Got first hand experience with AP1...YES PLEASE!

[chillaban]

Everytime when my AP1 Tesla drives me down the highway, I simply cannot think how better can it get beyond this. It is that good..

Well that's just it. Every time around California I have it drive me around even a slightly hilly + curvy highway, I wish it could actually handle those conditions. On level roads, AP1 does fantastically well, even in city-ish conditions with poor lane line consistency.

However, the moment you add a bit of terrain elevation to the equation, all bets are off. Taking a curve + variable hill seems to consistently throw off AP1 to the point that I'm not willing to let it drive.

AP1 performs great, sure (arguably the best L2 system on the market), but it still leaves me wanting more.

 

[davidc18]

closing in on 50,000 combined miles on AP 1.0. AP 2 has nothing to offer until it can take over driving on local roads including street lights and stop signs. Hope AP 2.0 gets its act together before it is time to trade in our 2015 70D.

 

[Electroman]

However, the moment you add a bit of terrain elevation to the equation, all bets are off. Taking a curve + variable hill seems to consistently throw off AP1 to the point that I'm not willing to let it drive.

I agree. But then I simply cannot fathom how AP1 or even AP2 can drive if it can't see the road ahead on a hill or a curve. In fact I believe even if you throw in another 10 cameras to AP2, it will make no difference. On a hill the car gets blind and it doesn't matter how many cameras and sensors you have. Its like having a 20-20 vision means nothing if it is pitch dark.

 

[chillaban]

I agree. But then I simply cannot fathom how AP1 or even AP2 can drive if it can't see the road ahead on a hill or a curve. In fact I believe even if you throw in another 10 cameras to AP2, it will make no difference. On a hill the car gets blind and it doesn't matter how many cameras and sensors you have. Its like having a 20-20 vision means nothing if it is pitch dark.

I think the key is latency from first sight to driving decision. The MobileEye system seems to have considerable lag both in the UI and the underlying model. You can really feel it when aggressively switching lanes and getting false FCW alerts for the car that was in your previous lane.

We can drive with split second visibility and we don't possess X-ray vision either.

 

[Nosken]

I agree. But then I simply cannot fathom how AP1 or even AP2 can drive if it can't see the road ahead on a hill or a curve. In fact I believe even if you throw in another 10 cameras to AP2, it will make no difference. On a hill the car gets blind and it doesn't matter how many cameras and sensors you have. Its like having a 20-20 vision means nothing if it is pitch dark.

This is where integration with all the map data that they are gathering will help complement what is "seen" by the sensors. "Soon". :- )

 

[David29]

Having an AP1 car, I do not currently envy the AP2 owners who are patiently (or impatiently) waiting for parity in the functionality. It must indeed be annoying not to have even features such as the automatic wipers and headlight dimming, let alone TACC and autopilot.

But someday that parity will come and eventually the full AP2 features will emerge. And at that point, what I worry about is that all the development effort will be focused on AP2 cars and the AP1 platform will be left out, even when it could be improved.

Of course, there is also the risk that the Tesla team is not up to the task presented by AP2 and the development could be unduly delayed, or possibly fail to achieve all the advertised goals, etc. With any luck, that won't happen...

 

[calisnow]

I agree. But then I simply cannot fathom how AP1 or even AP2 can drive if it can't see the road ahead on a hill or a curve. In fact I believe even if you throw in another 10 cameras to AP2, it will make no difference. On a hill the car gets blind and it doesn't matter how many cameras and sensors you have. Its like having a 20-20 vision means nothing if it is pitch dark.

I don't follow. How do humans - with two eyes - manage drive on hills and curves?

 

[malcolm]

....and it's not like radar can see in the dark or through fog, dust or even smoke.

Oh wait.

 

[Electroman]

I don't follow. How do humans - with two eyes - manage drive on hills and curves?

Good question. Imagine a hill followed by an immediate curve.

I was thinking and trying to mimic exactly what would I do - if lanes is the ONLY input given to me. Then I realized that humans intuitively take cues from the sorroundings and make judgment and a determination that there is a curve ahead. If you only have lane markers and nothing else to process - like in a pitch dark night and freshly laid dark asphalt and poor headlights that all you can see are the lanes and nothing else - I betcha we will do just as bad as AP1 in that scenario

 

[chillaban]

Good question. Imagine a hill followed by an immediate curve.

I was thinking and trying to mimic exactly what would I do - if lanes is the ONLY input given to me. Then I realized that humans intuitively take cues from the sorroundings and make judgment and a determination that there is a curve ahead. If you only have lane markers and nothing else to process - like in a pitch dark night and freshly laid dark asphalt and poor headlights that all you can see are the lanes and nothing else - I betcha we will do just as bad as AP1 in that scenario

Yeah, the only thing we do a lot better than AP1 is that we understand that guard rails and ditches and other cars / oncoming car lanes are reeeeallly bad to be near. In a pinch, we'd gladly depart lane lines to avoid hitting / being in any of those things.

AP1 is extremely literal about lane lines and centering between a pair of lane lines. I've said this before, but I have a feeling that EAP and especially FSD need a much better understanding of where it's okay vs not okay to drive, that isn't strictly based off lane lines.

 

[dhanson865]

#ITOLDYOUSO Tesla is dumping Mobileye???

I was 100% sure that dumping Mobileye was a mistake. Tesla's CEOs mouth is on loose motion, while his engineering team's ability to deliver is constipated.

Paired with their focus on Model 3, and Model S/X quality have gone way down. I kinda feel sorry for the folks who got suckered into paying full price for AP2 honestly.

Keep in mind Tesla didn't dump Mobileye, Mobileye dumped Tesla.

Elons plan was to use Mobileye hw in AP2 along with Tesla Vision hw. He had no plans to stop buying mobileye chips as soon as they stopped selling them to him.

So you crowing about how Tesla shouldn't have done what it didn't do is kind of off.