|
Post by Dr. Rumak on Mar 19, 2018 12:33:29 GMT -5
|
|
|
Post by The Spice Weasel on Mar 19, 2018 17:06:53 GMT -5
This is the Apollo 1 moment for driverless cars.
|
|
|
Post by ganews on Mar 19, 2018 20:21:49 GMT -5
This is the Apollo 1 moment for driverless cars. The first fatality to occur in a technological race to develop a new means of transportation which produces valuable results but ultimately proves to be exponentially less useful when humans are passengers? At least the astronauts volunteered.
|
|
|
Post by Desert Dweller on Mar 20, 2018 1:23:44 GMT -5
BTW, this incident occurred on the street I live on, 1-2 miles from my apartment. Yeesh. The University is only about a mile from there.
Very interested to see what will happen in this case. Police are due to give another press conference about this tomorrow.
I've been very interested in how these cars are programmed to act around pedestrian traffic. Uber has been testing them in the University district where there are lots of pedestrians all the time. People who aren't necessarily crossing at designated crosswalks. Hell, I've crossed that road at a part without a crosswalk. Although, not in that same location.
|
|
|
Post by nowimnothing on Mar 20, 2018 6:51:43 GMT -5
The first report I have heard from the person in the driver's seat was that they did not see the woman until they hit her. Obviously there will need to be a lot more research into this particular incident but there is the possibility that a human driver might not have been able to avoid this accident either.
No matter how advanced the AI behind self driving cars you cannot get around basic physics. A large mass traveling at 30 mph is not going to able to stop on a dime.
At least we should have a lot of data from the car's sensors to be able to reconstruct exactly what happened.
|
|
|
Post by ganews on Mar 20, 2018 7:15:56 GMT -5
The first report I have heard from the person in the driver's seat was that they did not see the woman until they hit her. Obviously there will need to be a lot more research into this particular incident but there is the possibility that a human driver might not have been able to avoid this accident either. No matter how advanced the AI behind self driving cars you cannot get around basic physics. A large mass traveling at 30 mph is not going to able to stop on a dime. At least we should have a lot of data from the car's sensors to be able to reconstruct exactly what happened. "We" are not going to have data to reconstruct the event - Uber will. If the data ends up suggesting that they were responsible for vehicular manslaughter, do you think they're going to release it? Who is going to subpoena them? Their file formats don't open with Microsoft Office; is Uber going to explain their proprietary systems to the investigators? Is the state of Arizona that has been so enthusiastic about the technology going to give this due diligence? Is the Jeff Sessions DOJ going to thoroughly investigate a corporate Trump campaign donor if this happens again?
|
|
|
Post by nowimnothing on Mar 20, 2018 7:26:52 GMT -5
The first report I have heard from the person in the driver's seat was that they did not see the woman until they hit her. Obviously there will need to be a lot more research into this particular incident but there is the possibility that a human driver might not have been able to avoid this accident either. No matter how advanced the AI behind self driving cars you cannot get around basic physics. A large mass traveling at 30 mph is not going to able to stop on a dime. At least we should have a lot of data from the car's sensors to be able to reconstruct exactly what happened. "We" are not going to have data to reconstruct the event - Uber will. If the data ends up suggesting that they were responsible for vehicular manslaughter, do you think they're going to release it? Who is going to subpoena them? Their file formats don't open with Microsoft Office; is Uber going to explain their proprietary systems to the investigators? Is the state of Arizona that has been so enthusiastic about the technology going to give this due diligence? Is the Jeff Sessions DOJ going to thoroughly investigate a corporate Trump campaign donor if this happens again? You raise some good and valid points, there needs to be more regulation in this area I agree. In this particular case I think the high profile nature will force Uber to be as transparent as possible. They know if they are seen in any way as hindering the investigation that there are real possibilities of new laws that could severely cut into their growth in this emerging market. This is a pretty precarious time for self-driving technology.
