As a sotware developer, I would expect “a huge step up” to go to v13.0, not a subversion.
Saw an ad yesterday (may have been a recorded program) for Ford auto-drive. Showed possibly non-actors taking their hands off the wheel and exclaiming how cool it was.
It is called ‘BlueCruise’ - this is not the commercial I saw, but it is an interesting one. Plus, Ford is apparently pushing it pretty hard if they have multiple ads on it.
As a manufacturing engineer, the software engineers developed lots of software for me that I helped test and that’s definitely how it’s typically done but that’s just a convention, not a natural law. They may have a different internal convention for when they rollover the base number or v13 is designated for a specific change rather that a big improvement on the current base. There are plenty of legitimate criticisms but this is trivial.
If you follow semantic versioning rules, you shouldn’t update the major version unless you are making backward-incompatible API changes. Updating the minor version (add functionality in a backward compatible manner) seems appropriate for this kind of update.
(Not that most software companies use strict semantic versioning. Usually “major releases” are done for marketing purposes only, not based on the semantics of what has changed.)
Version numbering is one of those things that can turn into religious war among software developers. 
Similar for part numbers and manufacturing engineers. I’m so glad to be retired
Here is how “cool” BlueCruise is:
https://www.ntsb.gov/investigations/Pages/HWY24FH006.aspx
One would think that NOT running into the rear of a stationary vehicle would be about the easiest thing to program. [although Tesla has this problem too.]
word in Michigan is the system will be adapted via OTA - update…
they plan to rename it to “BlueBruise”
ps: the fotos look of the accident do look ugly
A little tangential to the topic at hand, but I spent the better part of last week in a Kia with their lanekeeping system. I don’t think they advertise it as autonomous or anything, but dang, that thing is a disaster.
On the plus side, it does more or less function as a lanekeeping system and is (arguably) slightly better than nothing. But it behaves in an incredibly unsafe manner: it gives you almost no notification that it’s turned itself off. No sound, no steering wheel feedback–just a tiny icon that changes from green to gray and some road lines that go from white to light gray.
And it turns itself off all the time. Slightly faded road lines. Modestly inclement weather. Curves that are sharper than the usual gentle freeway curves. Phase of the moon. Too many times I found myself drifting out of my lane because it just shut itself off for no obvious reason.
Tesla has gotten some crap for the same sort of thing, but: it’s far more robust in the first place, and it gives you audio and tactical feedback that it’s happened. Kia’s version is actively unsafe in comparison.
As I suspected, Elmo just dropped the price from 199/month to 99/month. Oddly the non-transferable life time sub is $12k
I agree - doesn’t make sense, nor holds water…
upfront payment of 10+ years worth of subscription and no portability???
Even in a best case scenario, it will be way beyond year 11 until you brake (ha!) even - if you include finance cost.
Sure, where do I sign up??? /s
the only reasoning I can visualize: while the system is stil not up for prime time, they charge a rather nominal monthly fee, and then up it if they get it done. The fact that they do not lower the “purchasing” price, might be indicative that they might get there in the next few months…and dont want to leave any money on the table
The only argument you can make is that it’s a hedge against a future price increase but that’s weak. I suspect that that price will drop as well.
It sounds like the Hyundai version. I think it’s just a completely different goal compared to Tesla or even Nissan, with pros and cons.
Hyundai calls it lane assist, and it is not intended to steer for you. Instead, it makes little adjustments if you start to lose focus and drift. And it works fine for that on straight highways or very gentle curves. As you note, it shuts off when markings are unclear or on bigger curves. I haven’t noticed a problem in rain - it works fine for me, unless it’s a very hard rain that is obscuring the road.
The downside is that it doesn’t do as much as Tesla. You still need to keep your attention on the road and actively steer. But on the plus side, it doesn’t encourage distraction.
I agree, but it really, really needs an obvious auditory tone that it’s lost tracking. I get that it’s an “assist” system. Nevertheless, I need to know whether it’s actually assisting or not.
I disagree that it doesn’t encourage distraction. It has the same problem as all of these systems: it can work very well for long periods and then stop working for some reason. I found myself using the Kia system for tens of miles at a time, only touching the wheel every so often to eliminate the nags (it actually allows me to go longer than Tesla). But then I hit some section with poor markings and it shuts off.
There’s another aspect that’s harder to articulate, but I guess related to the “assist” nature that they’re going for. With Tesla, it feels like there’s a positive exchange of control between me and the car and back. If I don’t like what the car’s doing, I grab the wheel and as soon as I move it with enough force, Autopilot disengages with a sound and I know I have full control. With the Kia, the assist is very weak, but also doesn’t disengage if I do something myself like change lanes. Sort of a weird syrupy feeling to the car’s inputs. I prefer the positive exchange of control.
Tesla will allow you to make a quick change without disengaging in certain cases. There was construction on the freeway once and two lanes were merging into one. The car was confused and I was merging with a truck. I punched the accelerator and got past them and then ceded control back. FSD didn’t disengage.
There are a lot of things that you need to relearn because it’s not the same as an ICE car and the FSD is something that you do together with the car. You’re the supervisor and the car does most of the work. It’s certainly not for everyone but it suits me fantastically.
I’d argue it’s your Tesla FSD experience influencing how you tried to use Kia’s. I’d never dream of going any distance with only occasional touches of the wheel, because I know it’s not supposed to be used like that and isn’t effective or safe when used like that.
There’s another update that just dropped. It looks like minor tweaks.
2024.3.15 FSD v12.3.4
Yup, just got it installed over here. Haven’t had a chance to try it.
I was very impressed with how 12.3.3 handled a construction zone. The actual lane markers had nothing to do with where the lanes were–that was strictly defined by a zillion cones. It navigated through just fine, crossing yellow lines at weird angles and all that. Only complaint is that it should have slowed down a little easier before entering the cone region.
It sorta doesn’t matter, though. Like Tesla, it only has torque sensing on the wheel for engagement detection. So you can’t just follow the movements of the wheel–you have to apply pressure in some direction to keep it from nagging, which means that if the system disconnects for some reason, you’re now actively steering out of your lane. And since there isn’t even a physical “clunk” that you can sense, you have to either constantly monitor the display or be really sensitive to where you are in the lane.
And for the record, it was effective on the 1400 miles of Texas highways that I drove on. Or, 99% of them… hence the problem.
Maybe it would be better if the system had a large guard band. That is, it did nothing if you were (say) less than 2/3 off-center. It would still keep you in your lane if you got distracted or fell asleep or something, but you’d still be forced to steer manually for comfort reasons.
We’ve been using the FSD as much as possible during our free month and I feel like it’s gotten better over the last couple weeks, plus I’ve learned how to urge it to do stuff. For doing stuff, I’ve learned that when it comes up timidly to an intersection without electronic controls (stop lights), I can give it a bit of “gas” to get it to stop being timid and get it’s ass in gear. On today’s trip, it did have two stupid moves where I had to intervene. But, my biggest complaint is still the speed control for two reasons: When we are driving on a 60MPH highway and I have manually set the max to 67, I have to manually jab the gas to get it up to speed many, many times. I get tired of looking down and seeing it going 60. These are twisty roads, but come on. The other bigger complaint is the slow down from 60 to rolling into a town 30 MPH speed limit. It will fucking rip along at 50’ish until I disengage FSD to slow it the fuck down. This is where the MAX reading on the display is 30. But it is happy to rip through this small town. I don’t get it. I can even roll it down to a max of 20 and it just keeps going, or so it seems.
I got 12.3.4 and only took one short round trip drive so far, but the speed control seemed better. I hope it really is, and not just the quirks of traffic masking the problem.
In the autopilot menu is an option to set the speed based on how fast humans would typically drive in conditions. My first thought on seeing that was the car was driving like that setting was enabled, even when it was disabled. I’m really hoping the speed issues were something very simple, like a logic error on that setting. I never did try enabling it to see what happens.
For another odd Tesla quirk, the short drive I took today I do multiple times per week, and today it decided to go a slightly different route. There is nothing wrong with the route the nav picked, but in over five years it has never picked that route.