I don't believe there is a Type II category. The new sensors are going to require a new wiring harness and history suggests that this will not be a reasonably-priced upgrade. The problem with identifying Type III items is that there are many things that it will simply do better and more reliably with improved sensor suites. Assuming they stick with Mobile Eye solution, the next generation will likely offer 3 front-facing cameras which will improve steering and reactions to pretty much everything it is doing already -- that is really hard to quantify. It probably means reliable steering without lane markings, better reactions to lane incursions from traffic on side roads, more accurate lane placement for edge cases where the current suite struggles. On the other hand, it seems quite obvious that some of this will improve with the current hardware. Detection of vehicles approaching at high speed from behind you is clearly a Type III case as is probably improved detection of cars surrounding you (probably improving the range of the ultrasonic sensor array).
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
-
On today's TMC Podcast, we ask the question "What is the Tesla Cybercab?". Join us on YouTube live at 1PM and participate in the chat.
I don't believe there is a Type II category. The new sensors are going to require a new wiring harness and history suggests that this will not be a reasonably-priced upgrade. The problem with identifying Type III items is that there are many things that it will simply do better and more reliably with improved sensor suites. Assuming they stick with Mobile Eye solution, the next generation will likely offer 3 front-facing cameras which will improve steering and reactions to pretty much everything it is doing already -- that is really hard to quantify. It probably means reliable steering without lane markings, better reactions to lane incursions from traffic on side roads, more accurate lane placement for edge cases where the current suite struggles. On the other hand, it seems quite obvious that some of this will improve with the current hardware. Detection of vehicles approaching at high speed from behind you is clearly a Type III case as is probably improved detection of cars surrounding you (probably improving the range of the ultrasonic sensor array).
A couple of points on this. One is that I don't think the rear camera is part of the AP sensor suite, secondly probably more people park front-in than back-in so it's probably better to start with what's most popular. Finally there are stall situations where you almost have to park front first like this.
Attachments
SabrToothSqrl
Active Member
I'm also considering trading in my CPO 85 for a CPO or Inv. 85D.
Let me know the price difference you encounter, if you would.
thanks
Let me know the price difference you encounter, if you would.
thanks
It's going to be expensive.
My S85 was bought at $84,070 (after federal $7500); and they are offering trade in at $61,500. So $22k depreciation right there, for the first year. I was originally planning for $10k/yr depreciation when I bought it, so this trade is quite painful.
The inventory 85D is $93,550 after a $5,900 discount (i.e. sticker price is $99,450.) It qualifies for the federal $7500 rebate. In Washington state, I pay 9.5% sales tax on the difference in price, so my net cash outlay would be:
(93550 - 61500) * 1.095 - 7500 = $27,594.75
In return, I'm getting:
- $7,880 increase in sticker price (mostly due to bells and whistles like leather seats, premium lighting, etc.)
- AWD and Autopilot
- 9000 less miles on the odometer
I'm not sure how much everyone else values AWD and Autopilot - If I value them at $10k total, I would be paying approximately $1/mile for the 9000 less miles. Neither great nor terrible, it's just the way depreciation works :/
CHG-ON
Still in love after all these miles
Well, first off, if I can't park my car, I probably shouldn't be driving it. It is big. But I have adapted. It certainly did take some time to adjust. I used the auto parallel parking once and it worked great. Totally cool! I have no problems parking straight on: forward or reverse. Or parallel parking. Maybe I am lucky with that.
As for the rest of autosteer, I am very conservative and will not trust it until it can prove it's meddle. But I have found, through my testing, that it is actually getting better every day. I have been testing on back country roads (I know...not designed for that), gently nudging the wheel just enough to not cancel AS, and, much to my surprise, it is actually learning. It really is amazing. When I want it more centered in the lane, I nudge it for one, perhaps for two days, and then it seems to understand that I want it "there" in the lane and remembers it. I have taken it very, very slowly. It does seem to get better as the days pass. I will continue to keep my hands on the wheel and coax it. But I do see small incremental improvements every day. Very cool.
As for the rest of autosteer, I am very conservative and will not trust it until it can prove it's meddle. But I have found, through my testing, that it is actually getting better every day. I have been testing on back country roads (I know...not designed for that), gently nudging the wheel just enough to not cancel AS, and, much to my surprise, it is actually learning. It really is amazing. When I want it more centered in the lane, I nudge it for one, perhaps for two days, and then it seems to understand that I want it "there" in the lane and remembers it. I have taken it very, very slowly. It does seem to get better as the days pass. I will continue to keep my hands on the wheel and coax it. But I do see small incremental improvements every day. Very cool.
Soolim
Member
I don't see why the current hardware can't reliably drive me from San Francisco to Seattle while I watch movies or read a book (as in, awake, but not paying attention). It's not like the interstate lanes change. All that is required is maps of all the lanes. It shouldn't even need the camera for lane markings. If there is construction, google should know, and the car should alert me. Even if I have to manually pull off and supercharge, that's okay. The current cars can control the steering, and has enough vision around the car to see what's going on.