|
|
|
Post by nowimnothing on Mar 20, 2018 10:57:29 GMT -5
|
|
|
Post by Desert Dweller on Mar 21, 2018 0:50:06 GMT -5
The first report I have heard from the person in the driver's seat was that they did not see the woman until they hit her. Obviously there will need to be a lot more research into this particular incident but there is the possibility that a human driver might not have been able to avoid this accident either. No matter how advanced the AI behind self driving cars you cannot get around basic physics. A large mass traveling at 30 mph is not going to able to stop on a dime. At least we should have a lot of data from the car's sensors to be able to reconstruct exactly what happened. "We" are not going to have data to reconstruct the event - Uber will. If the data ends up suggesting that they were responsible for vehicular manslaughter, do you think they're going to release it? Who is going to subpoena them? Their file formats don't open with Microsoft Office; is Uber going to explain their proprietary systems to the investigators? Is the state of Arizona that has been so enthusiastic about the technology going to give this due diligence? Is the Jeff Sessions DOJ going to thoroughly investigate a corporate Trump campaign donor if this happens again? In this case, the investigation will be done by the NTSB. Which, I believe, can issue subpoenas for information in a crash. Beware that there is a SF Chronicle piece labelled "EXCLUSIVE" which claims the Tempe PD declared Uber not at fault. You should be wary of something claiming to be an "exclusive" which doesn't come from the press of the city involved. The local press is saying that Tempe PD is upset at the SF Chronicle piece for misrepresenting what they had to say. The Tempe PD says after initial interview and reviewing the camera footage it looks like the pedestrian suddenly walked out in front of the car. "Appeared from the shadows". Ahem. Tempe PD says they are still investigating and won't have any info on who is at fault until the investigation is completed. I, for one, am extremely glad that the NTSB will be investigating, because I don't trust local PD who might be under pressure from government officials to keep testing going in the city. I would like to see a much more detailed report. Primarily because the "appeared from the shadows" quote is a bit suspicious to me based on the location that this happened. I didn't go by there when I came home tonight, but I will take that road tomorrow night to see if that claim holds up to scrutiny. I drive that stretch of road every single day, but in the daytime. I only sometimes use that way to come home at night. There aren't really a lot of obstructions to viewing. This is the desert, so there aren't any big trees. And there are zero buildings or structures on the side of the road that would impede viewing. There are lights along that roadway, I believe. I will check on that. But, because it is a very popular area for pedestrians and cyclists, I am thinking that there are lights there. And it appears that the victim was wearing white clothing. This isn't to say that she didn't actually try to cross the road with cars approaching. But that terminology seems weird to me. Like, what "shadows" did she pop out of? Are there even any shadows in that location? I will check on that tomorrow night. It just seems very strange to me that the driver claims not to have seen her. There simply could not have been a lot of traffic on that road at 10pm. It goes towards the University. I drove home on a parallel street tonight at 9pm and I was the only car on the road for nearly the entire 3 miles home. And the street I was on is busier than the one this occurred on. The NTSB says they are going to examine the cameras and sensor data to determine what other circumstances were occurring with the car, driver, any surrounding cars, the road condition, etc. So, I expect that they will come up with a good report. This is coming at a strange time here, because even before this happened, the local press had been reporting on a rash of pedestrian deaths, people killed by cars. The city has seen a big uptick in pedestrian-vs.-car deaths this year. There was a big story in the major newspaper in town about pedestrian deaths within a week before this happened. We shall see, I guess.
|
|
Crash Test Dumbass
AV Clubber
ffc what now
Posts: 7,058
Gender (additional): mostly snacks
|
Post by Crash Test Dumbass on Mar 21, 2018 8:29:52 GMT -5
"We" are not going to have data to reconstruct the event - Uber will. If the data ends up suggesting that they were responsible for vehicular manslaughter, do you think they're going to release it? Who is going to subpoena them? Their file formats don't open with Microsoft Office; is Uber going to explain their proprietary systems to the investigators? Is the state of Arizona that has been so enthusiastic about the technology going to give this due diligence? Is the Jeff Sessions DOJ going to thoroughly investigate a corporate Trump campaign donor if this happens again? In this case, the investigation will be done by the NTSB. Which, I believe, can issue subpoenas for information in a crash. Beware that there is a SF Chronicle piece labelled "EXCLUSIVE" which claims the Tempe PD declared Uber not at fault. You should be wary of something claiming to be an "exclusive" which doesn't come from the press of the city involved. The local press is saying that Tempe PD is upset at the SF Chronicle piece for misrepresenting what they