|
Paul MaudDib posted:I'm surprised it hasn't already hit this thread, but the FAA has released their rules on drones. They are going as tough on them as they can get away with. All models (including R/C aircraft) that weigh more than 250 grams must be registered with the FAA, plus all the previous exclusion rules that applied. You need ATC approval to fly them anywhere near most major cities (since those are in controlled airspace), even under 500 feet. People who have tried to get approval so far have reported that the ATC guys basically have no idea what's going on let alone any process for approving or rejecting them. Some people have reported success as long as they stay under the treeline, others report the ATC basically giving them the official version of "huh?". So the FAA hasn't really even trained their staff one this one, to date. This is interesting at a few levels. Fist I used to fly balsa wood planes an helicopters 10 years ago and it seems silly that they're getting caught up in this. Even huge heavy gas planes/copters didn't need registration and have existed for decades. There's a pretty big line in my opinion between 'drones' which are self stabilizing and self navigating and the manual kinds of RC vehicles which crash if they go out of sight. Second, it's worth noting the pattern here. Drones were hyped for applications like pizza delivery and now regulations are coming down hard on them because of the reality of the risks associated with them. This is quite possibly the path vehicle automation is going to follow as soon as people realize that "super cruise control" or whatever is actually the same as "read your phone while doing 70 down the highway with half-baked technology at the wheel" and arguably not safer than regular old fashioned driving (or driving with backup-only systems like auto-breaking and lane departure alarms). And note that right at this moment drones are probably actually capable of package deliver (in a limited ways) but it doesn't exist and won't for a long time because of the myriad of real life safety/legal/regulatory problems that come with wide-scale adoption of actually disruptive technology.
|
#
?
Dec 16, 2015 20:03
|
|
|
|
| # ? May 4, 2024 20:13 |
|
|
Turtle Sandbox posted:Im surprised Glass was leading the charge over industrial applications, where you don't have to worry so much about looks and marketability and can focus on the basic issues the technology has like when we all had to learn how to make smart phones that didn't suck complete rear end. It wasn't though! Similar devices have been in industrial applications since at least 2000, probably earlier. And they still are. And since the industrial users don't have issues with wearing battery packs and the like, they're capable of much longer usable time between charges and all that good stuff. One of the crippling parts of the google glass thing is that they knew the average consumer wouldn't want to have to wear something connected to it to hold enough battery for a reasonable use time, hence the small amount of battery loaded into the one side, and the usable lifespan of 60-90 minutes of active stuff.
|
|
#
?
Dec 16, 2015 21:26
|
|
|
Slack is actually useful, maybe people are envious
|
|
#
?
Dec 16, 2015 21:48
|
|
|
Gail Wynand posted:Slack is actually useful, maybe people are envious There's also the interview where he openly mocked VC's for throwing money at him. Not saying he was wrong, just a little gauche.
|
|
#
?
Dec 16, 2015 21:53
|
|
|
Turtle Sandbox posted:Im surprised Glass was leading the charge over industrial applications, where you don't have to worry so much about looks and marketability and can focus on the basic issues the technology has like when we all had to learn how to make smart phones that didn't suck complete rear end. Glass isn't all that amenable to actual AR applications that involve real-time image processing, which is a lot of the useful ones. It was basically just the ability to overlay notifications pushed from the phone, plus to take a photo. Like for industrial applications you could have an assembly diagram open but it couldn't tell you that you forgot to tighten bolt 37. For medical applications you could overlay a pulse but not have it overlay a cut-here line of the incision you need to make. Rift with input cameras would be much better on that stuff. Unfortunately near-field depth perception and fine motor manipulation are apparently garbage at present, which is a big problem. Paul MaudDib fucked around with this message at 22:25 on Dec 16, 2015 |
|
#
?
Dec 16, 2015 22:23
|
|
|
Paul MaudDib posted:Glass isn't all that amenable to actual AR applications that involve real-time image processing, which is a lot of the useful ones. It was basically just the ability to overlay notifications pushed from the phone, plus to take a photo. or hololens, which is pretty much what MS has done with it since the announcement.
|
|
#
?
Dec 16, 2015 22:25
|
|
|
hololens is vaporware on the verge of fraud.
|
|
#
?
