|
Valve put out a video showcasing the automated production line for the Steam gamepad. https://www.youtube.com/watch?v=uCgnWqoP4MM
|
# ? Dec 11, 2015 04:17 |
|
|
# ? Apr 25, 2024 23:04 |
|
C.M. Kruger posted:Valve put out a video showcasing the automated production line for the Steam gamepad. I'm super into Factorio. How can I play this game?
|
# ? Dec 11, 2015 04:36 |
|
Dr. Arbitrary posted:I'm super into Factorio. How can I play this game? Get a degree in mechanical/electronic engineering! Automation engineering is a crazy good field to work in.
|
# ? Dec 11, 2015 08:48 |
|
Doctor Malaver posted:Those people saying how driverless cars can do only highways and other simple situations and are still decades away, should watch this TED talk. Or look at this: https://www.youtube.com/watch?v=3yCAZWdqX_Y This is a real car in production which you can buy right now. It's on a real city street and working. Sure, the outside conditions are nearly ideal, but "decades away" is nonsense.
|
# ? Dec 11, 2015 23:01 |
|
This is what the world will look like: https://www.jacobinmag.com/2011/12/four-futures/
|
# ? Dec 11, 2015 23:02 |
|
Tezzor posted:Or look at this: They're called "ideal conditions" for a reason - they're the easiest conditions anyone could possibly think of, where the fewest possible potential problems exist. "Drive straight, stop when the car in front of you stops" is not complex - but self-driving cars will not be a thing until the actually complex problems are solved.
|
# ? Dec 11, 2015 23:48 |
|
Bastard Tetris posted:Get a degree in mechanical/electronic engineering! Automation engineering is a crazy good field to work in. It's what I'm doing right now. Started out taking a couple welding classes at my local community college, had the "introduction to manufacturing technologies" class as a prerequisite, and now I'm doing the whole machining and CAD/CAM design course. This semester was a pain though. 20 students in the CNC programming class, 1 working CNC bridgeport and required pencil plots for each of the 15 projects (sure we've got some Haas training simulators that visualize everything, but they don't count towards grades). And we missed two weeks of classes because the instructor had to go to industry conventions. Probably going to write a letter to the dean complaining about the lack of equipment (and my grade), they could at least get a couple cheap CNC routers like the Shapeko or Chinese eBay specials so that more than a few people can do their projects per class.
|
# ? Dec 11, 2015 23:54 |
|
Main Paineframe posted:They're called "ideal conditions" for a reason - they're the easiest conditions anyone could possibly think of, where the fewest possible potential problems exist. "Drive straight, stop when the car in front of you stops" is not complex - but self-driving cars will not be a thing until the actually complex problems are solved. But they're a thing right now. You can buy one right this second. Maybe it can't drive itself literally 100% of the time, but depending on where you live you're looking at near-ideal conditions a huge percentage of the time you'd actually be driving.
|
# ? Dec 12, 2015 00:15 |
|
Main Paineframe posted:"Stop if an object is in your path" is not a particularly impressive programming task, nor is it a representative showing of the potential difficulties involved in self-driving cars. Once you have sensors capable of clearly distinguishing objects in the surrounding space, it's relatively easy. That's why the "snowstorm" was a particularly important part of his post that you glossed over - whether it's a calculator or a self-driving car, it can only do its job properly if it is getting clear and reliable input that the processing unit can make sense of. Try pushing the buttons on the calculator with a beach ball and you might find it becoming a whole lot harder to get it to accurately perform the calculations you want. Likewise, poor weather conditions or unusual road conditions may impact the ability of the sensors to provide clear input to the processing unit, rendering it unable to accurately respond to its surroundings. The inability of a self-driving car to tell what am object actually is can cause serious problems, like slamming on the brakes in 70mph highway traffic because a plastic bag caught by the wind drifted across its path. Humans are also not particularly good at driving in zero-visibility situations. Also, they drive way too aggressively at the same time. Trust me, I live in Michigan, I know about snowstorms. I lived in a town that got an average of 80 inches of snow per year, my sister went to college in a town that got an average of 10 feet per year. I've driven snowstorms in vehicles with no antilock brakes let alone traction control. The best defense is to just not drive into a blizzard (i.e. prefer "ideal conditions"), and if you're an idiot who needs to be back at university for finals tomorrow (hi there) and the polar vortex hits you just have to slow the gently caress down. That applies whether you're a computer or a human, if you can't see you can't see. But people are loving awful at actually doing it. I narrowly avoided a human semitruck driver who lost it on black ice, among countless other near-incidents on that odyssey. The sudden stop was caused because the state police closed the freeway so they could peel someone off. I didn't break 35 mph the entire time and including stops it took three times as long as the trip over, but I made it safe because I slowed the gently caress down. In the end computers are just gonna have to do the same thing, and they will crash a non-trivial amount if you insist on driving them into blizzards. Humans wipe out a shitload in blizzards too. Paul MaudDib fucked around with this message at 02:37 on Dec 12, 2015 |
# ? Dec 12, 2015 02:28 |
|
Main Paineframe posted:They're called "ideal conditions" for a reason - they're the easiest conditions anyone could possibly think of, where the fewest possible potential problems exist. "Drive straight, stop when the car in front of you stops" is not complex - but self-driving cars will not be a thing until the actually complex problems are solved. We saw that car changing lanes too so it doesn't just drive straight. And what do you actually want it to do? I'm perfectly fine with a car that will drive me "straight" from A to B and slow down or stop for obstacles. That's how I drive too. If you can't use it in a snowstorm, so you won't. You can't really drive a BMW convertible in a snowstorm either and yet convertibles are "a thing".
|
# ? Dec 13, 2015 14:50 |
I think the point people have been trying to get at is that presumably the greatest overall safety benefits from self-driving cars can only be attained if they are the only cars on the road, which is not exactly feasible due to the existence of things like 'snow'. That said I would not be surprised if long-haul trucking and urban taxis became wholly automated within our lifetime, but then, even freight trains aren't self-driving yet and that seems like a comparatively much simpler automation task.
|
|
# ? Dec 13, 2015 15:03 |
|
Trains aren't automated because the economic incentive is way less pronounced. You have one or two engineers for a train hauling millions of dollars worth of material vs a cab driver driving a single person. As for the snow scenario keep in mind that traction control on modern cars is pretty sophisticated and at the point that cars already will just ignore the driver's input in certain scenarios. Yes there will still be the blizzards where it will be impossible for anything but manual driving but in those situations 90% of people are hosed anyways so it's not like computers are doing any worse for it.
|
# ? Dec 13, 2015 16:16 |
|
Doctor Malaver posted:We saw that car changing lanes too so it doesn't just drive straight. And what do you actually want it to do? I'm perfectly fine with a car that will drive me "straight" from A to B and slow down or stop for obstacles. That's how I drive too. If you can't use it in a snowstorm, so you won't. You can't really drive a BMW convertible in a snowstorm either and yet convertibles are "a thing". Well as it exists now these things are safety/convenience feature for rich people. The point is that an automated uber isn't a thing until most every problem is solved. You can't have the uber fleet shutdown in a moderate snow storm. Snow is brought up because it interferes with vision based driving systems more than the driving implications. Though it's a combination of both. My Subaru eyesight has disabled itself in rain a few times presumably because it decided it couldn't make reliable decisions.