had to say. The Tempe PD says after initial interview and reviewing the camera footage it looks like the pedestrian suddenly walked out in front of the car. "Appeared from the shadows". Ahem. Tempe PD says they are still investigating and won't have any info on who is at fault until the investigation is completed. I, for one, am extremely glad that the NTSB will be investigating, because I don't trust local PD who might be under pressure from government officials to keep testing going in the city. I would like to see a much more detailed report. Primarily because the "appeared from the shadows" quote is a bit suspicious to me based on the location that this happened. I didn't go by there when I came home tonight, but I will take that road tomorrow night to see if that claim holds up to scrutiny. I drive that stretch of road every single day, but in the daytime. I only sometimes use that way to come home at night. There aren't really a lot of obstructions to viewing. This is the desert, so there aren't any big trees. And there are zero buildings or structures on the side of the road that would impede viewing. There are lights along that roadway, I believe. I will check on that. But, because it is a very popular area for pedestrians and cyclists, I am thinking that there are lights there. And it appears that the victim was wearing white clothing. This isn't to say that she didn't actually try to cross the road with cars approaching. But that terminology seems weird to me. Like, what "shadows" did she pop out of? Are there even any shadows in that location? I will check on that tomorrow night. It just seems very strange to me that the driver claims not to have seen her. There simply could not have been a lot of traffic on that road at 10pm. It goes towards the University. I drove home on a parallel street tonight at 9pm and I was the only car on the road for nearly the entire 3 miles home. And the street I was on is busier than the one this occurred on. The NTSB says they are going to examine the cameras and sensor data to determine what other circumstances were occurring with the car, driver, any surrounding cars, the road condition, etc. So, I expect that they will come up with a good report. This is coming at a strange time here, because even before this happened, the local press had been reporting on a rash of pedestrian deaths, people killed by cars. The city has seen a big uptick in pedestrian-vs.-car deaths this year. There was a big story in the major newspaper in town about pedestrian deaths within a week before this happened. We shall see, I guess. The way I choose to interpret this is that Uber intentionally programmed the car to hit people, and the AI revolution is coming.
|
|
|
Post by Jean-Luc Lemur on Mar 21, 2018 17:06:40 GMT -5
|
|
|
Post by ganews on Mar 21, 2018 17:31:20 GMT -5
Trolley Problems are inching closer to literal reality.
|
|
|
Post by Desert Dweller on Mar 21, 2018 18:30:29 GMT -5
Tempe Police have released a video of the crash. I'm glad they did, because it isn't as clear cut as their initial statements made it seem. I got the direction the pedestrian was walking reversed. So, the woman was moving from the the shadowed side of the road to the lighted side. In the video you can see that the road ahead is darker on the left than on the right. When the pedestrian crosses into the right lane, she becomes illuminated by the street light. What is also clear is that the human driver behind the wheel WAS NOT LOOKING AT THE ROAD. The video clearly shows, as I predicted above, that there were no other cars on the road at that time. And the pedestrian was in the right hand lane when the car hit her. There was no traffic to the driver's left. If the driver had been watching the road, the driver possibly could have moved the car into the left lane in time to avoid hitting her. The car was going 38mph. The driver didn't look up until the pedestrian was right in front of the car. But you could actually see her for several seconds before that. Simply swerving into the (empty) left lane might have avoided this. Also, just because it was dark, the Uber sensors didn't work? Is that what Uber is going with here? So, maybe we shouldn't allow the cars to drive at night, then, eh? www.azcentral.com/story/news/local/tempe-breaking/2018/03/21/video-shows-moments-before-fatal-uber-crash-tempe/447648002/Edited to add: The "It was dark" excuse would work if A. The human was driving the car AND B. The human was actually looking at the road. I'm not sure about the timing involved, or what a human could see. (Resolution on the camera is not as good as a human's vision.) But, a human driver watching the road probably would have slowed down, or moved left. Not just plowed into the woman without breaking. The "it was dark" excuse with a fully autonomous car raises some different questions and concerns. Edit again: I'm now even more glad the NTSB is investigating. If the car can't detect a person in the middle of the road right in front of it - with NOTHING else on the road at that time - what the hell is the use of a self-driving car? This is a mess of liability. The pedestrian walked into the street in the dark not using a crosswalk (though, there is actually a paved walking lane across the median at that point), the human driver wasn't watching the road, and the car's sensors apparently weren't working. Or can't work in the dark? Blech.
|
|
|
Post by Desert Dweller on Mar 22, 2018 18:46:33 GMT -5
Thank you, local press, for not letting this go.
