Forums

So Tesla has missed out on this?

So Tesla has missed out on this?

Anthony J. Parisio | January 12, 2020

Your point?

TeslaTap.com | January 12, 2020

Missed out on a prototype that is not in production? Nothing to miss out on. Glad someone else is validating Tesla's direction though - all camera/radar solution. The prototype is a little ugly - not as bad as Waymo and most other prototypes running around.

Uses a third-party system has some advantages - potentially many customers and large volumes sold. On the downside, these often end up far more complex with multiple customers all requiring slightly different requirements. With contracts, negotiations, customizations, and validations, it can take 2-3 years to end up in a vehicle. So these solutions tend to be way out of date with technology.

With Tesla doing it all in-house, the turn-around is far quicker. Still, FSD is really complex and is going to take time for anyone attempting it to get it right.

TeslaTap.com | January 12, 2020

"Using a third...."

inconel | January 12, 2020

I am excited to see a camera-only system that can handle complex urban driving. It validates Tesla direction and means we might not be that far away from FSD.

dborn @nsw.au | January 12, 2020

@Anthony Parisio. My point is Mobileye ver 3 as installed on my Signature car with parking sensors (Australia) is pretty damn good. Tesla would be far ahead of where it is now if they had stayed with Mobileye, whose entire focus is on self driving with no competing agenda such as car/truck, solar roof, grid scale energy, development. There was a huge lag period while they were developing their own system which was far poorer for some years after the “divorce”. Mobileye has the advantage of working with a much larger car population and had Tesla stayed with them would have been far larger still and would have dramatically accelerated development and deployment, to answer Tesla Tap. If version 5 is anything like as stable as ver3 is, it will be awesome!! Does Teslas system recognize red lights yet? I remain convinced that my version 3 chip is capable of that but it will never be implemented because of the divorce!! Mores the pity, since i plan on keeping my car for a VERY long time!

Daisy the Road ... | January 12, 2020

Yes, the Tesla will recognise red lights. You'll need to purchase FSD to get that though.

redacted | January 12, 2020

I put some kind of mobileye on my '13 MS. Mobileye promptly stopped supporting it, it never had a software update, never even recompiled the iOS app so it stopped working with iOS. In the meantime I had constant software updates from Tesla. Mobileye as a company is unreliable and I'm glad Tesla dropped it while it could.

redacted | January 12, 2020

I'm not bitter, of course.

Pungoteague_Dave | January 14, 2020

@dborn "My point is Mobileye ver 3 as installed on my Signature car with parking sensors (Australia) is pretty damn good. Tesla would be far ahead of where it is now if they had stayed with Mobileye, whose entire focus is on self driving with no competing agenda such as car/truck, solar roof, grid scale energy, development. There was a huge lag period while they were developing their own system which was far poorer for some years after the “divorce”. "

I agree completely. I just traded a '15 P85D (AP1) for a new Raven X. The Raven X is far superior in many ways, including self-driving tech. However, it is far worse than the Mobileye system at handling headlight high beams and some other functionality that was built in to the former platform. My X is constantly blinding both oncoming drivers and leading drivers in their rear view mirrors. The car simply cannot make out tail lights or oncoming headlights until far later than the Mobileye system could, or a human can. This makes the function nearly worthless, as a responsible driver self-over-rides the system - which makes your vintage Tesla more advanced in this respect, and puts fail to the Mobileyte divorce. This is an example of Tesla having a full five years to catch up yet has not been able to develop the tech inhouse that it already had from a third party a long time ago. But what was the real alternative? Tough call.

"Mobileye has the advantage of working with a much larger car population and had Tesla stayed with them would have been far larger still and would have dramatically accelerated development and deployment, "

Perhaps, but the collaboration had become toxic. Israeli "invented here and you'll take what we have" intransigence and demanding customer was a bad mix.

"Does Teslas system recognize red lights yet?"

Yes, it does. The latest software download shows a lot of interesting things onscreen, although does not do anything with the data. It shows traffic signals, stop signs, bicycles, pedestrians fairly accurately. It knows more types of adjacent vehicles. It is also often incorrect, showing electrical boxes as trash cans, etc. But Tesla is doing some neat stuff here. However, as you point out, it has to defray the costs over a tiny increment of installed cars that have the required hardware, while a dedicated specialist supplier has far more resources and can spread the costs over multiples of the installed base. Like the law of gravity, the system with more users and a dedicated focus must eventually be better. Which leaves me torn - when there is no third party solution, or one that isn't available to buy anymore, one must improvise. Tesla has done that a lot, to good effect. I believe that Mobileye walked away, shocking Tesla, not the other way around - Tesla was doing things with AP that Mobileye thought legally irresponsible.

But looking at how Tesla operates, making things happen when everyone else says "can't be done", we see lots of examples. The nav/mapping system kluge was just that, with conflicts between Garmin/TomTom/google etc. but the fact was that we had a huge hi-rez solution in our cars when none other existed, bugs and all. Things get better all the time, but at least Tesla is willing to put itself out there with products that may not be as fully baked as if they came from a German marque, but at least we have them.

We only have so many years in life, and I've personally already had more than seven years as a Tesla owner, driving the future. Lots of rough edges in the process, but a hugely enhanced life for 10% of the allotted time here, and counting. The alternative would have been far less interesting or enjoyable in many ways. So we take what we can get. In this case, it was inhouse automotive vision, with all the attendant flaws. Will we ever see FSD on the cars that supposedly have the tech today? I doubt it. But so what? Just today I drove 200 miles in mist and sporadic rain. The car threw dozens of camera obscured messages and shut down all AP and even cruise control functions many times (after a red screen barf session without warning), citing individual or several times multiple camera blockages. It's not a great thing for a car to be driving in traffic and then suddenly have it turn on full regen braking and give up its control of steering and acceleration, with cars all around. Yes, I was ready, had no issue, but seems to me that this should not be a feature of any consumer product being put on public roads. There's no way to react instantaneously when it does one of these "I quit" tantrums without warning, so the car committed numerous brake checks in traffic. The cameras were fine, just wet or sometimes only misted damp. If camera vision is that easy to shut down, given Elon's insistence that cameras and radar is all that he needs to pull off FSD, these cars simply cannot be allowed out on streets alone. Yet that's Elon's claim. Not buying it, but still a happy customer.