Poll: Will HW2 Autopilot be capable of full highway driving with no driver input by the end of 2018?
- Thread starter strangecosmos
- Start date
Tam
Well-Known Member
How would you summarize it in 10 words or less? It’s hard to find a short-hand for the functionality Elon seems to be describing: Enhanced Autopilot as it currently exists, but it is capable of handling divided highway driving from on-ramp to exit most of the time, requiring no steering, acceleration, or braking from the driver — although monitoring and occasional intervention is still required. What short-hand do we use for that? Maybe “fully highway capable Level 2 autonomy”?
This is a needlessly rude comment. People were not clear on the exact functionality Elon is promising, so I clarified by quoting his exact description of the functionality. I am just providing clarification. I haven’t stated any opinion on the matter. Besides, people should be able to disagree without insults and name-calling. FYI, I’m just going to add you to my ignore list if you aren’t respectful.
Other comments made by Elon on Twitter and by both Elon and Tesla VP Stuart Bowers on the Q2 earnings call indicate that Elon/Tesla is claiming Autopilot v9 will have the functionality described on the Joe Rogan podcast. The only thing that is still unclear to me is whether this is planned for Autopilot v9.0 or some later v9.x release.
Autopilot v9.0 is “hopefully” supposed to come out towards the end of September, according to Elon. The early access release of Autopilot already seems to be a bit delayed from what Elon initially said, and in the past Elon has been notoriously off with his timelines for Autopilot functionality. “The end of 2018” gives Elon 3 months to be late with Autopilot v9.0. Which, of course, may not be enough. He has been much more late with Autopilot stuff in the past.
The poll essentially breaks down into three questions:
1. Will Autopilot v9.0 be released by the end of 2018?
2. Will Autopilot v9.0 have the functionality described on the Joe Rogan podcast?
3. If not, will some later Autopilot v9.x release that does have that functionality be released by the end of 2018?
I am not sure if Elon/Tesla is even claiming that v9.0 will have the on-ramp to exit functionality, but Elon’s comments on the Q2 earnings call, at the shareholders’ meeting in June, and on the Joe Rogan podcast suggest to me it is either planned for the v9.0 release or a v9.x release planned to come a few months after.
Last edited:
HankLloydRight
No Roads
<opinion>
What Elon fantasizes about and talks about just is not reality (IMHO).
Yes, EAP or FSD (I don't know which) will be able to go on-ramp to off-ramp and change lanes itself, etc. But only when every other driver on the road acts in an intelligent, predictable, and correct manner. But unless every other car on the road is AP equipped, we know that's just not true and will never be true.
</opinion>
Just this summer I was driving on the highway at about 85mph on AP1. There was a black wooden 6' ladder in the middle of my lane. I didn't see it until the truck in front of me passed over it. It almost blended into the pavement. The radar didn't sense it. I had to take evasive maneuvers to avoid hitting the ladder. Luckily there wasn't anyone to the left of me so I could steer around it. But I just don't see AP1 or AP2 being able to handle every-day edge cases like this. If AP plowed into that ladder, it would have caused significant damage and most likely loss of control and a bad accident.
Later that same day, there was a large plastic trashcan lid in the middle off the road, I'd guess rising up 6"-7" off the road surface.. AP1 radar DID pick that up and alert me to what it thought was a car I was about to hit. At least it detected it this time, but still, I don't know if AP (any version) is going to be able to make very fast, very aggressive avoidance maneuvers like that.
I'd sure like to see something like that demoed and proven wrong.
What Elon fantasizes about and talks about just is not reality (IMHO).
Yes, EAP or FSD (I don't know which) will be able to go on-ramp to off-ramp and change lanes itself, etc. But only when every other driver on the road acts in an intelligent, predictable, and correct manner. But unless every other car on the road is AP equipped, we know that's just not true and will never be true.
</opinion>
Just this summer I was driving on the highway at about 85mph on AP1. There was a black wooden 6' ladder in the middle of my lane. I didn't see it until the truck in front of me passed over it. It almost blended into the pavement. The radar didn't sense it. I had to take evasive maneuvers to avoid hitting the ladder. Luckily there wasn't anyone to the left of me so I could steer around it. But I just don't see AP1 or AP2 being able to handle every-day edge cases like this. If AP plowed into that ladder, it would have caused significant damage and most likely loss of control and a bad accident.
Later that same day, there was a large plastic trashcan lid in the middle off the road, I'd guess rising up 6"-7" off the road surface.. AP1 radar DID pick that up and alert me to what it thought was a car I was about to hit. At least it detected it this time, but still, I don't know if AP (any version) is going to be able to make very fast, very aggressive avoidance maneuvers like that.
