|
Orvin posted:Is there a good place to look up home level 2 charger reviews? We were having the discussion about standard household outlets being fragile for extended heavy use, but you'd be getting a 14-50 outlet most likely, which is like an even bigger dryer plug. I haven't heard about problems with them? I got the The mach e appears to have a 10.5kw charger according to Clipper creek's website, so if you wanted the absolute fastest charging, you'd need a 60 amp circuit with a 48amp charger, but I'm not sure if it's worth the extra few $ to you or not. Would be cool to brag about 
Kia Soul Enthusias fucked around with this message at 19:36 on Mar 7, 2021 |
#
?
Mar 7, 2021 18:26
|
|
|
|
| # ? May 26, 2024 12:37 |
|
|
MrOnBicycle posted:The only thing I'm worried about with the Ioniq 5 is what the interior quality will be like. Hyundai/KIA generally do OK these days on interior. My ex had a Kia Soul (helmet!) and while I hated driving it, the interior quality was surprisingly decent for its price point and they throw in a lot of features that are uptrim from other manufacturers.
|
|
#
?
Mar 7, 2021 18:30
|
|
|
Orvin posted:Is there a good place to look up home level 2 charger reviews? I'd just get the portable ford charger and have a 14-50 outlet installed.
|
|
#
?
Mar 7, 2021 19:08
|
|
|
ilkhan posted:No bias there. Literally the only person saying that we are anywhere close to fully functional self driving cars owns a car company promoting their self driving tech, but you seem to believe that poo poo without concern for bias. Indiana_Krom posted:Though it wouldn't surprise me if in the next 2-3 years the forever FSD beta becomes more reliable and safer than human drivers in good conditions, not really a stretch because autopilot is already significantly safer than human drivers on freeways in good weather. The data to determine if autopilot is actually safer is insufficient because Tesla does not provide the requisite amount of detail. https://www.forbes.com/sites/bradte...y-be-right/amp/ Highway driving is a much easier problem to solve than the complex set of challenges that come with driving in areas with pedestrians, cyclists, intersections, construction, cars entering and exiting the road from multiple angles, etc. Not to mention insufficient or incorrect GPS data being more likely on surface streets than highways. The idea that you can take a large set of data, throw a neural net at it, and solve any problem is techbro fantasy. Tesla autonomy day presentations are still focused on solving problems like “how do we properly detect a stop sign behind foliage” and “how do we differentiate a bike on a rack on the back of a car from a bike in the road.” They are still at the “properly identifying the environment based on still images” phase. Getting from there to all of the things a real driver needs to do to safely navigate the roads is a significantly harder problem than anything we’ve successfully applied machine learning to thus far.
|
|
#
?
Mar 7, 2021 19:30
|
|
|
YOLOsubmarine posted:Literally the only person saying that we are anywhere close to fully functional self driving cars owns a car company promoting their self driving tech, but you seem to believe that poo poo without concern for bias.. The bias emerges not from skepticism about self-driving, which any sensible person can see is many years out, but from the idea that pedestrians should wear some sort of sensor target rather than telling drivers to drive their drat cars.
|
|
#
?
Mar 7, 2021 19:36
|
|
|
Zorak of Michigan posted:The bias emerges not from skepticism about self-driving, which any sensible person can see is many years out, but from the idea that pedestrians should wear some sort of sensor target rather than telling drivers to drive their drat cars. Sure, but regulators, at least in the US, don’t seem especially interested in doing their jobs so buying a gadget to prevent your neighbors self driving car from running you down seems like exactly the sort of solution we will end up landing on. Instead of stifling innovation we’ve encouraged it by creating an entirely new problem and then selling the solution to it!
|
|
#
?
Mar 7, 2021 19:54
|
|
|
I saw a prototype EV with manufacturer's plates here in Orange County, CA yesterday morning. Unfortunately I didn't have my phone with me to get a picture and it drove off as I walked closer to it. At first I thought it was a Lucid Air but doesn't seem to match exactly what's on their website. I don't know what else it could have been because it looked like a production-ready car.
|
|
#
?
