|
LITERALLY MY FETISH posted:First off, you can learn to weld well enough for a certification in under 6 months and there will never be enough welders to address the demand, so that's bogus as gently caress from the start. 
|
#
?
Apr 19, 2017 20:40
|
|
|
|
| # ? May 2, 2024 11:35 |
|
ReverendCode posted:You are agreeing with me very aggressively here, and I am not sure why. I am saying that automation does not have to replace workers on a 1 to 1 basis in order for automation to cause people to lose jobs. We don't agree because you don't know what the gently caress you're talking about and are trying to simplify it down. This isn't even "do research" territory, this is "talk to people who have actually worked in the restaurant industry at any level" territory. Here. Feel free to tell me about how the restaurant industry is losing jobs because of automation. I'll wait. A Buttery Pastry posted:Imagine there was 15% unemployment, with no prospects of the jobs most of those people held ever returning. Would there still never be enough welder to address the demand? What if unemployment was 25%? I mean, sure, you're technically correct in regards to being able to learn a new trade, but I think what people generally mean when they talk about a new trade is one which is not already saturated. The field I'm entering, court reporting, is in danger of losing ground to electronic recording not because it's better or cheaper, but because there aren't enough people getting through court reporting school to fill the jobs. I brought up welding specifically because I know about the education process and job prospects for it, not because it's special. There are a lot of high skill trades that are in desperate need to be filled because idiots keep lining up for english BAs instead of looking at actual career prospects. It's a little thing you may have heard of called "the skill gap" and it's actually a really big loving problem that automation can't magically make disappear. The issue people run into with learning new trades isn't that those trades aren't being used or are being phased out due to automation, it's because the training itself is being phased out through a combination of lack of interest due to the college boom, a lack of funding for training from the government, a weakening of unions, and a reluctance on the part of business to foot the bill when they haven't had to in the past because trade schools used to be a lot bigger.
|
|
|
#
?
Apr 19, 2017 21:16
|
|
|
Submarine Sandpaper posted:
well none of these are unsolvable problems unless you're a lazy tesla product designer who goes "nah we merely implied teslas can self drive instead of outright saying it so don't bother putting in more than one type of sensor" not running into a picture of a tunnel isn't hard and stupid drivers actively driving into walls isn't a self driving car problem anyway also i assume that sony truck got stolen despite being driven by humans so i'm not sure how that proves self driving trucks are never going to be practical
|
|
#
?
Apr 19, 2017 21:19
|
|
I CANNOT EJACULATE WITHOUT SEEING NATIVE AMERICANS BRUTALISED!
|
LITERALLY MY FETISH posted:There are a lot of high skill trades that are in desperate need to be filled because idiots keep lining up for english BAs instead of looking at actual career prospects. It's a little thing you may have heard of called "the skill gap" and it's actually a really big loving problem that automation can't magically make disappear.
|
|
#
?
Apr 19, 2017 21:21
|
|
|
blowfish posted:well none of these are unsolvable problems tesla killed a dude because the car thought a truck was the horizon. Dynamically update a truck's route with a detour or w/e and paint a tunnel, reap the benefits.
|
|
#
?
Apr 19, 2017 21:26
|
|
|
Submarine Sandpaper posted:tesla killed a dude because the car thought a truck was the horizon. to be fair the dude should have been paying attention instead of watching anime or whatever driver assist tech is nowhere near good enough to ignore the road so really it's just an expensive and complicated form of suicide if you let robojesus take the wheel
|
|
#
?
Apr 19, 2017 21:27
|
|
|
LITERALLY MY FETISH posted:We don't agree because you don't know what the gently caress you're talking about and are trying to simplify it down. This isn't even "do research" territory, this is "talk to people who have actually worked in the restaurant industry at any level" territory. So, the beginning of this particular discussion was discussing the rise of take-out ordering apps for "fast-casual" dining establishments, which is a literal example of the amount of work needed in the front of house being reduced, by automation even. This is a real thing that has, and is currently taking place, so again, I have no idea why you are so incredibly angry about this thing. Re: Court reporting. My first thought would be, How do you know the rise of automation is a result of the lack of recorders, and not the other way around? Edit: but it is likely that the rise of the machines will happen with or without an increase in the minimum wage, so why should automation influence whether or not we adjust it? ReverendCode fucked around with this message at 22:08 on Apr 19, 2017 |
|
#
?
Apr 19, 2017 21:58
|
|
|
It's super weird how "real" the skills gap is, yet it hasn't resulted in higher and higher wages as demand meets supply, isn't it?
|
|
#
?