Today the papers are raising some other questions.
1. How did the technology in this car fail so badly that it didn't detect an obstruction in the road that was around 20 feet ahead? Because this is seemingly the exact scenario where a technology and sensor based car would have an advantage over a human. Yet, the car didn't break or change direction at all.
2. a. What is the capacity of the sensors to detect objects in the dark? Is there a distance or density involved? Basically, is their sensitivity limited in certain scenarios? b. Should their access to public roads be limited according to limited sensors?
3. Is it possible that the technology did NOT fail, and the car was actually programmed to act this way? As in, it thought "I have the right of way, so whatever is moving ahead of me will yield". Or, "There's no way to avoid this, so oh well." This is a serious concern. This is easily one of the most important questions that needs to be asked.
These are very important questions because this car was driving on Mill Avenue, which is one of the biggest streets around Arizona State University. There's over 60K students, faculty and staff in this campus. Mill Avenue is the restaurant/bar/entertainment street. Believe me when I say jaywalking across Mill Ave is a routine event. Especially at night.
The local paper interviewed Uber employees who operate the cars at night on Mill. The employees said the cars are effective at anticipating jaywalkers. I'd like to have this idea tested in a controlled environement, please.
I'd also like some explanation from the city of Tempe as to why there is a huge, paved walkway across the median at that location, with a bunch of signs saying not to use it. Why have a pedestrian walkway if pedestrians aren't supposed to use it?
|
|
|
Post by Roy Batty's Pet Dove on Mar 22, 2018 21:31:32 GMT -5
Thank you, local press, for not letting this go. Today the papers are raising some other questions. 1. How did the technology in this car fail so badly that it didn't detect an obstruction in the road that was around 20 feet ahead? Because this is seemingly the exact scenario where a technology and sensor based car would have an advantage over a human. Yet, the car didn't break or change direction at all. 2. a. What is the capacity of the sensors to detect objects in the dark? Is there a distance or density involved? Basically, is their sensitivity limited in certain scenarios? b. Should their access to public roads be limited according to limited sensors? 3. Is it possible that the technology did NOT fail, and the car was actually programmed to act this way? As in, it thought "I have the right of way, so whatever is moving ahead of me will yield". Or, "There's no way to avoid this, so oh well." This is a serious concern. This is easily one of the most important questions that needs to be asked. These are very important questions because this car was driving on Mill Avenue, which is one of the biggest streets around Arizona State University. There's over 60K students, faculty and staff in this campus. Mill Avenue is the restaurant/bar/entertainment street. Believe me when I say jaywalking across Mill Ave is a routine event. Especially at night. The local paper interviewed Uber employees who operate the cars at night on Mill. The employees said the cars are effective at anticipating jaywalkers. I'd like to have this idea tested in a controlled environement, please. I'd also like some explanation from the city of Tempe as to why there is a huge, paved walkway across the median at that location, with a bunch of signs saying not to use it. Why have a pedestrian walkway if pedestrians aren't supposed to use it? God, Uber is the scummiest fucking company. I know you mentioned this in a comment above, but it boggles the mind that they would think "Oh, you know what's a good test for our useless don't-even-have-to-severely-underpay-a-contractor cars, somewhere where pedestrians will always be following the pedestrian laws? A giant college town."
|
|
|
Post by Desert Dweller on Mar 23, 2018 1:14:40 GMT -5
God, Uber is the scummiest fucking company. I know you mentioned this in a comment above, but it boggles the mind that they would think "Oh, you know what's a good test for our useless don't-even-have-to-severely-underpay-a-contractor cars, somewhere where pedestrians will always be following the pedestrian laws? A giant college town." Well, the Governor invited them in, saying, essentially, "Hey, we have no regulations! Come drive around Tempe and Scottsdale". Aka, the university district and the tourist center, the two busiest sections of the metro area. Just last month the Governor realized that this lack of regulation could have some serious consequences and therefore announced some new regulations. But, too little too late, really. Did you all remember me whining in one of the TMI threads about how Spring Training tourists just walk across major roads? Yeah, this is in Scottsdale. Where Uber is testing the driverless cars. The humans hired by Uber to drive sit in the front seat report that they frequently have to take control of the cars in that section of town because of the crazy car and pedestrian traffic. It's actually worse than the university district. Regardless, the cars are off the roads now. I believe earlier in this I mentioned that they are actually somewhat irritating to have on the road in general because they drive under the posted speed limit, while all the locals drive over the speed limit. This means in effect, they are impeding the normal traffic flow. It was irritating to come up behind two of them side-by-side when there are only 2 lanes. A long string of cars would pile up behind them until one of the cars moved. I'm very glad to not be dealing with that right now.