Mar 7, 2021 19:56
|
|
|
heated game moment posted:I saw a prototype EV with manufacturer's plates here in Orange County, CA yesterday morning. Unfortunately I didn't have my phone with me to get a picture and it drove off as I walked closer to it. At first I thought it was a Lucid Air but doesn't seem to match exactly what's on their website. I don't know what else it could have been because it looked like a production-ready car. Was it a sedan or SUV/Truck?
|
|
#
?
Mar 7, 2021 20:22
|
|
|
YOLOsubmarine posted:Literally the only person saying that we are anywhere close to fully functional self driving cars owns a car company promoting their self driving tech, but you seem to believe that poo poo without concern for bias. Ah yes the car companies Waymo, zoox, and comma.ai .
|
|
#
?
Mar 7, 2021 20:22
|
|
|
The Gunslinger posted:I totally missed all of the Ioniq 5 stuff, it looks...really good? Good amount of range, solid aesthetics, great use of space for cabin and cargo, infotainment looks decent and supposedly will use the SK batteries. This is the first other EV I've seen that makes me want to get rid of my Model 3 and go with that instead. It would be a big step up in several departments without the big price jump of the Y. Anyways really interested in it! I don't know what the price will be but I imagine I could probably swap cars without any cash involved and the writeoffs on this one will have eaten the depreciation for me. Looks like a better deal than the ID4 to me. It does look pretty good so far (though the design of the rear doesn't really do it for me, looks too angular and way too much plastic). Question is when it will be available for delivery and what the pricing will be. ID4s are being delivered to customers in two weeks here in Sweden, and so far we have no clue what the pricing is or when you'll even be able to order an Ioniq 5. So the main advantage of the ID4 is that you can actually buy one, I suppose.  But on paper the Ioniq looks better spec wise.
|
|
#
?
Mar 7, 2021 20:37
|
|
|
Shamino posted:Ah yes the car companies Waymo, zoox, and comma.ai . Comma.ai is just open source ADAS currently and Hotz has been generally skeptical of the self driving hype. He thinks Tesla will get to L5 first but hasn’t made any predictions on when that will be other than “longer than Elon Musk thinks.” Waymo is focused on commercial applications like taxis and delivery and trucking. And it only works on areas that they have mapped so it has inherent limitations that human drivers don’t. It cannot operate in unknown situations. Krafcik said Waymo would be selling driverless cars in 2017 and then in 2020 but hasn’t yet made any new predictions that I’m aware of. They seem pretty focused on the extremely slow rollout of driverless taxis to more markets. Zoox is literally a purpose built delivery vehicle for Amazon and also requires detailed environment mapping. I don’t think anyone anywhere is pretending it’s right around the corner. Musk is basically the only person in the industry that’s claiming that L5 autonomy is right around the corner (and has been for years now).
|
|
#
?
Mar 7, 2021 21:46
|
|
|
DoomTrainPhD posted:Was it a sedan or SUV/Truck? A sedan but it was pretty big. Definitely not a Rivian product!
|
|
#
?
Mar 7, 2021 21:52
|
|
|
Shamino posted:Ah yes the car companies Waymo, zoox, and comma.ai . Waymo themselves have stated their cars can not do level 5 and they have problems with level 4 unless they are in a tightly controlled / mapped enviroment. Construction zones gently caress them. Snow / rain still confuse the cars. Anyone with a passing familiarity of the 1% problem of L5 knows it's at least a decade and most likely more away. YOLOsubmarine posted:The idea that you can take a large set of data, throw a neural net at it, and solve any problem is techbro fantasy. Tesla autonomy day presentations are still focused on solving problems like “how do we properly detect a stop sign behind foliage” and “how do we differentiate a bike on a rack on the back of a car from a bike in the road.” They are still at the “properly identifying the environment based on still images” phase. Getting from there to all of the things a real driver needs to do to safely navigate the roads is a significantly harder problem than anything we’ve successfully applied machine learning to thus far. While I absolutely agree that the joke that is called "AI" and trying to run scripts against a enormous dataset is laughable for L5 , I kinda think the Tesla approach of pulling their data from real world driving like they are is probably the best approach IF you want to attempt L5. Waymo has a lot of issues where their "AI" kept getting cars rear ended as they wouldn't do the creative rule breaking humans understand is necessary and just stop in traffic. That said either approach is still yeeeaaaaaaaaaaaaaaaaaars away from even doing L4 properly and is always interesting to read how esp Waymo were so bullish on the subject are admitting "Okay wait no, this is in fact really loving hard"
|
|
#
?