I'd sure like to see something like that demoed and proven wrong.
boonedocks
Active Member
I am sorry, I wasn’t meaning to sound rude. You are just siting web links / clips / tweets as if they have any validity. In the realm of Elon Speak these things are to be taken as completely anecdotal and with trainload of salt. You keep re-posting and siting instanced and those of us that have been around circa pre-2015 as Tesla owners know better. We are just trying to keep the facts separate from the smoke.
Dr. J
Active Member
Dr. J
Active Member
I believe 100% of us are in the "don't know" category, but 90%+ of us (including me) are deservedly jaded/skeptical enough to disbelieve any time frame Elon gives, and so are a "no." In reality, those who know don't say, and those who say don't know. Including Mr. Musk.
I think we need to separate two questions:
1. Will the software and hardware have the capability of doing something?
2. Will such capability be released to customers?
The first question is strictly technical. I suspect Tesla already has a version that does most of those things that Musk was talking about and it may very well be finalized before the end of the year.
The second question is a business decision, that depends on many things not just the technical capability. How reliable the software is? How many mistakes does it make? Is it safe for an average user? What are the liability risks? Are there any regulations that may prevent the use of those features? And so on. I am skeptical about it being fully released this year. Perhaps some of the features, but not all.
1. Will the software and hardware have the capability of doing something?
2. Will such capability be released to customers?
The first question is strictly technical. I suspect Tesla already has a version that does most of those things that Musk was talking about and it may very well be finalized before the end of the year.
The second question is a business decision, that depends on many things not just the technical capability. How reliable the software is? How many mistakes does it make? Is it safe for an average user? What are the liability risks? Are there any regulations that may prevent the use of those features? And so on. I am skeptical about it being fully released this year. Perhaps some of the features, but not all.
Last edited:
Flawed logic. That doesn't mean that nothing he says should ever be believed. Especially given that everything he said on the Joe Rogan podcast has been backed up by descriptions of recent v9 development builds.
I think they're close. The Nav understands what lane you need to be in for interchanges, and the car does a great job of holding lanes and changing lanes under turn signal direction and reacting to other cars. I think it will be fairly simple to connect the dots and get the car shifting to the lane it needs on its own.
The joker in the deck is road debris, deer, potholes, and construction zones. It can hand back control or drop back to a driver monitored AP mode for construction zones. The big question will be whether they can sort out the code needed to see and avoid the debris/potholes/deer, or see it in time to hand over with enough time for the driver to understand the situation and react correctly coming into it cold.
It'll be a while longer before it understands the best lane to be in during a traffic jam, I imagine...
No, @strangecosmos was just posting an accurate transcription (and a link to the video) of what Elon said on the Joe Rogan podcast. There is nothing gullible or misleading about this.
He’s wrong more often than he’s right when it comes to timelines, and his behavior of late on social media is extremely erratic. He doesn’t get the benefit of the doubt from me at this point until he’s consistently following through on claims and promises.
I don’t think he means to come across as an over-promise master, but if someone says he’s gonna do 99 things and delivers 2 of them on time, I’m not going to have too much faith in future promises.
I agree with you in most cases. He's pretty much always off-base, until he's not. And I honestly think that his description of v9 sounds pretty close to what they're planning to release: Tesla Version 9 update: getting closer to release, now fixes one of the biggest issues with Model 3 UI, and more
Until it’s considered amongst the sea of other things Elon has claimed would happen that haven’t yet.
Maybe he’s turning over a new leaf, and will become super reliable and accurate with this sort of stuff. Here’s hoping.
Again, it's important to separate 1) the reliability of statements from Elon (or anyone else) from 2) whether it is 'gullible/misleading' to quote them for discussion in a thread specifically about the topic at hand.
Chargerfan
Member
I wasn't an early adopter of the EAP when I received my MX in June. I didn't see this feature as more than what it was stated to be... beta test performance .
When the feature was recently offered as a 2 week free trial I put it through my nonscientific but consistent 60mi daily commute and weekend errand paces.
It's not reliable enough at this time for safe autonomous city traffic. Stop and go straightaway travel using the speed pacing feature (the auto steer feature with frequent nags, off) was great but not essential for the requested cost . Not sure how many in my purchase situation opted for the "discounted" $5500 OTA download .
However , I will look to add when it is out of beta phase when I feel the ROI is more favorable with improved programing.
Wasn't version 9 supposed to address the doubts when processing speed was to be sigificantly faster?
Exactly. The question is too vague.
Will it be able also at highway speeds be able to reliably detect and stop for stopped vehicles like fire trucks? How about splits w/gore points like what killed Walter Huang in his Model X (highway 85 and 101 in Mountain View)?
I answered no.
Similar threads
- Poll