Apr 19, 2017 22:24
|
|
|
LITERALLY MY FETISH posted:I brought up welding specifically because I know about the education process and job prospects for it, not because it's special. BLS doesn't seem to agree with you: quote:Welders, Cutters, Solderers, and Brazers. Employment of welders, cutters, solderers, and brazers is projected to grow 4 percent from 2014 to 2024, slower than the average for all occupations. A shortage of qualified workers in skilled trades doesn't actually mean that demand for those skills is massive, it just means that people aren't entering those fields. An English BA might honestly be more useful than a welding certification (although I guess having both would be even better?) just because any degree at all will open more doors by getting you through HR screens. Paradoxish fucked around with this message at 22:32 on Apr 19, 2017 |
|
#
?
Apr 19, 2017 22:29
|
|
|
boner confessor posted:or they could reference an internal database of road networks? that's how GPS units work and have worked for years and those dont try to scan road signs or anything. many highways are experimenting with variable speed limits, it would be trivial to have them broadcast a local "the speed limit is now 45" signal for robot vehicles Maintaining an internal database of every single road in America is an enormous undertaking, and updating it in a timely manner brings it into the realm of insanity. It just doesn't matter that much for GPS manufacturers because a certain degree of error is acceptable in a GPS. As for having local governments set up devices everywhere to broadcast the speed limit to self-driving cars, that's basically a non-starter because it requires municipalities all over the country to spend large amounts of money on upgrading their infrastructure for the sake of vehicles that no one owns because they don't work without that infrastructure. Airbnb has demonstrated that people are even comfortable with allowing strangers into their personal homes without the homeowner there to supervise them. And yes, Airbnbers have come home to things ranging from messes and garbage to used condoms to having their homes outright robbed (and sometimes all of the above). The property-owning middle-class seems to be downright addicted to the promise of turning their property into money without thinking too much about the externalities. boner confessor posted:i never said blindly trust the gps? i said gps provides route and navigational data. you're so desperate to win this argument you've forgotten to actually read my posts. go yell at a pillow or something if you need to vent If the route and navigational data is wrong, that tends to cause problems with navigating along a route. Besides, you're the one who said it didn't matter if the sensors couldn't read certain things because they'd be able to rely on navigational data instead. Now that people are pointing out the unreliability of navigation data, you're saying it doesn't matter if the data is inaccurate because the sensors will be able to handle it. Those sound pretty contradictory to me!
|
|
#
?
Apr 19, 2017 22:44
|
|
|
Submarine Sandpaper posted:tesla killed a dude because the car thought a truck was the horizon. blowfish posted:unless you're a lazy tesla product designer who goes "nah we merely implied teslas can self drive instead of outright saying it so don't bother putting in more than one type of sensor"
|
|
#
?
Apr 19, 2017 23:26
|
|
|
Main Paineframe posted:Maintaining an internal database of every single road in America is an enormous undertaking, and updating it in a timely manner brings it into the realm of insanity. it's a big undertaking, but it's not as big as developing a self driving car in the first place. there are numerous private firms that maintain road databases, google has their own proprietery one and you can access it from any mobile device for free. waze also exists. people keep saying this is just too impossible to be true but you can access multiple different up to date road databases right now, here, i'll help you: http://maps.google.com this is a map of every single road that exists in america, updated in a timely fashion. please explain how i am wrong Main Paineframe posted:It just doesn't matter that much for GPS manufacturers because a certain degree of error is acceptable in a GPS. it matters just as much for people as for self driving cars, i don't see why one is necessarily more sensitive to error than another? Main Paineframe posted:As for having local governments set up devices everywhere to broadcast the speed limit to self-driving cars, that's basically a non-starter because it requires municipalities all over the country to spend large amounts of money on upgrading their infrastructure for the sake of vehicles that no one owns because they don't work without that infrastructure. you didn't read my post correctly. i was talking about variable speed limits, which imply specialized infrastructure already exists - the speed limit signs which can change the displayed speed. they look like this:  so, you're talking about the infeasibility of upgrading infrastructure during the process of upgrading infrastructure... like i dont see how it would be difficult to strap a transponder onto this sign but ok Main Paineframe posted:Airbnb has demonstrated that people are even comfortable with allowing strangers into their personal homes without the homeowner there to supervise them. And yes, Airbnbers have come home to things ranging from messes and garbage to used condoms to having their homes outright robbed (and sometimes all of the above). The property-owning middle-class seems to be downright addicted to the promise of turning their property into money without thinking too much about the externalities. from what i've read most airbnb is actually commercial operators who don't actually live in the units they rent, instead of just using airbnb as a way to get around leasing restrictions. the hotel industry lobby published a report that indicates like 80% of airbnb rentals are not owner-occupied https://www.ahla.com/press-release/new-study-shatters-airbnb-homesharing-myth regardless, uber will certainly collapse long before true self driving cars are viable Main Paineframe posted:If the route and navigational data is wrong, that tends to cause problems with navigating along a route. how often have you used google maps and it was wrong? Main Paineframe posted:Besides, you're the one who said it didn't matter if the sensors couldn't read certain things because they'd be able to rely on navigational data instead. Now that people are pointing out the unreliability of navigation data, you're saying it doesn't matter if the data is inaccurate because the sensors will be able to handle it. Those sound pretty contradictory to me! i did? quote me. i'm pretty sure you just didn't read my posts correctly. it seems to be a trend
|
|
#
?