Mar 7, 2021 22:16
|
|
|
CAT INTERCEPTOR posted:Waymo themselves have stated their cars can not do level 5 and they have problems with level 4 unless they are in a tightly controlled / mapped enviroment. Construction zones gently caress them. Snow / rain still confuse the cars. Right now I can goto Arizona and ride around in a Waymo taxi with no driver, it'll even drop me off at the door of a busy department store with no issue. Structural problems where traffic laws were made to be broken but are obeyed by an AI are frictional and will be solved by municipalities eventually addressing them. People like to claim self driving will never happen because "what will happen in the . 00001% edge cases" but ultimately you can look at the aviation industry for that. If there is a weather event or road closure event the car simply won't continue through and wait for the issue to be resolved or turned over to a human driver.
|
|
#
?
Mar 7, 2021 22:23
|
|
|
Shamino posted:Right now I can goto Arizona and ride around in a Waymo taxi with no driver, it'll even drop me off at the door of a busy department store with no issue. Cool, so you also don’t think Tesla or anyone else will achieve L5 autonomy because what you’ve described is not L5 autonomy and not what Tesla is promising. You’re describing L3/L4 autonomy, i.e. “fully autonomous, except when it isn’t.” If a driver is still necessary to take over quickly when the system reaches its limits then it’s not an actually fully self driving is it?
|
|
#
?
Mar 7, 2021 22:46
|
|
|
Shamino posted:Right now I can goto Arizona and ride around in a Waymo taxi with no driver, it'll even drop me off at the door of a busy department store with no issue. quote:Waymo themselves have stated their cars can not do level 5 and they have problems with level 4 unless they are in a tightly controlled / mapped enviroment. Guess what that small area of Arizona is? Also interestingly I was just reading about Waymo and the fact that those "autonomous cars" have in fact a remote driver ready to take over and they do so regularly. quote:Structural problems where traffic laws were made to be broken but are obeyed by an AI are frictional and will be solved by municipalities eventually addressing them. I think there is a term for that approach that also shows just how badly it works. quote:People like to claim self driving will never happen because "what will happen in the . 00001% edge cases" but ultimately you can look at the aviation industry for that. If there is a weather event or road closure event the car simply won't continue through and wait for the issue to be resolved or turned over to a human driver. There are a lot of pilots in this forum that can confirm flying != driving in the sheer scope of the auto drive/fly problem and there is simply no way to compare the relatively straightforward autopilot of a plane to the highly sophisticated problem of driving. Also edge cases like "construction" and "rain" and "snow" I hardly call 0.0001%
|
|
#
?
Mar 7, 2021 22:55
|
|
|
YOLOsubmarine posted:Literally the only person saying that we are anywhere close to fully functional self driving cars owns a car company promoting their self driving tech, but you seem to believe that poo poo without concern for bias. CAT INTERCEPTOR posted:Waymo themselves have stated their cars can not do level 5 and they have problems with level 4 unless they are in a tightly controlled / mapped enviroment. Construction zones gently caress them. Snow / rain still confuse the cars. Full L5 (no steering wheel in the car) is a long long ways away. Full L4 (it just does there unused) is not. ilkhan fucked around with this message at 23:51 on Mar 7, 2021 |
|
#
?