|
# ? Dec 13, 2015 16:35 |
|
Doctor Malaver posted:We saw that car changing lanes too so it doesn't just drive straight. And what do you actually want it to do? I'm perfectly fine with a car that will drive me "straight" from A to B and slow down or stop for obstacles. That's how I drive too. If you can't use it in a snowstorm, so you won't. You can't really drive a BMW convertible in a snowstorm either and yet convertibles are "a thing". Convertibles didn't destroy the professional driver industry. Neither did cruise control. There will be steps to automate the easy but boring drudgery and grunt work of driving, but until cars can reliably self-drive the last mile, which isn't happening anytime soon, they won't put much of a dent in driver employment. Paul MaudDib posted:Humans are also not particularly good at driving in zero-visibility situations. I didn't say "zero-visibility". Self-driving cars don't see using a pair of human eyes mounted on a head that can be rotated, parked comfortably behind a gigantic windshield which is equipped with several different methods of snow removal. They use lasers, LIDAR, radar, ultrasound, and tiny cameras, and the trend for all of those is toward smaller and smaller. It takes a lot less snow to block a pinhole camera or confuse a low-power LIDAR image than it does a pair of human eyes, and these sensors are not nearly as well-equipped for clearing snow. Even rain is still a serious obstacle for self-driving cars with a quite visible effect on effectiveness - sensors, particularly LIDAR and cameras, are simply much more affected by conditions that impact visibility even slightly. Current autonomous driving features don't fare well in Michigan winters: quote:“We’re a lot farther from general use self-driving cars than those in Silicon Valley would like you to believe. The radar sensor in the front and the rear camera are completely covered. While the snow was falling, I had to turn off the parking assist because the falling snow was triggering the ultrasonic sensors causing the system to beep continuously while there was nothing around the vehicle.”
|
# ? Dec 13, 2015 17:06 |
|
How would self-drivin cars work in Atlanta, where an inch of snow shuts down everything
|
# ? Dec 13, 2015 18:35 |
|
DOOP posted:How would self-drivin cars work in Atlanta, where an inch of snow shuts down everything In the specific instance of the south in a snowstorm self driving cars might actually be fantastic compared to the comical alternative.
|
# ? Dec 13, 2015 18:49 |
|
asdf32 posted:In the specific instance of the south in a snowstorm self driving cars might actually be fantastic compared to the comical alternative. Yeah, lest anyone forget what happened last year in Raleigh, North Carolina after they got 2.5 inches of snow:
|
# ? Dec 13, 2015 19:20 |
|
Main Paineframe posted:Convertibles didn't destroy the professional driver industry. Neither did cruise control. There will be steps to automate the easy but boring drudgery and grunt work of driving, but until cars can reliably self-drive the last mile, which isn't happening anytime soon, they won't put much of a dent in driver employment. Most these systems are strapped onto rather than purpose built for cars which themselves are mostly for the purpose of testing software. No we won't see them have any significant impact this decade but these are incremental rather than revolutionary improvements and improvement in computer navigation/image processing has been accelerating the last few years. I would be more surprised if we didn't start seeing (with human backup) true autonomous vehicles by 2025.
|
# ? Dec 13, 2015 21:32 |
|
my wife bought a roomba. human labor will continue to be valuable at least in vacuuming for some time.
|
# ? Dec 21, 2015 05:36 |
Because if there's one thing the impending techno-dystopian future is lacking, it's low-paying menial service jobs.
|
|
# ? Dec 21, 2015 05:38 |
|
Real hurthling! posted:my wife bought a roomba. human labor will continue to be valuable at least in vacuuming for some time. Are they still pieces of poo poo? The one I bought 3 years ago was so bad, it's sitting in a closet because cleaning the thing out is more trouble than just not using it.
|
# ? Dec 21, 2015 07:22 |
|
I've had nothing but great luck with Roombas, and have evangelized them to others. If you let a vacuum run every day there's much less dust and dirt than would accumulate under manual vacuuming, so the roomba essentially does less vacuuming more often.
|
# ? Dec 21, 2015 07:29 |
|
Counter point - pet hair likes to wrap itself around the brush and it's a pain in the rear end. Also, my house is too big. I'd need like 3 of them. And it likes to strand itself on my patterned area rug - it thinks the black areas are a cliff and ends up getting confused and just sitting there, screaming at me to move it somewhere else. It's really only useful if you happen to have the ideal home for it.
|
# ? Dec 21, 2015 07:33 |
|
https://www.autonews.com/article/20151218/OEM11/151219874/humans-are-bumping-into-driverless-cars-and-exposing-a-key-flawquote:Turns out, though, their accident rates are twice as high as for regular cars, according to a study by the University of Michigan’s Transportation Research Institute in Ann Arbor, Mich. Driverless vehicles have never been at fault, the study found: They’re usually hit from behind in slow-speed crashes by inattentive or aggressive humans unaccustomed to machine motorists that always follow the rules and proceed with caution. So on the one hand, human drivers. On the other, later in the article there's anecdotes like a Google car backing up traffic by going 24 in a 35.