Apr 19, 2017 23:33
|
|
|
boner confessor posted:
boner confessor posted:
Right, the map service that still says the parking lot behind my apartment is a through street to the next street over, despite the fact that it literally ends against a brick wall 10 feet under the road Google thinks it connects to. And which took 2 months to finally get updated to not show that over in Quincy Center the road that used to go straight through the new "Quincy Square" park had been shut down and permanently rerouted - by the time the map data update came the road had been completely torn up instead of simply blocked off. And a whole bunch of other problems like that that I've personally experienced just in the past 2 years of living here. Also had plenty of data issues when I was down in Blacksburg, usually in the form of semi-private farm access roads being marked as normal public roads and trying to tell you to drive on them. Also had plenty of issues all over the place. Does it take work for you to be so stupid you think Google Maps is infallible or what.
|
|
#
?
Apr 19, 2017 23:47
|
|
|
just to be absolutely clear in case anyone is confused on how driving works, there are two significant activities a driver must undertake when driving: hazard avoidance. not crashing your vehicle into some obstacle. maintaining an appropriate speed and space cushion for immediate proximate conditions navigation. using your vehicle to travel from an origin to a destination along some route the navigation part is pretty well solved with technology, using gps devices. these exist and you can buy them in stores, today. you can even use your phone to access a database of road networks and road conditions, updated in near real time, over existing cellular data networks. there is no reason whatsoever to believe that this technology cannot also be used for a self driving car's automatic wayfinding. you could even transmit supplementary data, such as road metadata (including posted speed limit) that the vehicle would take into account when driving. this data is relatively small, such that it could be stored internally to a vehicle's navigational system and updated on some regular period. i believe tesla currently does this on a three month cycle hazard avoidance. this is the difficult part of a self driving car. the vehicle navigates itself through space and time, not hitting other objects. this currently does not exist in any meaningful way and is entirely theoretical as to how much capacity it has for intelligent decision making while collecting data about road conditions we could also attempt to read signs for navigational (street signs) and hazard avoidance (speed limit, warning signs such as stop or yield) purposes. However, given that it is likely that some of this signage may be missing, obscured, or otherwise temporarily illegible, it seems like a super good idea to also transmit the most up to date signage information along the current route for supplementary purposes, as a double check on the sensor collection system. tesla also currently collects road data and transmits it back to a central server, demonstrating the technological feasibility of real time updates to the road network i dont see how any part of this post is controversial fishmech posted:Right, the map service that still says the parking lot behind my apartment is a through street to the next street over, despite the fact that it literally ends against a brick wall 10 feet under the road Google thinks it connects to. And which took 2 months to finally get updated to not show that over in Quincy Center the road that used to go straight through the new "Quincy Square" park had been shut down and permanently rerouted - by the time the map data update came the road had been completely torn up instead of simply blocked off. And a whole bunch of other problems like that that I've personally experienced just in the past 2 years of living here. Also had plenty of data issues when I was down in Blacksburg, usually in the form of semi-private farm access roads being marked as normal public roads and trying to tell you to drive on them. Also had plenty of issues all over the place. i dont see how this matters? do you think gps will cause self driving cars to drive into walls or something? you're not making a whole lot of sense here fishmech, i think you're just upset that i pointed out you were wrong and you'll never go away until you get your revenge. please make more sense or stop quoting me, thanks or better yet, propose a navigational system that doesn't rely on gps positioning to locate a vehicle's position on a stored map boner confessor fucked around with this message at 00:00 on Apr 20, 2017 |
|
#
?
Apr 19, 2017 23:48
|
|
|
boner confessor posted:
You said: "this is a map of every single road that exists in america, updated in a timely fashion." So i gave you evidence of it not being updated in a timely fashion, in one of the largest metropolitan areas in the country as well as other places. You said: "how often have you used google maps and it was wrong?" so I gave you times when it was wrong. What don't you get here? Did you suffer a stroke? You specifically made a claim and asked a question predicated on Google Maps not being wrong often enough to be a problem.
|
|
#
?
Apr 20, 2017 00:00
|
|
|
fishmech posted:You said: "this is a map of every single road that exists in america, updated in a timely fashion." So i gave you evidence of it not being updated in a timely fashion, in one of the largest metropolitan areas in the country as well as other places. im sorry i made you mad fishmech and you are desperate to prove you're more correct than me but this still has nothing to do with what i'm saying. this is still the only feasible method for navigation, you're trying to say "it's not 100% perfect" as an answer to a guy who claimed "it's not technically feasible" despite it existing today and you're discounting the idea that it could be massively improved given a more pressing need like reliance for navigation or actual real time feedback from a fleet of vehicles (which already happens today, see tesla and waze) like i said, please calm down and read what i'm saying instead of trying so hard to burn me you embarass yourself. if you can't settle down and stop being such a terrible pedant i dont see the point in responding to your anger boner confessor fucked around with this message at 00:05 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 00:03
|
|
|
Jesus Christ you're both insufferably stubborn enough to argue this until kingdom come, please just take it to PMs. Or start a new thread for self-driving car talk.
|
|
#
?
Apr 20, 2017 00:10
|
|
|
boner confessor posted:im sorry i made you mad fishmech and you are desperate to prove you're more correct than me but this still has nothing to do with what i'm saying. this is still the only feasible method for navigation, you're trying to say "it's not 100% perfect" as an answer to a guy who claimed "it's not technically feasible" despite it existing today and you're discounting the idea that it could be massively improved given a more pressing need like reliance for navigation or actual real time feedback from a fleet of vehicles (which already happens today, see tesla and waze) This is a pretty lame attempt to backpedal from the fact you made an easily disproven claim and associated stupid question. Google Maps is a perfect example of how databases you might expect to stay updated in a timely basis actually aren't, but you can't admit that for some reason. The thing you're claiming "existing today" for literally does not exist, this was already proven.
|
|
#
?
Apr 20, 2017 00:21
|
|
|
fishmech posted:This is a pretty lame attempt to backpedal from the fact you made an easily disproven claim and associated stupid question. it's not disproven just because you deemed it so. millions of people every day find it an acceptable navigational solution, so does every major car manufacturer with built in gps navigation, but but based on nothing more than the strength of your own arrogance you've decided you're correct because it's not perfect? whatever dude. you're not bothering to read and interpret my words, instead choosing to nitpick them for debate points. i'm not going to sit here and argue reality with you just so you can achieve the psychic satisfaction of being correct on the internet. i'll let you continue to insist that google maps isn't broadly used and quckly updated if it makes you feel better in your own imagination google maps is updated in a timely fashion and it's nobody's fault but yours if this facet of reality annoys you boner confessor fucked around with this message at 01:01 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 00:59
|
|
|
https://www.youtube.com/watch?v=Y7QZgH1eP2o (USER WAS PUT ON PROBATION FOR THIS POST)
|
|
#
?
Apr 20, 2017 01:04
|
|
|
Automated vehicles are going to use google maps to navigate anyways
|
|
#
?
Apr 20, 2017 01:05
|
|
|
Control Volume posted:Automated vehicles are going to use google maps to navigate anyways they can't, it's impossible. cannot be done
|
|
#
?
Apr 20, 2017 01:09
|
|
|
boner confessor posted:it's not disproven just because you deemed it so. No, it's disproven because it's simply true that it isn't reliably updated. This is really simple stuff that you're denying for no apparent reason. boner confessor posted:i'm not going to sit here and argue reality Then why are you trying to argue reality, by claiming something that's easy to prove false. Lmao you just can't admit that google maps ain't infallible and frequently does not get its data updated, it's crazy.
|
|
#
?