Dec 16, 2015 22:31
|
|
|
I don't follow this thread but is the gist of it that todays tech was yesterdays tulips?
|
|
#
?
Dec 16, 2015 23:02
|
|
|
shrike82 posted:hololens is vaporware on the verge of fraud. usually the term vaporware is used when things don't have an actual product (thus the "vapor" part, as in thin air) rather than when the product sucks for consumers and is an expensive joke
|
|
#
?
Dec 17, 2015 01:03
|
|
|
Trabisnikof posted:usually the term vaporware is used when things don't have an actual product (thus the "vapor" part, as in thin air) rather than when the product sucks for consumers and is an expensive joke Hololens might as well be. The demos they've shown dramatically overstate the viewport size.
|
|
#
?
Dec 17, 2015 14:28
|
|
|
Since this is the self-driving car thread, AFAICT: http://www.bloomberg.com/features/2015-george-hotz-self-driving-car/ quote:The car does, more or less, have it. It stays true around the first bend. Near the end of the second, the Acura suddenly veers near an SUV to the right; I think of my soon-to-be-fatherless children; the car corrects itself. Amazed, I ask Hotz what it felt like the first time he got the car to work.
|
|
#
?
Dec 17, 2015 19:32
|
|
|
This is the Uber thread, right? I just read this comment on HN and thought it was pretty interesting:quote:An interesting problem of working with "real" cabs is that you have to compete with street hails (that is, people doing the traditional stand-on-the-curb-and-wave-their-arms deal). This is a surprisingly big issue.
|
|
#
?
Dec 31, 2015 09:12
|
|
|
Cicero posted:This is the Uber thread, right? I just read this comment on HN and thought it was pretty interesting: This is from a taxi app in Australia:  https://www.reddit.com/r/australia/comments/3xithw/no_uber_in_my_home_town_so_i_downloaded_the_local/
|
|
#
?
Dec 31, 2015 11:18
|
|
|
Subjunctive posted:Since this is the self-driving car thread, AFAICT: Dude seems smart, but insufferable. It'll be cool to see what he does as he gets older and hopefully more humble.
|
|
#
?
Dec 31, 2015 16:31
|
|
|
Ccs posted:Dude seems smart, but insufferable. It'll be cool to see what he does as he gets older and hopefully more humble. Looking at the picture of him at his desk, he definitely made sure to put that bitcoin mug where we could see it and know he's supports bitcoin, especially since it's at an angle it would never be at if he had actually been using it. The usual paradox where really smart engineers are often especially stupid outside of their specialization.
|
|
#
?
Jan 8, 2016 16:23
|
|
|
Did someone order some cheap LIDAR?quote:Yesterday at CES, Quanergy, an automotive startup based in Sunnyvale, Calif., held a press conference to announce the S3, a solid-state LIDAR system designed primarily to bring versatile, comprehensive, and affordable sensing to autonomous cars. The S3 is small, has no moving parts, and in production volume will be US $250 or less. According to Quanergy, the S3 is better than traditional LIDAR systems in every single way, and will make it easier and cheaper for robots of all kinds to sense what’s going on in the world around them. Still about a year out before production and the demo didn't quite meet what was advertised on the floor, but wow that's considerably cheaper than google's $75,000 whirligig.
|
|
#
?
Jan 9, 2016 08:37
|
|
|
uninterrupted posted:Among other things, they've been dragging their feet on implementing threading. Threading is awful.
|
|
#
?
Jan 9, 2016 19:57
|
|
|
Threading live chat sounds like about the worst possible way to implement real time chat
|
|
#
?
Jan 9, 2016 20:28
|
|
|
JawnV6 posted:There's also the interview where he openly mocked VC's for throwing money at him. Not saying he was wrong, just a little gauche. Got a link to this? shrike82 posted:hololens is vaporware on the verge of fraud.
|
|
#
?
Jan 9, 2016 21:15
|
|
|
ANIME AKBAR posted:Okay next time I see my advisor I'll let him know that thing in the lab doesn't actually exist and he should check into a psych ward. It's going to be a real product, you just wait!
|
|
#
?