I don't care if the car can't drive me home if I am drunk. That's what Uber is for. I just want on ramp to off ramp on road trips, with the ability for the car to alert me when something weird is coming up.
I don't care if the car can't drive me home if I am drunk. That's what Uber is for. I just want on ramp to off ramp on road trips, with the ability for the car to alert me when something weird is coming up.
I don't think it's that simple, you're missing a lot of "what is scenarios" between having autopilot and semi-autonomous driving, which semi-autonomous driving should be able to handle while you read your book, just a few random ones out of the top of my head
1. The car next to you is drunk, tries to swipe into your lane. The side collision avoidance can only do so much without changing lanes.
2. The car behind you tries to rear-end you
3. There is unexpected construction, SURPRISE! google doesn't know yet, you're the first to show up there!
4. There is road debris that will puncture your tire
5. There is an emergency vehicle pulled over on the shoulder, the law (in many states) is to slow down or change lanes. But wait, the Tesla can't change lanes, and I'm not sure it can detect emergency vehicles to slow down for
6. etc. etc. etc.
mikeash
Active Member
I think it's instructive to imagine yourself as the autopilot's "brain," driving with the AP sensors, and imagine what you would be able to do.
This is the scenario: you're in command of the car as usual, with full access to steering, brakes, and other controls. For vision, you have a single camera that faces forward and can't be turned. Thus: you can't see to the sides, or behind you. (Maybe you can see behind you if the rear camera is connected to the system, but then you have a fisheye lens with almost no distance vision in that direction.) You also have a radar that lets you "feel" cars ahead of you in a cone out to a few hundred feet. And you have ultrasonic sensors that let you "feel" obstacles to the sides. You have GPS and high-quality maps.
Now imagine driving the car like this. How would you do. What situations could you handle, and in what situations would you require assistance?
You can see how this is enough the current autopilot functionality. If you were placed in this situation and asked to stay in a lane, match the speed of the car ahead, change lanes on command, dodge side collisions, or parallel park, you could do it.
How about future functionality? Imagine the "summon" feature that people keep talking about. You could do that! You'd have to take it slow and be careful, because you can't spot things coming in from the sides very well, turns require extreme care because your vision only turns when the car does, etc. You could do it, but at low speeds. Which is exactly what "summon" is all about anyway.
How about automatic city driving? Some of this could work. You could recognize stoplights and stop for them automatically. You could stop for stop signs, but you probably couldn't then proceed through them safely, because you can't see cross traffic well. You can't make sharp turns very well because, again, your vision turns with the car. Jaywalkers could be a serious challenge, though, with limited side vision.
The software isn't human-equivalent, but from this thought experiment it looks like the larger limitation is the sensors. Whatever a human could do with these sensors, the software will probably be able to do reasonably soon.
This is the scenario: you're in command of the car as usual, with full access to steering, brakes, and other controls. For vision, you have a single camera that faces forward and can't be turned. Thus: you can't see to the sides, or behind you. (Maybe you can see behind you if the rear camera is connected to the system, but then you have a fisheye lens with almost no distance vision in that direction.) You also have a radar that lets you "feel" cars ahead of you in a cone out to a few hundred feet. And you have ultrasonic sensors that let you "feel" obstacles to the sides. You have GPS and high-quality maps.
Now imagine driving the car like this. How would you do. What situations could you handle, and in what situations would you require assistance?
You can see how this is enough the current autopilot functionality. If you were placed in this situation and asked to stay in a lane, match the speed of the car ahead, change lanes on command, dodge side collisions, or parallel park, you could do it.
How about future functionality? Imagine the "summon" feature that people keep talking about. You could do that! You'd have to take it slow and be careful, because you can't spot things coming in from the sides very well, turns require extreme care because your vision only turns when the car does, etc. You could do it, but at low speeds. Which is exactly what "summon" is all about anyway.
How about automatic city driving? Some of this could work. You could recognize stoplights and stop for them automatically. You could stop for stop signs, but you probably couldn't then proceed through them safely, because you can't see cross traffic well. You can't make sharp turns very well because, again, your vision turns with the car. Jaywalkers could be a serious challenge, though, with limited side vision.
The software isn't human-equivalent, but from this thought experiment it looks like the larger limitation is the sensors. Whatever a human could do with these sensors, the software will probably be able to do reasonably soon.
SabrToothSqrl
Active Member
Thank you for the details, bottom Line: $28,000 for AWD and AP.
Yea... I can't personally swing that. if I'm going to drop that much $, it's going to be for all that wrapped in an X.
I paid $75,100 for my CPO S in Feb of 2015. Trade in is $48,800. Yea... so... not going to happen. I'll recheck my options after more info/test driving an X.
Thank you very much. This thought experiment does clear my mind quite a bit. So - biggest issue is lack of side vision.
I guess I could live with the current gen hardware for a couple of years - it seems that reliable side-vision-assisted autopilot/autonomy is a few years away at least. In the meanwhile I'll just apply my biological peripheral vision :tongue:
