|
Cicero posted:Why? New semis start around $100K a pop. That's before whatever it will cost for automato and a sensor suite that can actually handle the aggressive lack of standardization in loading dock layouts well enough to actually make the last mile portion of shipping possible. Who do you think can afford to do a full fleet replacement on a whim, exactly? Hell, the Post Office couldn't afford to replace their delivery fleet, much less the freight trucks. FedEx and UPS would go right under. That is, I note, totally ignoring the maintenance costs of adding all this automatiin.
|
# ? Apr 12, 2016 02:50 |
|
|
# ? May 22, 2024 09:17 |
|
Somewhat related to the current discussion: is Google getting too big? Other companies have such a narrow focus and strategy, even companies of similar size. I can't see Google/Alphabet being coherent across its entire company, and I don't understand the long-term strategy of cars, Nest, search, browser, OS, service provider, biotech and venture capital.
|
# ? Apr 12, 2016 02:57 |
|
Liquid Communism posted:New semis start around $100K a pop. That's before whatever it will cost for automato and a sensor suite that can actually handle the aggressive lack of standardization in loading dock layouts well enough to actually make the last mile portion of shipping possible. Who do you think can afford to do a full fleet replacement on a whim, exactly? * Having someone to do the loading/unloading is fine, I'm talking about driving on the roads. That's where the safety concern is. quote:Hell, the Post Office couldn't afford to replace their delivery fleet, much less the freight trucks. FedEx and UPS would go right under.
|
# ? Apr 12, 2016 02:58 |
|
Aliquid posted:Somewhat related to the current discussion: is Google getting too big? Google's strategy is the same as it has been the last 10 years: make money on ads, waste it on random 'cool' companies.
|
# ? Apr 12, 2016 03:01 |
|
Aliquid posted:Somewhat related to the current discussion: is Google getting too big? Good news, they're quietly selling off everything outside of their main ad driven business.
|
# ? Apr 12, 2016 03:02 |
|
Aliquid posted:Somewhat related to the current discussion: is Google getting too big? They don't have a long term strategy for Nest, which is why it's going under inside a year, two tops. That or losing the current CEO. Related to car chat, the ability of the government/your car insurer to always know where you've been is going to be an issue.
|
# ? Apr 12, 2016 03:05 |
|
computer parts posted:Good news, they're quietly selling off everything outside of their main ad driven business. Wasn't that why Alphabet was created, so that Google would just be the ad-driven search-and-web business?
|
# ? Apr 12, 2016 03:10 |
|
bartkusa posted:What if the autonomous car lies to you about its intentions?
|
# ? Apr 12, 2016 03:15 |
|
Cicero posted:* The reduced need for drivers will incentivize trucking companies to automate once the technology is good. It'll be hard for individual operators to compete with not having to pay a driver; there was just a test in Europe that had one driver leading a convoy of automated trucks, which is one step down that road. You're still not considering the full picture. A road convoy is useless for handling the reason that we have semis in the first place, that being that the population of the US is so spread out that rail service isn't practical to get goods delivered where they need to be in a timely manner any more. Road trains will work fine for going distribution center to distribution center, but they are useless if you don't have multiple trucks worth of load going to the same locality. I mean, if we really wanted to have fun I could just ask you how an automated truck handles harvesting corn, but I don't want to blow your mind over having to handwave who's paying to replace all that equipment once the automation fairies figure out how to act autonomously offroad.
|
# ? Apr 12, 2016 03:22 |
|
Once a significant amount of cars on the road have the ability to run in automated mode, and have the ability to communicate with each other (is there a standard for this being drafted?) what I suspect you'll see is manually operated cars requiring some kind of transmitter that tells the automated cars around them to pay active attention to these cars since that would be easier than treating all the cars around them as equally unpredictable. Really 20 years from now Google is gonna be making bank selling your destination data to the government so they can optimize their smart traffic lights' cycles on the fly Hopefully the security for the Internet of things™ gets better soon, it seems to be pretty terrible right now!
|
# ? Apr 12, 2016 03:38 |
|
Liquid Communism posted:You're still not considering the full picture. A road convoy is useless for handling the reason that we have semis in the first place, that being that the population of the US is so spread out that rail service isn't practical to get goods delivered where they need to be in a timely manner any more. Road trains will work fine for going distribution center to distribution center, but they are useless if you don't have multiple trucks worth of load going to the same locality. quote:I mean, if we really wanted to have fun I could just ask you how an automated truck handles harvesting corn, but I don't want to blow your mind over having to handwave who's paying to replace all that equipment once the automation fairies figure out how to act autonomously offroad.
|
# ? Apr 12, 2016 03:40 |
|
Liquid Communism posted:You're still not considering the full picture. A road convoy is useless for handling the reason that we have semis in the first place, that being that the population of the US is so spread out that rail service isn't practical to get goods delivered where they need to be in a timely manner any more. Road trains will work fine for going distribution center to distribution center, but they are useless if you don't have multiple trucks worth of load going to the same locality. They actually might not be a bad idea if you aren't starting from the assumption that the purpose is to outright replace all the drivers or that all the trucks need to be going to the same destination. You could have one lead driver directing several trucks that are going in the same general direction, with each truck just peeling off when it needs to to get wherever it's going. That'd be a pretty huge reduction in workload assuming the lead truck/driver was switched off regularly. Of course, then you've just got a semi-automated convoy and convoys are hilariously dangerous for pretty much everyone else on the road.
|
# ? Apr 12, 2016 03:46 |
|
Liquid Communism posted:New semis start around $100K a pop. That's before whatever it will cost for automato and a sensor suite that can actually handle the aggressive lack of standardization in loading dock layouts well enough to actually make the last mile portion of shipping possible. Who do you think can afford to do a full fleet replacement on a whim, exactly? Automation would be fairly simple to add to modern semis. There's few industries more analytics hungry than logistics, so pretty much every truck sold today has a suite of programs allowing wireless monitoring of every system in the vehicle, and most of them are able to take control of the control systems in the trucks remotely (mostly because they're very poorly designed). Last mile isn't very important in wider logistics, as most of it would be yard-to-yard, and shunt trucks could do the work after any trailers were dropped off. It's still a million years away and I don't agree with the guy you responded to, but there's less obstacles than you're talking about.
|
# ? Apr 12, 2016 04:04 |
|
I hope Calico and Verily keep running, because those could lead to interesting discoveries. If Google backs off being a mega-corporation that tries to do life-altering thing rather than just making a ton of money as a search engine/ad platform, I'll be disappointed.
|
# ? Apr 12, 2016 04:05 |
|
Arsenic Lupin posted:Wasn't that why Alphabet was created, so that Google would just be the ad-driven search-and-web business? That or an excuse to make a bunch of "CEO" titles to pad people's resumes. In either case though, Alphabet is selling off the other stuff. The Robotics boondoggle is going away, and who knows how long their other pet projects will remain (Fiber will probably be around for a bit because it gives them all the user data they could want).
|
# ? Apr 12, 2016 04:05 |
|
computer parts posted:That or an excuse to make a bunch of "CEO" titles to pad people's resumes. Would you say Alphabet is reducing itself into an Abjad?
|
# ? Apr 12, 2016 04:12 |
|
cheese posted:I love when we go full circle and come back to the awkward reality of automated cars: they will exist largely to fill the role that should be occupied by widespread public transportation. Yeah basically. An automated car is a bus with stops that don't require you to walk as far and without drunk strangers on it.
|
# ? Apr 12, 2016 07:58 |
|
^^But poooors. Really, avoidance of other people is at least half of every tech innovation. Techlords still think they're the bullied.
|
# ? Apr 12, 2016 08:31 |
|
Cicero posted:* The reduced need for drivers will incentivize trucking companies to automate once the technology is good. It'll be hard for individual operators to compete with not having to pay a driver; there was just a test in Europe that had one driver leading a convoy of automated trucks, which is one step down that road. The problem with self-driving car chat is that you always get people blowing up one-off experiments and individual features into the whole future of how everything will work. The kinds of stuff you are talking about is decades away at best, and the scenarios you are envisioning may not even occur in our lifetime. Feasibility isn't swayed by how awesome something would be, or else we would all be riding around in jet packs. Trying to argue that market forces cause everything to once something is adopted is foolish, especially if we have no clear idea what the end product will actually look like.