Mar 7, 2021 23:43
|
|
|
ilkhan posted:Of course Tesla is biased. But there's also videos of cars driving point to point without the driver touching the wheel. That is generally called evidence. I don't think they're going to be perfect, and they certainly are not perfect today, but better than humans is a bar they aren't far from achieving, if they haven't already. There are definitely some situations that are more difficult for them, and it will take a long time to get those sorted out. There's also evidence of autodriving cars utterly failing in the simplest of conditions. https://www.youtube.com/watch?v=x4fdUx6d4QM quote:Construction zones and snow/rain still gently caress with human drivers too. And yet flesh and blood still can work out way to handle them quickly while the computers on the autonomous car don't have any clue and need to call home for assistance. Full l4 is still years away. Dont kid yourself or believe Tesla for a second on this one.
|
|
#
?
Mar 7, 2021 23:56
|
|
|
YOLOsubmarine posted:Literally the only person saying that we are anywhere close to fully functional self driving cars owns a car company promoting their self driving tech, but you seem to believe that poo poo without concern for bias. Thank you for this post. The other major issue with neural net relates to comprehensibility of the actual model that was produced by training. Today it can be nearly impossible to understand a complex network. Lawyers are going to have a field day with this stuff.
|
|
#
?
Mar 8, 2021 00:32
|
|
|
YOLOsubmarine posted:Zoox is literally a purpose built delivery vehicle for Amazon and also requires detailed environment mapping. I don’t think anyone anywhere is pretending it’s right around the corner. I hadn't kept up with them, i've only ever seen their modded Toyota SUVs around SF. Didn't know they were gonna make hardware as well as just the self-driving stuff.
|
|
#
?
Mar 8, 2021 00:46
|
|
|
CAT INTERCEPTOR posted:There's also evidence of autodriving cars utterly failing in the simplest of conditions. I love this one because there's an entire team of people sat behind screens monitoring absolutely every single thing in the car. There's no way to get a more controlled environment than a closed race track where its the only car and it still stoves itself into the wall. "but there was an error" Yes, and despite the various failsafes, it still crashed.
|
|
#
?
Mar 8, 2021 01:37
|
|
|
I like to think that the car's AI was so good it became self-aware and committed suicide.
|
|
#
?
Mar 8, 2021 01:58
|
|
|
Interesting - the guy from Out of Spec Reviews tried using Plug & Charge on his Mach E. It was limited to 30 kW and took almost a minute to start session. He tried all the stalls at the station, then tried his app, and it charged at 120 kW. This isn't just an issue with Plug & Charge right? It's something Ford hosed up? https://twitter.com/Out_of_Spec/status/1368723834644348929
|
|
#
?
Mar 8, 2021 02:53
|
|
|
RZA Encryption posted:I like to think that the car's AI was so good it became self-aware and committed suicide. That’s nonsense. They’ll hide in the garage and refuse to go out on the roads with the other killer drivers.
|
|
#
?
Mar 8, 2021 02:56
|
|
|
Westy543 posted:Interesting - the guy from Out of Spec Reviews tried using Plug & Charge on his Mach E. It was limited to 30 kW and took almost a minute to start session. He tried all the stalls at the station, then tried his app, and it charged at 120 kW. I think it could be EA. I have seen reviewers with the opposite problem, trying to use the EA app or paying at the station, and it screws up. But when they try plug and charge, it magically works. But there is something a little funky with the Mach-E charge curve. The Out of Spec guy posted a 0-100% DC charge video a day or two ago, and it drops off from 150kW based on time, not thermals. He speculates a bit in the video about Ford being conservative, and an update in the works.
|
|
#
?
Mar 8, 2021 03:01
|
|
|
CAT INTERCEPTOR posted:There's also evidence of autodriving cars utterly failing in the simplest of conditions. That's not a Tesla, zoox or a Waymo car. Also hundreds of thousands of flesh and blood drivers fail every year and 30,000+ of them end up dead.
|
|
#
?
