Autonomous Car Progress

[Daniel in SD]

So this is the first time MobileEye has access to the "entire" stack, whereas Tesla has had it for years, on multiple continents?

So if this Mobileye system is better than FSD will it mean that they accomplished more in less time?
Or will we say that they've been working on this since 2014, way longer than Tesla?
It's crazy how much spin there is on this forum.

 

[diplomat33]

So this is the first time MobileEye has access to the "entire" stack, whereas Tesla has had it for years, on multiple continents?

No. Mobileye has had access to the entire stack for years. This is just the first time that Mobileye is deploying the entire solution as one plug-and-play package to automakers. In the past, AFAIK, Mobileye would provide the EyeQ chip but let the automaker do their own software. Now, Mobileye is providing the entire ADAS solution to the automaker.

 

[zone121]

So I watched the annual meeting this week and Elon spent about two minutes talking about autopilot and autonomous driving. He said there would be a huge software update in a month or something but quickly passed over it as if he was trying to avoid the subject. So, should I not expect my Tesla to be autonomous by the end of the year? I think he said last year that it would be basically self driving by 2q2020 but of course we are well past that point.

 

[diplomat33]

So I watched the annual meeting this week and Elon spent about two minutes talking about autopilot and autonomous driving. He said there would be a huge software update in a month or something but quickly passed over it as if he was trying to avoid the subject. So, should I not expect my Tesla to be autonomous by the end of the year? I think he said last year that it would be basically self driving by 2q2020 but of course we are well past that point.

No, do not expect your car to be autonomous by the end of this year (IMO).

 

[zone121]

No, do not expect your car to be autonomous by the end of this year (IMO).

You think it will ever be autonomous? Or is this all marketing hype? I just bought the car on the hopes of it someday being somewhat if not fully autonomous.

 

[diplomat33]

You think it will ever be autonomous? Or is this all marketing hype? I just bought the car on the hopes of it someday being somewhat if not fully autonomous.

IMO, it is a lot of marketing hype. Tesla has a long ways to go before our cars are truly fully autonomous (ie can drive anywhere like there is no person in the car). I also don't believe the hardware is good enough for true full autonomy. But they might get close to fully autonomous. I am sure our cars will probably be able to drive some routes with little to no driver interventions at some point in the future. Autopilot is already very useful in some cases and it will get better. So I think Autopilot will still be very helpful even if it never achieves true full autonomy.

 

[daniel]

I believe Tesla's cars will some day be truly and fully driverless Level 5. I do not believe that the cars being produced by Tesla today with the present hardware will ever be able to drive on public roads without a qualified driver in the driver's seat. And I do not know if Tesla will sell a truly driverless L5 car in what remains of my lifetime.

 

[NHK X]

Looks great. Thanks for sharing.

I like that Mobileye is delivering the whole ADAS solution.

We knew it was only a matter of time. Looks like Mobileye is ready to start deploying their camera vision part, along with their driving policy to deliver robust ADAS.

The clock is ticking for Tesla.

Clock is ticking yes, but bring on the competition. Good for me as the end user.

 

[DanCar]

You think it will ever be autonomous?

It is already level 2 autonomous. When will it be level 3? I'm hoping in 3 years. Level 4 where we can sleep? Only if you want to take a few minute nap. Drive all night while you are asleep, I'll estimate 2030. Level 5, not happening this decade.

Or is this all marketing hype?

A lot of wishful thinking with people who have a bias and a history of success, resources, and power.

I just bought the car on the hopes of it someday being somewhat if not fully autonomous.

Pssst, when people aren't looking, a lot of people use it that way. "somewhat" autonomous. I suspect the number of people with cheat devices is higher than people would estimate.

 

[Bladerskb]

 

[banned-66611]

Late 2011

Either they are emulating Elon Musk's "optimistic" predictions of they meant 2021.

 

[J1mbo]

No. Mobileye has had access to the entire stack for years. This is just the first time that Mobileye is deploying the entire solution as one plug-and-play package to automakers. In the past, AFAIK, Mobileye would provide the EyeQ chip but let the automaker do their own software. Now, Mobileye is providing the entire ADAS solution to the automaker.

Maybe you mean for EyeQ5?

I wonder if they have kept the requirement for the OEM to provide them with always-on connectivity.

 

[tmp131]

Looks great. Thanks for sharing.

I like that Mobileye is delivering the whole ADAS solution.

We knew it was only a matter of time. Looks like Mobileye is ready to start deploying their camera vision part, along with their driving policy to deliver robust ADAS.

The clock is ticking for Tesla.

From what I read so far, it seems like Mobileye is ahead of Tesla. After all it was Mobileye that initially supplied Tesla with the camera and software until they split. Mobileye just didn't sit and do nothing. There is a reason Intel bought Mobileye for $15 billion. I am not surprised that they might be ahead of Tesla. Maybe this autopilot rewrite will change things, but I am not so sure.

 

[DanCar]

I suspect Mobileye is always going to be ahead in the lab, i.e. demo videos, but always behind in U.S. implementation.

 

[diplomat33]

I suspect Mobileye is always going to be ahead in the lab, i.e. demo videos, but always behind in U.S. implementation.

That could change if Mobileye can find automakers to partner with. They 've already made a deal with Geely and will be deploying their ADAS in hundreds of thousands of Chinese EVs. If they do the same in the US, then Mobileye will not be behind Tesla in implementation either.

 

[daniel]

I suspect Mobileye is always going to be ahead in the lab, i.e. demo videos, but always behind in U.S. implementation.

It's hard to compare because of different business models: Tesla builds cars and constantly provides its latest improvements and features to customers who choose to pay for them. (My EAP used to be so slow to begin moving when stopped behind a car and that car started to move that I had to touch the pedal to get it started. I recently noticed that it now starts moving much sooner, and I no longer have to touch the pedal.)

Mobileye does not build cars. It's developing a system to sell to carmakers. So, basically, Tesla will always be ahead in implementation until a carmaker buys the Mobileye tech and starts selling cars with it. Then suddenly that carmaker will have arrived and Tesla will still be trying to figure out roundabouts and unprotected left-hand turns without 360° cameras.

 

[banned-66611]

Mobileye only releases stuff when it's ready, Tesla throws out alpha software for customers to test.

 

[diplomat33]

Mobileye only releases stuff when it's ready, Tesla throws out alpha software for customers to test.

Yep. Which is why I trust that Mobileye's SuperVision system will be great when it is released.

 

[DanCar]

Mobileye only releases stuff when it's ready, Tesla throws out alpha software for customers to test.

The problem with self-driving cars is that when it's ready is very subjective. There's an infinitely long tail of possible issues. If you're waiting to solve all those issues nothing will be ready.

 

[diplomat33]

The problem with self-driving cars is that when it's ready is very subjective. There's an infinitely long tail of possible issues. If you're waiting to solve all those issues nothing will be ready.

They can be ready in the sense of being able to perform a specific use in a specific ODD at the desired safety level. For example, you can have a highway driving system that is designed to drive hands-free on certain highways, in certain conditions, with the driver keeping their eyes on the road at a safety level X that is deemed good enough. So no, it can't do every single driving case but it is not designed to do every single case. It only needs to do what it is designed to do. If it can, then it is ready to be deployed in that limited role.