, or until they have a car that's not trying to ram concrete pillars.

Andrej has said some stuff in public that certainly implies he's prioritizing the 'party line' on stuff like HD maps and lidar than his academic stats knowledge.

> Adding LIDAR would only complicated it.

More sensors is better. They just have not yet learned how to merge data. Example is drone autopilot, that have IMU and GPS and uses Kalman filtering.

> If Radar says one thing and Vision says another over and over who do you trust?

Neither? If you don't have a tiebreaker you disengage and let the human decide. We solved this problem in systems long ago. You either fail open or fail closed depending on the system. In this case it seems like you'd always fail to the safest option which is alert the human.

So basically you're driving full speed against an obstacle and you get the commands handed out. It sounds like a slapstick comedy where the pilot says "lol bye you're on your own" and jumps out the plane door.

I think that is still desirable behavior when compared to a slapstick comedy where the pilot says "don't worry bruh I got this" and then proceeds to plow directly into a mountain side

I get your point and I agree to it. For the nitpicking, I'd just not use the word "desirable" when the slapstick situation would be rather "less bad"... because "desirable" means a positive outcome while neither one is anything positive.

I learned on HN long ago that planes usually have three sensors for critical systems so that if one is disagreeing but the other two agree, you still have a reasonable degree of certainty. Having a single sensor that seems to insist on steering toward obstacles doesn't seem great to me.

Tesla’s margin on the Model 3 is 35%:

https://thenextavenue.com/2020/06/10/teslas-model-3-profit-m...

It's not that high once you strip out the bitcoin gains and regulatory credits. https://archive.is/WAvau

Actually BTC gains don’t count, only losses do. They will take a minor hit on their earnings coming up but will still made record revenue and profit, so like Method-X claimed - Tesla has good margins.

Tesla's Full Self Driving package was between 8 and 10,000 dollars per vehicle. That should cover it.

The hardware suite is the same on every car nowadays. They want FSD to be a downloadable app like a subscription. Which means you would have to put LIDAR on every car even if it didn't convert to FSD.

Also, Elon has gone on Twitter rants several times before against LIDAR. That horse has already left the stable and it ain't coming back.

>They want FSD to be a downloadable app like a subscription.

Well that's just too bad.

While I'm sure it would be inconvenient to their business plan, it remains the case that they're charging ten grand for that software patch, and have no possible argument relating to margin.

This seems backwards; it's a premium feature, the sort everyone else is adding to increase margins.

Teslas' margins are several times larger than other automakers ... but the company valuation isn't particularly justified even at the current high margins.

Reply on news.ycombinator.com