Mar 8, 2021 09:50
|
|
|
Shamino posted:That's not a Tesla, zoox or a Waymo car. Also hundreds of thousands of flesh and blood drivers fail every year and 30,000+ of them end up dead. That's a relevant self driving failure from a team that is experienced. There have been many more high profile self driving failures and well documented problems that could be and have been bought up that have no realistic solution in the next decade Frankly the real solution is advanced driving but of course thats not going to happen. Self driving is a Techbro overhype and underdeliver.
|
|
#
?
Mar 8, 2021 10:48
|
|
|
Shamino posted:That's not a Tesla, zoox or a Waymo car. Also hundreds of thousands of flesh and blood drivers fail every year and 30,000+ of them end up dead. It's a race car, with self-driving specifically tuned for race driving and tuned for that particular track, which is probably the most controlled environment you can get, with no other cars, no pedestrians, no dogs running around, no street furniture. And it still massively hosed up, by taking itself out in the fastest possible way.
|
|
#
?
Mar 8, 2021 10:54
|
|
|
Its an experimental race car running experimental software and experimental hardware during a beta test. It was running off of false data provided by an error in data recording during an initialization lap. Yes this is a autonomous car that failed catastrophically but its not really comparable to the "self driving" cars that are on the road today. Please note how i put that in quotes because i do not support the notion that those cars are actually self driving or will be any time soon. I just want to point out that using that particular crash to point out that self driving cars aren't ready yet is disingenuous. There are enough videos of Teslas and other vehicles failing badly at driving themselves that support the point better.
|
|
#
?
Mar 8, 2021 11:59
|
|
|
Shai-Hulud posted:Its an experimental race car running experimental software and experimental hardware during a beta test. It was running off of false data provided by an error in data recording during an initialization lap. Yes this is a autonomous car that failed catastrophically but its not really comparable to the "self driving" cars that are on the road today. 4 out of 6 self driving cars in the same event failed to complete the required 3 laps of the closed circuit which they don't even have to do at speed. Yes this car fell foul of an error in data during the initialisation lap but here's the kicker, that error in data was confirmed by an engineer on the team but none of the vehicle operators actually acted on it. The question is why the car logged the error to begin with, why it was spotted and nothing done about it and quite why a self driving car couldn't see a solid wall despite everything else. The excuse was basically "we didn't expect the error" and in the programming the car is only looking for certain errors, this error occurred outside of what the software written was looking for so just ignored it and stoved itself into a wall. It just shows the level we're at with it when the first thing a self driving car does with information outside of the scope it has been programmed for is 'poo poo pants and crash'. It's a little worrying but shows exactly why it is going to be a long, long time before self driving becomes a reality on anything other than controlled roads with sensors absolutely everywhere on them. So this post isn't a total diss at self driving cars, here's a Roborace car doing something cool at Goodwood and dropping the driver off so he can go say hello to the crowd. https://www.youtube.com/watch?v=lYxCOjIXFOA Olympic Mathlete fucked around with this message at 12:53 on Mar 8, 2021 |
|
#
?
Mar 8, 2021 12:49
|
|
|
Noticed in another Mach-E video that the US spec Mach-Es include a EVSE with interchangeable 5-15 and 14-50 plugs. Like the Tesla UMC. Nice to see more OEMs doing this.
|
|
#
?
Mar 8, 2021 15:09
|
|
|
Shai-Hulud posted:I just want to point out that using that particular crash to point out that self driving cars aren't ready yet is disingenuous. While this is absolutely true, that video is goddamn hilarious. As soon as the car went active it took the fastest possible route right into the wall. Actual full self driving is starting to feel like one of those things that is perpetually ten years away. All the collision avoidance features that are coming along in tandem should make things safer in general though at least.
|
|
#
?
Mar 8, 2021 15:09
|
|
|
With how many times Musk has lied about FSD being right around the corner while taking people's money for it I don't understand why anyone would still give him the benefit of the doubt. Why is his promise *this time* so special from the last dozen times he's promised it?
|
|
#
?
Mar 8, 2021 15:43
|
|
|
Nothing. It's still a perpetual 10 years away thing. Except for all the intermediate steps and features they add / have added along the way. They can mostly navigate parking lots, they can mostly navigate surface streets, they can mostly navigate freeways. There's a lot of glue to put those together seamlessly, and a lot of work to do on each section. But the individual parts are getting visible progress. Not perfection, but progress.
|
|
#
?
Mar 8, 2021 16:10
|
|
|
I think the right answer is between "never going to happen" and "happening tomorrow." A lot of this is going to depend on infrastructure and mass adoption of vehicles that have the capability to begin with.
|
|
#
?
Mar 8, 2021 16:41
|
|
|
ilkhan posted:They can mostly navigate parking lots, they can mostly navigate surface streets, they can mostly navigate freeways. This reminds me of a friend who worked for JLR, they of course get in competitors cars often to see what they're doing. He brought over a lovely £100k Mercedes S class one time which is a lovely car and it has self parking mode. So we go out in it, drive to the shop and he tries it out. It 'sees' the cars parked, you select a space you want it to drive into and it tries to whip it in in reverse. Well after what seemed like an eternity of thinking it starts to slot itself into a spot without too much trouble but a little quicker than anybody in the car is comfortable with. A few days later he takes it back to work and goes to park it in the compound with all the other cars. A few of his colleagues are there picking up other stuff so he decides to show off by getting it to park itself while he has his hands out of the window. He picks a spot, the car figures out what to do and starts to move and stick its butt into the space but again at a speed higher than it probably should. It doesn't stop itself and fires into the front bumper of the £250k Rolls Royce Cullinan parked there. Luckily no damage was done but everyone that got to witness learned never to trust the self-parking of an S Class.
|
|
#
?
Mar 8, 2021 16:54
|
|
|
A car that can function as well as human driver “mostly” (and this definitely isn’t true of Teslas even now) means the human driver must be attentive *always*. And if that’s the case then what you’ve built is ADAS, not an autonomous car. And driver assistance systems are great! Coupled with an attentive driver they will certainly make cars safer. But they aren’t going to allow you to sleep through your commute or pull up your laptop and get some work done because the car may require you to take over on very short notice and if you have not been attentive you will lack the necessary context to make a quick decision. If your car can get you to work without intervention 99% of the time that’s not very useful if the 1% of the time it drives over a pedestrian or into the other lane or even just stops unexpectedly in the flow of traffic. It’s like a robot surgeon that is 99% as good as a real surgeon. Who the gently caress is going to use that?
|
|
#
?
Mar 8, 2021 19:20
|
|
|
In my experience, Autosteer can barely navigate a two lane road here if the markings drop off briefly. Full self driving is a long ways off. I am fine with a really good L2+ which is what Tesla is building toward now. I don't know how they will ever make it work for cities, they represent a ridiculous number of edge case scenarios that I'm not sure you could ever account for without car to car networking, pedestrian awareness systems and stuff like that. Regardless Autopilot is really nice for highway driving and if all they do is make a really good L2+ system for that then I'm ok with it. I am not paying $10k for software that doesn't follow me to my next car no matter how good it is.
|
|
#
?
Mar 8, 2021 19:31
|
|
|
CAT INTERCEPTOR posted:Guess what that small area of Arizona is? Also interestingly I was just reading about Waymo and the fact that those "autonomous cars" have in fact a remote driver ready to take over and they do so regularly.
|
|
#
?
Mar 8, 2021 20:12
|
|
|
|
| # ? May 26, 2024 12:37 |
|
|
Cicero posted:This is false, there are no "remote drivers" who "take over". A remote driver would be directly operating the car. That's obviously stupid and dangerous and Waymo doesn't do it (though IIRC there was at least once startup that was planning to anyway). Waymo has remote driving coaches that give the car higher level directives when it's confused (e.g. yes, you can go around that construction zone), but the car itself is still charge of driving. The car is not in charge of driving if it's getting instructions from home base. And simply put if it's getting instructions, that person doing the decision making is a human driver and they have taken over. Air quote "driver" and "taken over" if you want. The simple fact is that even Waymo have been walking back their promises and what they have told the public to expect
|
|
#
?
Mar 8, 2021 21:29
|
|