|
# ? Dec 21, 2015 16:55 |
|
spoon0042 posted:https://www.autonews.com/article/20151218/OEM11/151219874/humans-are-bumping-into-driverless-cars-and-exposing-a-key-flaw Yes, Google Cars are currently registered as Neighborhood Electric Vehicles which are not allowed to drive faster than 25mph or travel on roads that have speed limits faster than 35mph. quote:Low-speed vehicle is a federally approved street-legal vehicle classification which came into existence in 1998 under Federal Motor Vehicle Safety Standard 500 (FMVSS 500). There is nothing in the federal regulations specifically pertaining to the powertrain. In other words, the Google Car was doing exactly what their legal classification allowed them to do, and that was an example of an overzealous cop fishing for his ticket quota. 25 in a 35 zone is annoying, but that's why there's passing zones. Obstructing traffic tickets are basically a catch-22 anyway since you can get ticketed if you're doing slower than traffic speeds but also can get ticketed because you're going faster than the speed limit. I actually welcome the fact that self-driving cars are going to cut waaaaay down on that bullshit (and also speedtraps). Cops will have to find a new source of revenue to fund their sick fleet of Dodge Chargers. Paul MaudDib fucked around with this message at 18:17 on Dec 21, 2015 |
# ? Dec 21, 2015 18:12 |
|
I think if anything the strictly law-abiding self driving cars are exposing the fact that the current state of road rules throughout the world is completely broken by design.
|
# ? Jan 4, 2016 19:55 |
|
SwissCM posted:I think if anything the strictly law-abiding self driving cars are exposing the fact that the current state of road rules throughout the world is completely broken by design. I like that the real barrier for robot cars is that human laws are fundamentally illogical and cannot all be obeyed at the same time.
|
# ? Jan 4, 2016 20:40 |
|
Maybe eventually road laws will be rewritten to emphasize transportation efficiency and safety instead of revenue generation.
|
# ? Jan 4, 2016 23:00 |
|
When cars are, for the most part, unable to break the law, that will have to be the case. I can't wait for the inevitable police union fight against self driving cars.
|
# ? Jan 4, 2016 23:04 |
Paul MaudDib posted:Yeah, lest anyone forget what happened last year in Raleigh, North Carolina after they got 2.5 inches of snow: Why, exactly, did Raleigh turn into Mad Max after a moderate dusting of snow?
|
|
# ? Jan 4, 2016 23:05 |
|
Jazerus posted:Why, exactly, did Raleigh turn into Mad Max after a moderate dusting of snow? snowstorms in the south tend to produce driving catastrophes for three major reasons the ground is warmer, leading to higher formation of ice under the snow compared to the north, where the ground typically stays frozen and dry snow is so rare that most local agencies have little resources to deal with the snow. the entire city of atlanta has like maybe four plow trucks and keeps no salt or deicing chemicals on hand a higher proportion of people have no idea how to drive in snowy weather, and you really only need a handful of idiots to cause traffic jams typically a southern city's response to potential snow is just call a snow day, cancel school and tell people to stay home. in 2014 the snow started in the middle of the day, cities suddenly canceled school, everyone who had kids had to go pick them up at the same time and then you end up with a massive traffic jam https://www.youtube.com/watch?v=ctvRTFmZ-lA boner confessor fucked around with this message at 23:45 on Jan 4, 2016 |
# ? Jan 4, 2016 23:31 |
|
I'm involved in local politics and I've had a lot of fun talking to politicians about the consequences of automated driving when it comes to tickets. There are substantial chunks of the budget paid for by tickets, and there's also a lot of jobs that revolve around the ticket process, from police to write them to the courts and clerks that process them. I have a feeling that a lot of cities are going to run face first into this problem and then have serious problems when their police departments start getting creative to make up for lost traffic ticket revenue.