Apr 20, 2017 01:19
|
|
|
Mr. Belding posted:But how could a computer calculate complex angular momentum formulae more quickly and accurately than a 45 year old guy with a GED, an STD, and an amphetamine habit? When I go into the field to explain how to secure cargo properly for ocean transport I use the Willie the meth head rule. If Willie can't understand and apply what I'm trying to teach, then I need to rework how I'm explaining it. Before I moved away from the south this was a literal thing. We would literally field test what we were trying to teach on the actual Willie. Now it's more an abstract philosophy. But, a lot of this math for the maritime stuff was constructed by drunks for drunks to apply in the field. So yes I believe one can teach a "45 year old guy with a GED, an STD, and an amphetamine habit" how not to overturn a truck. It's just hard and frustrating. By the time I left Willie could explain what a sail area was and why it mattered when he was securing a big yacht. He could even do the math for the simple rule of thumb model in his head. But that took years. But the point is even something relatively complicated like stability, can be boiled down to a page and taught to a literal drunk. It's easier if one uses older references, stuff written in the 40s, 50s and 60s was often written for in language for someone who only made it to the 9th or 10th grade. I actually have a manual for operating a steam plant written by the navy for a ninth grade reading level!
|
|
#
?
Apr 20, 2017 03:03
|
|
|
boner confessor posted:it matters just as much for people as for self driving cars, i don't see why one is necessarily more sensitive to error than another? Drivers are a lot less sensitive to error because if the GPS does something wrong, the driver can navigate by looking at signs and their surroundings. It's not a problem if the speed limit for the current road was decreased eight months after the last map update, because the driver is expected to be relying on signs as their primary means of determining road conditions and relying on the GPS as merely a source of advice. boner confessor posted:you didn't read my post correctly. i was talking about variable speed limits, which imply specialized infrastructure already exists - the speed limit signs which can change the displayed speed. they look like this: I didn't say it was difficult, I said it was expensive, and that municipalities are unlikely to have any interest in spending that much money to add infrastructure exclusively for the use of cars that don't exist yet because they rely on that infrastructure. boner confessor posted:from what i've read most airbnb is actually commercial operators who don't actually live in the units they rent, instead of just using airbnb as a way to get around leasing restrictions. the hotel industry lobby published a report that indicates like 80% of airbnb rentals are not owner-occupied "Mostly" is not the same as "entirely". There's plenty of examples of people who really did rent out their own home.
|
|
#
?
Apr 20, 2017 14:22
|
|
|
TheKingofSprings posted:I find it hard to imagine a data gap provided cars are able to "talk" to other nearby cars in the event of a failure of the main sensor what do you do in the case of a liar?
|
|
#
?
Apr 20, 2017 15:22
|
|
|
Main Paineframe posted:Drivers are a lot less sensitive to error because if the GPS does something wrong, the driver can navigate by looking at signs and their surroundings. It's not a problem if the speed limit for the current road was decreased eight months after the last map update, because the driver is expected to be relying on signs as their primary means of determining road conditions and relying on the GPS as merely a source of advice. self driving cars will exactly replicate this behavior. they currently examine their surroundings to maintain the hazard navigation function. since self driving cars mimic the behavior of humans, why do you think self driving cars would glitch out due to imperfect stored information if humans don't? google maps and waze can also stream near real time updates on road conditions to your phone, they do this today, right now, and people use this functionality to avoid traffic. there's no reason to believe that speed limit or route data would even be missing or out of date in the first place. it's the same as saying that the technology used now is the exact same as the technology which will be used in 5-10 years because you don't see how it could possibly be improved, despite the fact that there are numerous firms working on real time updates right now, in reality Main Paineframe posted:I didn't say it was difficult, I said it was expensive, and that municipalities are unlikely to have any interest in spending that much money to add infrastructure exclusively for the use of cars that don't exist yet because they rely on that infrastructure. it's not that much more expensive then the cost of the variable speed limit sign and necessary recieving equipment. like if it costs $5000 to upgrade to a variable speed limit sign, and $5100 for a variable speed limit sign + speed limit broadcaster for robot cars, then where is the problem? Main Paineframe posted:"Mostly" is not the same as "entirely". There's plenty of examples of people who really did rent out their own home. and plenty of examples of people who stopped doing it because their home was trashed boner confessor fucked around with this message at 16:27 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 16:24
|
|
|
boner confessor posted:google maps and waze can also stream near real time updates on road conditions to your phone, they do this today, right now, and people use this functionality to avoid traffic. there's no reason to believe that speed limit or route data would even be missing or out of date in the first place. It's not objectionable to say that at some near-to-mid-future point, a combination of large navigation systems and small perceptual systems will obsolete human drivers. Saying that ALL those systems are perfect TODAY is clearly wrong. It's also worth noting there are more failure modes than "outdated db" or "highway melted," google maps thought my hotel in China was in the water because they used the national maps. "The corporation wants to play nice with a government uncommitted to high-fidelity data." I remember the first big mistake I saw a GPS router make, it was unaware of a closed exit. The exit ramp ended up underneath the elevated highway, so without a 3D gps lock it didn't recognize we'd missed the exit and took a half mile or so to provide a re-routing. 30MPH speed limit difference between those two, if that sort of thing seems significant. Just the amount of work to integrate navigation & local sensors to pull off these tricks is staggering, I think it's a disservice to the folks who will be building those to take the position that things are perfect now.
|
|
#
?
Apr 20, 2017 17:41
|
|
|
JawnV6 posted:It's not objectionable to say that at some near-to-mid-future point, a combination of large navigation systems and small perceptual systems will obsolete human drivers. Saying that ALL those systems are perfect TODAY is clearly wrong. can you quote where i said it's perfect today? all i've said is that the technology to do this exist, it's good enough for millions of people to use, and there's no reason to believe it won't improve over the next decade this whole derail started because someone said "the only practical way for a self driving car to know the speed limit is complicated sensors" and i pointed out that you can reference a database with stunning accuracy, at which point the debate shifted to bickering over whether 99.5% vs 99.9% accuracy is sufficient to be considered practical JawnV6 posted:I remember the first big mistake I saw a GPS router make, it was unaware of a closed exit. The exit ramp ended up underneath the elevated highway, so without a 3D gps lock it didn't recognize we'd missed the exit and took a half mile or so to provide a re-routing. 30MPH speed limit difference between those two, if that sort of thing seems significant. and yet you didn't follow the navigational directions blindly, crashing your vehicle. why would you assume a self driving car would? plenty of people keep implying this would happen itt but nobody seems to want to attempt an explanation as to why boner confessor fucked around with this message at 17:48 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 17:44
|
|
|
boner confessor posted:self driving cars will exactly replicate this behavior. they currently examine their surroundings to maintain the hazard navigation function. since self driving cars mimic the behavior of humans, why do you think self driving cars would glitch out due to imperfect stored information if humans don't? The whole reason you proposed relying on stored information for things like speed limits in the first place is so that cars wouldn't have to have sensors capable of reading road signage. If the car can read street signs, then there's no need to keep a stored database of speed limit information in the first place. And if the car can't read street signs, then how the heck do you expect it to know that speed limit changed since the last database update?
|
|
#
?
Apr 20, 2017 18:15
|
|
|
jfc just hire rednecks to stick rfid tags on speed limit signs
|
|
#
?
Apr 20, 2017 18:26
|
|
|
Main Paineframe posted:The whole reason you proposed relying on stored information for things like speed limits in the first place is so that cars wouldn't have to have sensors capable of reading road signage. no, this is something you assumed. you said it wasn't practical to store this information, i pointed out that not only is it practical, but there are many consumer devices which do exactly this on the market today. a self driving car would necessarily have sensors to read road signs because it would need to have those sensors to self drive in the first place. i dont know why you would leap to this interpretation of my words because if you think about it for ten seconds it's pretty impractical. i can only assume you're choosing not to think about what i could be saying in favor of having some fake argument to beat up? Main Paineframe posted:If the car can read street signs, then there's no need to keep a stored database of speed limit information in the first place. there are many reasons why those street signs may be illegible, such as bad positioning, being obscured by vegetation or other vehicles, dirty, broken, or missing, etc. i'd bet money that a stored database of information would be more reliable and have less missing data than every street sign along a given route being readable. like if you're next to a large truck in heavy traffic that's multiple signs you can't read simply because the truck is in the way Main Paineframe posted:And if the car can't read street signs, then how the heck do you expect it to know that speed limit changed since the last database update? it's not necessary to update these databases frequently today, but you could update them more frequently in the future. there's no technical limitation to updating a database, only a limitation imposed by lack of current need. plus, if a car has route information streamed to it (which is exactly how google maps and other mapping tools work for mobile devices) there's no reason to believe that this information wouldn't be up to date given that there are already systems which exist today which utilize real time user feedback. this is why i keep bringing up waze, which is a real piece of software which exists today, you can download it right now, and it does what you say can't be done - provides real time road conditions along a route, including modifications such as speed limit reductions, accidents, detours, etc. google maps does the same thing boner confessor fucked around with this message at 18:31 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 18:28
|
|
|
boner confessor posted:
Well, hey, it looks like we've wrapped right around to the beginning of this argument: Main Paineframe posted:The only realistic and practical way for self-driving cars to determine what the posted speed limit is involves using complicated sensor suites to detect and read physical street signs It's certainly technically feasible to have a database with speed limits for every street in the country that's updated on a daily basis, but the logistics of actually doing in a timely, cost-effective, and reliable way are a considerable hoop to jump through, especially if it's just a backup system that the car doesn't even need most of the time. It's not a solved problem: Waze has problems with both timeliness and reliability, thanks to its reliance on unpaid volunteers responding to anomymous crowdsourced reports in their free time.
|
|
#
?
Apr 20, 2017 19:48
|
|
|
Main Paineframe posted:It's certainly technically feasible to have a database with speed limits for every street in the country that's updated on a daily basis, but the logistics of actually doing in a timely, cost-effective, and reliable way are a considerable hoop to jump through, especially if it's just a backup system that the car doesn't even need most of the time. It's not a solved problem: Waze has problems with both timeliness and reliability, thanks to its reliance on unpaid volunteers responding to anomymous crowdsourced reports in their free time. and you dont think thousands of vehicles scanning the immediate area and providing adaptive feedback wont work? this is somehow too outlandish in a scenario where we have cars driven by AI? especially considering that an extremely small fraction of that data will actually need to be updated on a daily or even hourly basis, given that speed limits rarely actually change? all the pieces are there, you're just refusing to put them together and declaring it impractical for reasons that you don't seem to want to elaborate on. like if it's such a considerable hoop to jump through why does it happen right now every day e: the more i research this topic the more absurd this derail gets as i discover people are arguing against the practical implementation of systems which have existed for years. it's been possible to transmit digitally encoded traffic info and alerts via fm radio for decades boner confessor fucked around with this message at 20:12 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 19:58
|
|
|
boner confessor posted:can you quote where i said it's perfect today? all i've said is that the technology to do this exist, it's good enough for millions of people to use, and there's no reason to believe it won't improve over the next decade I gave specific examples of things that will NOT improve over the next decade, would you care to address any of those or just ignore them? boner confessor posted:this whole derail started because someone said "the only practical way for a self driving car to know the speed limit is complicated sensors" and i pointed out that you can reference a database with stunning accuracy, at which point the debate shifted to bickering over whether 99.5% vs 99.9% accuracy is sufficient to be considered practical Having two inputs doesn't reduce the chance for error, it generally doubles it. If an analogy helps: Twin-engine planes that require a skilled pilot to use when one blows out are more dangerous because you've doubled the chance of engine failure, not safer because there's a backup. "Bb..but I added an engine! how could that make it WORSE??" boner confessor posted:and you dont think thousands of vehicles scanning the immediate area and providing adaptive feedback wont work? this is somehow too outlandish in a scenario where we have cars driven by AI? especially considering that an extremely small fraction of that data will actually need to be updated on a daily or even hourly basis, given that speed limits rarely actually change?
|
|
#
?
Apr 20, 2017 22:23
|
|
|
boner confessor posted:e: the more i research this topic the more absurd this derail gets as i discover people are arguing against the practical implementation of systems which have existed for years. it's been possible to transmit digitally encoded traffic info and alerts via fm radio for decades
|
|
#
?
Apr 20, 2017 22:28
|
|
|
JawnV6 posted:You've been going point by point with fishmech for pages over specific cases, you may not have said The Word "Perfect" but you're pretty tied to the infallibility-or-quick-correctability of these systems in the face of specific contrary evidence and handwaving it away. so what evidence do you have that there aren't numerous navigational apps which provide real time traffic data? JawnV6 posted:I gave specific examples of things that will NOT improve over the next decade, would you care to address any of those or just ignore them? did you? you just complained that your gps was wrong a couple of times. i dont see you saying that they can't or won't improve JawnV6 posted:Because we're talking about digital systems. When conflicts like "navigation says 60MPH, my local sensors predict 30MPH with 89% probability, road hazard 25MPH with 15% probability," someone has to resolve that with a decision system coded long before. Slamming on the brakes and going 30MPH on a highway is dangerous even if sensors deem it safe, speeding up to 60 on an offramp is dangerous even if navigation thinks it's fine, there's heaps of failure modes from any deviation other than correct. You're placing undue certainty on half the system to get things "right" with contradictory inputs rather than thinking the bad input will override the sum total of the decision. it's pretty easy to say "this theoretical system which doesn't exist won't function correctly if you design it in a dumb way" but i don't see what that has to do with the fact that the technology i describe already exists and if it's not already sufficient for the purposes i described, could easily be improved upon once there is an extant need for it JawnV6 posted:If you're just ginning up hypothetical systems whole cloth with features that don't exist on any shipping car, you're not talking abut "this decade," we're indulging a lovely attempt at sketching sci-fi. Throw me a character or a conflict. i'm sorry my description of potential future applications of technology differs from your own imagination? im not sure what your problem here is - vehicles ship today with built in gps navigation, gps navigation is also possible with a range of consumer devices. it's not my problem if that irritates you JawnV6 posted:A system that broadcasts at 2% of 2G speeds to an area of 69,400 square miles: appropriate for per-road local traffic data. We're not arguing against "practical implementation of systems which have existed for years," we're arguing against your ridiculous attempts to stretch them beyond recognition and still claim they're up to snuff for massively larger problems. i dont see why not, but the only thing people have been arguing for a couple pages now is just "it CAN'T work that way!" without any sort of argument as to why. this is why i keep repeating the amusing fact that, yes, it currently does work that way, and people just think these systems for transmitting traffic information will fail in the future because... why exactly? here's a real thing that exists right now, please tell me why this technology or future versions of it would not be applicable to self driving vehicles https://en.wikipedia.org/wiki/Traffic_message_channel as soon as i ask the question "you've asserted it won't work, but why wouldn't it work" people seem to not want to continue the derail, i wonder why that is? check this out: quote:Higher-end models of personal navigation assistants come with a built-in TMC receiver,[25] and depending on the country, the service is available in Eclipse, Garmin, iPhone (Navigon), Navman, Navway, Mio, Pioneer, TomTom and Uniden navigation systems, as well as in Volvo, BMW and Ford Falcon navigation systems, among many others. please tell me why all these manufacturers can't be using the technology they're using to do what you say can't be done e: i guess i'll knock this off for now because the inevitable future of self driving cars using internal databases of road networks supplemented by real time data transmitted via some central location is extremely obvious and uncontroversial to me, but for some reason this concept is causing multiple people to have fits for reasons i can't even understand and at this point i'm just antagonizing the mentally ill boner confessor fucked around with this message at 23:05 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 22:49
|
|
|
boner confessor posted:and you dont think thousands of vehicles scanning the immediate area and providing adaptive feedback wont work? this is somehow too outlandish in a scenario where we have cars driven by AI? especially considering that an extremely small fraction of that data will actually need to be updated on a daily or even hourly basis, given that speed limits rarely actually change? The question then becomes "why bother?" If the vehicles are already capable of gathering this data, analyzing it, and picking out the relevant bits without any external help...then why bother with a database in the first place? By automating away the data collection for the database, you've also automated away any need for the database. Navigation systems are useful because they tell you what you don't already know. It's like keeping an internal database of stop signs - you don't need to know where a stop sign is unless you're approaching one, but if you're approaching one you don't need a database to tell you where it is because you can see it with your own eyes.
|
|
#
?
Apr 20, 2017 23:34
|
|
|
Main Paineframe posted:The question then becomes "why bother?" If the vehicles are already capable of gathering this data, analyzing it, and picking out the relevant bits without any external help...then why bother with a database in the first place? redundancy. it's also easier to transmit or calculate "go to node 29025 along line 56720 no faster than 40mph, stop, turn left, go to node 570014 along line 34890 no faster than 60mph, stop, turn right" etc. so on because you already have to have this solution in place for a self driving car to figure out how to get to a destination in the first place. like cars aren't going to get where they're going just from sensors alone. how would they know which roads to take? you must supply a route to a self driving car, and it's trivial to supply road metadata along with that route, so... Main Paineframe posted:It's like keeping an internal database of stop signs - you don't need to know where a stop sign is unless you're approaching one, but if you're approaching one you don't need a database to tell you where it is because you can see it with your own eyes. when you're driving a route you know well, do you visually watch out for signage or do you just know what the appropriate actions are at each intersection from memory? boner confessor fucked around with this message at 23:44 on Apr 20, 2017 |
|
#
?
Apr 20, 2017 23:39
|
|
|
|
| # ? May 2, 2024 11:35 |
|
|
boner confessor posted:how often have you used google maps and it was wrong? Google Maps is really awful in rural areas with unmaintained winter roads, as just one example. In central VT, NH, ME it will regularly tell you to take roads that will leave you stranded. I see this as entirely solvable, but it unquestionably is a problem that requires a lot of resources.
|
|
#
?
Apr 21, 2017 00:03
|
|