Jan 9, 2016 23:08
|
|
|
I certainly don't think it's going to sell as a toy for playing videogames or something, but it's definitely valuable for training and education, and for remote control in industry. A few thousand dollars a piece is nothing to a big private university, or the department of defense.Freakazoid_ posted:Did someone order some cheap LIDAR?
|
|
#
?
Jan 9, 2016 23:54
|
|
|
LIDAR has been used for decades for similar things as you do with RADAR. And if you combine them as a backup for each other, it's quite helpful for uses like this. Do keep in mind that LIDAR systems tend to use non-normal-visible-spectrum light.
|
|
#
?
Jan 10, 2016 00:05
|
|
|
ANIME AKBAR posted:I certainly don't think it's going to sell as a toy for playing videogames or something, but it's definitely valuable for training and education, and for remote control in industry. A few thousand dollars a piece is nothing to a big private university, or the department of defense. Don't frequency shame
|
|
#
?
Jan 10, 2016 00:08
|
|
|
ANIME AKBAR posted:So it operates just like a radar, but with light? Why the hell would you do that? Does it still work if you get the surface wet or dirty? LIDAR has been a thing for a while: https://en.wikipedia.org/wiki/Lidar Its very useful for higher accuracy measurements, higher than radar or sonar can generally give. For example, it is used for highly detailed height maps of the ground for land management and resource extraction.
|
|
#
?
Jan 10, 2016 00:14
|
|
|
fishmech posted:LIDAR has been used for decades for similar things as you do with RADAR. And if you combine them as a backup for each other, it's quite helpful for uses like this. And a drop of water will refract IR and UV just the same. Anyways, I can believe that radar or LiDAR could work for an autonomous vehicle. But a dozen vehicles crowded at one intersection, all peppered with multiple transceivers, all firing at the same time? Good luck. hobbesmaster posted:Don't frequency shame Trabisnikof posted:LIDAR has been a thing for a while: https://en.wikipedia.org/wiki/Lidar ANIME AKBAR fucked around with this message at 00:28 on Jan 10, 2016 |
|
#
?
Jan 10, 2016 00:21
|
|
|
ANIME AKBAR posted:Anyways, I can believe that radar or LiDAR could work for an autonomous vehicle. But a dozen vehicles crowded at one intersection, all peppered with multiple transceivers, all firing at the same time? Good luck. I don't think that problem is as vast as it might seem, we've done tons of work on using phase and frequency shifts to improve fiber bandwidth for example. I think it should be easy enough to keep track of which coherent beam traveling at the speed of light is which.
|
|
#
?
Jan 10, 2016 00:25
|
|
|
When you have a one-dimensional medium like fiber, with only one transceiver at each end, yeah it's doable. Now you have several independent sets of transceivers operating in a 3D medium. Again, good luck.
|
|
#
?
Jan 10, 2016 00:33
|
|
|
ANIME AKBAR posted:When you have a one-dimensional medium like fiber, with only one transceiver at each end, yeah it's doable. Now you have several independent sets of transceivers operating in a 3D medium. Again, good luck. You're missing my point, it is actually still rather easy to use slight variations in the transmitter frequency, the coherent nature of laser light, and the fact that we're already using high precision timing to disregard another transmitter's direct beam or spurious reflection because the timing won't make sense and the light will be wrong. Compared to the more challenging question of "will the AIs all stop and refuse to go when a nervous pedestrian acts like they're crossing the road but won't actually do it" figuring out "is this my transmission" is easy. Trabisnikof fucked around with this message at 00:42 on Jan 10, 2016 |
|
#
?
Jan 10, 2016 00:39
|
|
|
Typo posted:Or it could be that companies like amazon and uber provides better services for the costumer than existing alternatives I can't speak for Uber, but I've studied Amazon. The language they use internally is "an obsession with the customer". It's one of their fundamental operating principles. I know with certainty that it is something they try to teach employees too. I'm understating it. Amazon literally is built on an attempt to be structured organizationally, ideologically, and physically (in the sense of it's physical logistics ) to better serve customers. The cutting edge is "Customer facing horizontal organizations that are process (rather than function) focused" If one digs into it it is remarkably similiar to ideas like the rhizome, the main example I can think of is Christopher: http://www.martin-christopher.info/publications/logistics-and-supply-chain-management
|
|
#
?