|
# ? Jan 4, 2016 23:44 |
|
Main Paineframe posted:They're called "ideal conditions" for a reason - they're the easiest conditions anyone could possibly think of, where the fewest possible potential problems exist. "Drive straight, stop when the car in front of you stops" is not complex - but self-driving cars will not be a thing until the actually complex problems are solved. The complex problem, image recognition, is already mostly solved (certainly enough to drive). You can get a half dozen functioning, realtime algorithms that performs at near or better than human levels on imagenet tasks running on your computer right now. The hard part of getting a computer to see and recognize objects is already done. The issue is, like in all machine learning tasks, getting enough usable data. We have sensors that give us good enough resolution to see in snow and fog. The problem is they cost significantly north of 10 grand and break if you look at them funny. Making sensors cheaper and more durable isn't really some daunting task.
|
# ? Jan 4, 2016 23:47 |
|
some sort of fish posted:The complex problem, image recognition, is already mostly solved (certainly enough to drive). You can get a half dozen functioning, realtime algorithms that performs at near or better than human levels on imagenet tasks running on your computer right now. The hard part of getting a computer to see and recognize objects is already done. The issue is, like in all machine learning tasks, getting enough usable data. We have sensors that give us good enough resolution to see in snow and fog. The problem is they cost significantly north of 10 grand and break if you look at them funny. Making sensors cheaper and more durable isn't really some daunting task. "Cheaper and more durable" normally come free with tooling for mass production.
|
# ? Jan 5, 2016 00:35 |
|
KillHour posted:"Cheaper and more durable" normally come free with tooling for mass production. Cheaper yeah, more durable no. That's a quality issue, and quality is expensive.
|
# ? Jan 5, 2016 00:40 |
|
It should be noted that these sort of sensors had, afaik, almost no large scale commercial application until quite recently. The latest buzz in the valley is interconnectivity, "the internet of things", and drones, which has cratered the cost of cameras and other sensors as companies cram them into everything they can.
|
# ? Jan 5, 2016 00:50 |
|
computer parts posted:Cheaper yeah, more durable no. That's a quality issue, and quality is expensive. More durable, too, when you're talking about sensitive electronics. They have to be able to survive an assembly line and shipping, which is more than you can say for most test/research equipment that lives its live a spec of dust or errant vibration away from being a million-dollar paperweight. Hand-built electronics are janky as gently caress.
|
# ? Jan 5, 2016 00:58 |
|
Popular Thug Drink posted:snowstorms in the south tend to produce driving catastrophes for three major reasons Plus I doubt anyone has winter tires, and summers are terrible on any amount of snow or ice. I know because I got caught out last year and had to drive home on high performance summer tires. Had no grip once there was any amount of snow on the ground at all, like even if you could see the asphalt through. I mean it's hilarious to watch unfamiliar people freak out, but I think it's a perfectly reasonable response to such rare events. We don't prepare for tornadoes in central europe either. Anyway, I wouldn't say that machine vision is solved as such - otherwise you'd just put a stereo HD camera like what people have and be done with it. Instead there are a bunch of other stuff like LIDAR, radars, ultrasonic sensors, and manually prepared data necessary to make this work under good conditions. A person can look at a completely snow covered road and just with a few clues estimate where the edges are, mentally divide the roads into lanes and pick one to drive in. If there's some magical sensor that would let a car do that, I'm very interested.
|
# ? Jan 5, 2016 02:36 |
|
Is there any reason why self driving cars wouldn't limitedly connect to one another anyway? Things would start getting crazy fast and efficient, especially for complicated high intersections in rush hour.
|
# ? Jan 5, 2016 03:11 |
|
|
# ? Apr 25, 2024 23:04 |
|
Nevvy Z posted:Is there any reason why self driving cars wouldn't limitedly connect to one another anyway? Things would start getting crazy fast and efficient, especially for complicated high intersections in rush hour. they certainly will, as it's way easier to handle some navigational tasks in a swarm. each individual vehicle would have less computational work to do if all the vehicles surrounding it are broadcasting their location, direction, and speed but this won't become really useful until a high enough proportion of cars are self driving. for maybe a couple decades you're still going to have a large number of manually operated vehicles on the road
|
# ? Jan 5, 2016 03:42 |