|
# ? Apr 12, 2016 09:17 |
|
Peztopiary posted:^^But poooors. Really, avoidance of other people is at least half of every tech innovation. Techlords still think they're the bullied. i dunno, i don't want to be around yuppies and assorted nouveau riche types either~
|
# ? Apr 12, 2016 09:21 |
|
Cicero posted:What does harvesting corn have to do with banning manually driven cars from public roads? Truck goes in the field. Corn goes in the truck. Truck goes on the road to the co-op or silage site. It's not rocket science here. Banning manually driven cars from public roads, with the rhetoric given, runs into a whole lot of things we do that aren't 'commute from point a to point b in city traffic'.
|
# ? Apr 12, 2016 11:32 |
|
if automated cars would become mandatory half of Paris would lose their parking space
|
# ? Apr 12, 2016 14:12 |
|
robot cars are gonna kill a whole lotta cats and dogs
|
# ? Apr 12, 2016 14:16 |
|
Ccs posted:I hope Calico and Verily keep running, because those could lead to interesting discoveries. If Google backs off being a mega-corporation that tries to do life-altering thing rather than just making a ton of money as a search engine/ad platform, I'll be disappointed. Calico is actually an interesting topic for this thread, since it definitely seems to have been set up along the lines of the 'thanatophobe silicon valley STEMlord' trying to beat ageing and death. I work in biotech and Calico raised a lot of eyebrows when they started out by splashing a ton of money and making a lot of noise about trying to cure aging, without going into any specifics about how they are going about it. Since then everything has been almost totally shrouded in mystery except for the fact that they are still pouring huge amounts of money into hiring very influential people are partners, but still basically no details about how or what they are actually up to apart from the various non-profits, universities and researchers they have teamed up with. its so completely out of the realm of what normal biotech & pharma companies do i.e. create and sell drugs, that no-one has much of a clue what is going to come of this. It doesn't appear that they are trying to make medicines for marketisation because presumably they would have announced something in order to obtain funding. Obviously it would be cool if this all works out and they have a plan, but I think a lot of people have seen too many tech companies come in with some sort of WORLD CHANGING IDEA, get a tonne of funding and then come out the other side with no product or a a product that cannot possibly make any money.
|
# ? Apr 12, 2016 14:50 |
|
Yeah it seems crazy but hey go for it. The EU did a semi-similar thing (in the "it's crazy" sense) called the "human brain project" that I was NOT a fan of because it was government money but if Google wants to throw the money around then sure. There is a dynamic where a certain type of BIG THINKER (the grandiose and borderline con artist type) in a field manages to pull big money from like-minded or gullible rich people but Calico isn't like that, as far as I know, it's real goals seem to be the more practical "well what low-hanging fruit would add years to people's lives that might not be harvested by the usual biotech profit motive". But I'm not sure what's really going on in there. edit: well the other reason I am not a fan of brain project is that I'm actually positive it's crazy to try to do that with the current state of knowledge. I don't know enough about the biology of aging and age related diseases to actually know how much of a windmill that problem is. pangstrom fucked around with this message at 15:14 on Apr 12, 2016 |
# ? Apr 12, 2016 15:09 |
|
corn in the bible posted:robot cars are gonna kill a whole lotta cats and dogs Don't worry, they solved this by making them stop for anything in the road, up to and including plastic bags. It's shocking really, who could have possibly expected that something completely incapable of thought would have trouble distinguishing between a piece of debris, a small animal, and a toddler?
|
# ? Apr 12, 2016 15:13 |
|
The Larch posted:Don't worry, they solved this by making them stop for anything in the road, up to and including plastic bags. It's shocking really, who could have possibly expected that something completely incapable of thought would have trouble distinguishing between a piece of debris, a small animal, and a toddler? That's not a hard computer vision problem, and training data isn't scarce. Classification is well-understood, you could configure it to only hit certain breeds of dog. An unknown thing on the road should be avoided, just as by humans, but the set of unknown things will drop quickly over time.
|
# ? Apr 12, 2016 15:19 |
|
We have kind of beat the robo car thing into the ground but, for the "heh dumb machines" camp, please take a quick look around at your fellow drivers next time you're on the road. I know every time I have to slow down for a traffic snag I look in the rear view and pray the guy behind me isn't texting. I've seen a serious car wreck happen almost once a year over the last 5 years and I drive very little. People do a lot of things that are well beyond the bleeding edge of AI but moving vehicles through space isn't going to be one of them for long.