Jan 10, 2016 06:29
|
|
|
424331 miles, 69 events where the driver had to take over or bad things would have happened, 13 times the driver had to take over to prevent a crash, 10 of those times because of the autocar's fault. quote:Of the 69 reportable safe operation events, 13 were “simulated contacts” -- events in which, upon replaying the event in our simulator, we determined that the test driver prevented our vehicle from making contact with another object. The remaining 56 of the 69 events were safety-significant because, under simulation, we identified some aspect of the SDC’s behavior that could be a potential cause of contacts in other environments or situations if not addressed. This includes proper perception of traffic lights, yielding properly to pedestrians and cyclists, and violations of traffic laws. To be clear, however, these 56 events during the reporting period would very likely not have resulted in a real-world contact if the test driver had not taken over. give it up autocar failures
|
|
#
?
Jan 13, 2016 06:15
|
|
|
Trabisnikof posted:424331 miles, 69 events where the driver had to take over or bad things would have happened, quote:To be clear, however, these 56 events during the reporting period would very likely not have resulted in a real-world contact if the test driver had not taken over.  I, for one, am shocked that in-progress technology that is not available to consumers still makes mistakes. What is this world coming to? It's not good enough now, which logically means it'll never be good enough, ever. Everyone knows technology never improves. Pack it in, guys, we're going home.
|
|
#
?
Jan 13, 2016 07:14
|
|
|
Cicero posted:
And technology always improves to the point of absolute perfection and there are never longstanding bugs or problems, especially with the first generation of a consumer product. Not to say that we should give up on self-driving cars. It's just that there's this assumption that they will not just be better than human drivers, but overwhelmingly better than human drivers right out of the gate, is just not supported by the data we have.
|
|
#
?
Jan 14, 2016 00:47
|
|
|
e_angst posted:Not to say that we should give up on self-driving cars. It's just that there's this assumption that they will not just be better than human drivers, but overwhelmingly better than human drivers right out of the gate, is just not supported by the data we have.
|
|
#
?
Jan 14, 2016 01:10
|
|
|
Cicero posted:It's likely they will be better than human drivers 'out the gate' simply because car companies will wait until they're that good before a consumer release. The first generation of self-driving cars is going to be under an enormous amount of scrutiny, car makers know that and have no interest in being sued into oblivion. But I thought Uber was making self-driving cars?
|
|
#
?
Jan 14, 2016 01:41
|
|
|
There's a big difference between being sued because you violated protectionist taxi regulations and being sued because you killed a bunch of people.
|
|
#
?
Jan 14, 2016 03:17
|
|
|
ayn rand hand job posted:But I thought Uber was making self-driving cars? Uber is telling the money faucets "oh yeah, we're totally working on that" but it's unlikely they're actually accomplishing anything.
|
|
#
?
Jan 14, 2016 03:48
|
|
|
If not uber then gm and lyft.
|
|
#
?
Jan 14, 2016 03:52
|
|
|
Are human driver accidents ever tallied on a per-mile basis? Some googling suggests there is, but I can't find any actual reports. Without a comparison, I can't tell if 6149.724 miles per accident is worse than human drivers on a per-mile basis. Also, nice one google.
|
|
#
?
Jan 14, 2016 04:02
|
|
|
Freakazoid_ posted:Are human driver accidents ever tallied on a per-mile basis? Some googling suggests there is, but I can't find any actual reports. Without a comparison, I can't tell if 6149.724 miles per accident is worse than human drivers on a per-mile basis. fwiw that's less than half as many miles as an average american drives per year (and still less than the least frequent age cohorts) https://www.fhwa.dot.gov/ohim/onh00/bar8.htm Most of the reported statistics seem primarily concerned with fatal collisions rather than just "accidents" though.
|
|
#
?
Jan 14, 2016 04:23
|
|
|
|
| # ? May 4, 2024 20:13 |
|
|
to be more prosaic about it, the Google car is far worse than I am at driving since I haven't had a car collision in ten years 
|
|
#
?
Jan 14, 2016 04:24
|
|