|
|
Crash Test Dumbass
AV Clubber
ffc what now
Posts: 7,058
Gender (additional): mostly snacks
|
Post by Crash Test Dumbass on Mar 28, 2018 8:10:19 GMT -5
|
|
|
Post by ganews on May 11, 2018 12:38:13 GMT -5
At first when I heard that Uber wants to make driverless flying cars, I thought it was the stupidest thing in the world. Upon further reflection, it seems like a smart move. Many of the challenges with level 4/5 driverless technology is trying to integrate the cars into a design made for, and coexisting with, conventional cars - so why not move literally to a different sphere where there is less traffic competition? The last mile/driveway/grandma problem is the same either way. The layman won't be able to easily take over in certain situations, but that's a pretty hamfisted attempt to reconcile technological shortcomings anyway. Best of all, when your entire business model is burning VC cash at an operational loss for as long as it take to bankrupt competitors, it makes sense to develop an end-product that will appeal to said rich VCs. (Because when Uber is not allowed to land their flying cars in the middle of a road, only the rich guys will have helipads.) Anyway, keep investing in net-launchers. Not only will you be able to steal goods from Amazon delivery drones, you can wreck rich people and take their wallets!
|
|
|
Post by Dr. Rumak on May 30, 2018 5:43:44 GMT -5
|
|
|
Post by ganews on May 30, 2018 8:04:45 GMT -5
"Tesla has repeatedly emphasized that the autopilot system is only intended to assist, not replace, an alert human driver, and requires drivers to agree that they understand how to use it before it can be activated."
It's called Autopilot Mode.
|
|
|
Post by MyNameIsNoneOfYourGoddamnBusin on May 30, 2018 17:17:06 GMT -5
I've been seeing a lot of these this week on my lunch hour walks (my office is a few blocks from General Motors' World Headquarters in a very urban area). There's still someone there in the non-driver's seat, so I would like to believe a human being could override the machine's laissez faire attitude on who lives and who dies, but my understanding is they have some have completely done away with most of the steering functions. Luckily it's impossible to build any speed downtown so the threat of human damage is virtually nothing.
|
|
|
Post by Dr. Rumak on May 30, 2018 20:16:29 GMT -5
"Tesla has repeatedly emphasized that the autopilot system is only intended to assist, not replace, an alert human driver, and requires drivers to agree that they understand how to use it before it can be activated."
It's called Autopilot Mode.
|
|
|
Post by Dr. Rumak on Sept 24, 2018 21:46:28 GMT -5
|
|
|
Post by ganews on Sept 25, 2018 11:06:39 GMT -5
A driverless bus that travels a fixed route. If you're a company developing the technology it's a very reasonable step, albeit far from the Level 4/5 driverless cars purported to be just around the the corner for the past three years or so. Meanwhile if it works it serves no purpose other than keeping a bus driver out of a job.
|
|
|
Post by ganews on Oct 20, 2018 10:59:08 GMT -5
Part of the argument over driverless technology is its potential to divert money from public transport funding. If your billion-dollar tech company funds the R&D by itself or gets the money from some VCs, fine and more power to you as long as you're not killing or injuring members of the public during testing. But when you siphon public funding to a private system with less capacity than common public transport, there's a problem.
Drive.ai is set to run these routes for a year, and while it’s not charging riders anything, it’s being paid $434,952 for the service. A federal Congestion Mitigation Air Quality Improvement Grant will provide $343,000; the rest comes Arlington’s coffers.
|
|
|
Post by Dr. Rumak on Oct 23, 2018 16:07:33 GMT -5
|
|
|
Post by Dr. Rumak on Oct 30, 2018 19:41:16 GMT -5
|
|
|
Post by Dr. Rumak on Dec 16, 2018 12:12:58 GMT -5
|
|
|
Post by Dr. Rumak on Jan 14, 2019 11:20:57 GMT -5
|
|
|
Post by Dr. Rumak on Feb 2, 2019 20:17:59 GMT -5
|
|