|
# ? Apr 12, 2016 15:23 |
|
Why do insufferable nerds always dream of a world full of autonomous self-driving cars?? Driving cars yourself is fun. gently caress the self-driving car crew.
|
# ? Apr 12, 2016 15:25 |
|
pangstrom posted:We have kind of beat the robo car thing into the ground but, for the "heh dumb machines" camp, please take a quick look around at your fellow drivers next time you're on the road. I know every time I have to slow down for a traffic snag I look in the rear view and pray the guy behind me isn't texting. I've seen a serious car wreck happen almost once a year over the last 5 years and I drive very little. People do a lot of things that are well beyond the bleeding edge of AI but moving vehicles through space isn't going to be one of them for long. And why, exactly, do you expect autonomous cars to be better at this than humans are?
|
# ? Apr 12, 2016 15:28 |
|
The Larch posted:And why, exactly, do you expect autonomous cars to be better at this than humans are? Because they have perfect attention.
|
# ? Apr 12, 2016 15:33 |
|
Because I am not just some computer fetishist and I'm not rooting for anything. It's a hard problem but there has already been a lot of progress and a lot of the edge case stuff is, probabilistically speaking, less of an issue than the stuff humans do gently caress up but computers do not gently caress up. We'll see where we are in 10 years and I'm happy to be proven wrong but pretty sure I won't be.
|
# ? Apr 12, 2016 15:35 |
|
How are u posted:Why do insufferable nerds always dream of a world full of autonomous self-driving cars?? Driving cars yourself is fun. gently caress the self-driving car crew. Famously, we prefer trains.
|
# ? Apr 12, 2016 15:36 |
|
pangstrom posted:We have kind of beat the robo car thing into the ground but, for the "heh dumb machines" camp, please take a quick look around at your fellow drivers next time you're on the road. For real. It is amazing how many people are staring at their smartphones while stopped at a light, and it's terrifying to think how many more are doing that while not stopped at a light.
|
# ? Apr 12, 2016 15:39 |
|
pangstrom posted:Because I am not just some computer fetishist and I'm not rooting for anything. It's a hard problem but there has already been a lot of progress and a lot of the edge case stuff is, probabilistically speaking, less of an issue than the stuff humans do gently caress up but computers do not gently caress up. We'll see where we are in 10 years and I'm happy to be proven wrong but pretty sure I won't be. This is something that will be interesting. Computers are likely to make different mistakes than humans do, but that means that the computer collisions that do happen may well look like things that should have been obvious. Modern nets can outperform humans on facial recognition, but the ones they get wrong are usually clearly distinct to a human doing the same comparison. It could be hard to gather data on collisions that don't happen, and on outcomes of ones that do (beyond fatal/not). If better micro-reactions reduce the severity of injuries, I wonder where that'll show up. I'm waiting for L2 in vehicles to progress far enough that insurance companies adjust rates -- in either direction! -- for vehicles that have the capability.
|
# ? Apr 12, 2016 15:46 |
|
Yeah it's akin to Watson's goofy Jeopardy errors. A little scarier when a car does it, of course.
|
# ? Apr 12, 2016 15:54 |
|
pangstrom posted:Yeah it's akin to Watson's goofy Jeopardy errors. A little scarier when a car does it, of course. Or the multiple pieces of facial recognition software that thought black people were shadows.
|
# ? Apr 12, 2016 16:03 |
|
The Larch posted:Or the multiple pieces of facial recognition software that thought black people were shadows. AFAIK, that phenomenon is still not well understood, but detection is easier than recognition, and a person is easier than a face (sometimes).
|
# ? Apr 12, 2016 16:07 |
|
Cicero posted:The difference there is that those things only affect the people using them. If some other dude decides to not wear seatbelts, it's no skin off my back. Explain how my emissions grandfathered car only affects me, or are you denying that man made global warming is real? Edit: it's a model t that burns leaded gasoline and I drive through school zones every morning.
|
# ? Apr 12, 2016 16:15 |
|
|
# ? May 22, 2024 09:17 |
|
How are u posted:Why do insufferable nerds always dream of a world full of autonomous self-driving cars?? Driving cars yourself is fun. gently caress the self-driving car crew. Driving nice cars casually is fun. Driving lovely cars is not.
|
# ? Apr 12, 2016 16:15 |