Hey, there! Log in / Register
Driverless cars can now roam entire city of Boston
By adamg on Wed, 06/20/2018 - 1:02pm
Boston today gave a company experimenting with driverless cars permission to expand its testing from the Seaport to the entire city. Don't worry, at least during testing, you'll still have somebody to stare down behind the wheel:
In all cases while testing, a professionally trained safety driver is behind the wheel prepared to take over as necessary. Additionally, nuTonomy, and its parent company Aptiv, utilizes a test engineer in the passenger seat to monitor system performance. Each vehicle is equipped with a suite of sensors that provide a 360-degree view of the surroundings. Testing may occur in both day and nighttime hours, and during some inclement weather, such as light precipitation, fog, and low temperatures.
Boston transportation officials say nuTonomy has had a good track record over its 18 months of testing in the Seaport.
Creating a policy on the operation of autonomous vehicles on City of Boston streets is a priority of the Go Boston 2030 Transportation Plan as the technology is capable of significantly enhancing mobility for Boston residents, particularly for senior citizens and people with disabilities.
City officials are hoping that the next generation of robotmobiles will cut down on the number of fatalities caused by human mistakes. One of the conditions of the expansion is that the vehicles always observe speed limits.
Neighborhoods:
Topics:
Ad:
Support Universal Hub
Help keep Universal Hub going. If you like what we're up to and want to help out, please consider a (completely non-deductible) contribution.
Comments
One of the conditions of the
It'll be easy to tell which cars are the driverless ones then.
Doesn't say they'll abide by
Doesn't say they'll abide by them, just that they'll see them. Still more than most drivers though.
4 fatalities
4 deaths so far in the US, 1 pedestrian in Tempe AZ in March, and 3 "drivers" when autonomous driving was engaged , says WIKI
40,000 Fatalities
40,000 deaths in the US in 2017, by actual drivers.
https://www.nsc.org/road-safety/safety-topics/fatality-estimates
I’m Sure You Weren’t
Trying to support the post you were replying to, but there have got to be more than 10k miles driven by humans for every mile driven by automotons, right? Id say the n is too low to draw a conclusion but I this is how we should be thinking about these stats.
good one!
Surprised to see you are the first to comment on this, too. Do you have a google alert set up with certain trigger words?
Awesome! The only thing I don
Awesome! The only thing I don't like about cars is the drivers! Glad to see them done away with.
Let’s reduce the number of drivers and ...
...the number of cars too. Get rid of them both.
Don't worry? Didn't that last
Don't worry? Didn't that last 'driverless' accident have a driver in the car at the time?
Yes, but dig this: 100% of
Yes, but dig this: 100% of accidents with non-driverless cars had drivers in them!
Sort of
There was a person sitting in the driver's seat who was supposed to take action if there was a problem, but Uber had gotten rid of the second person whose job it was to monitor the system and made the "driver" do it. Apparently it didn't occur to them that maybe giving someone two jobs wasn't a great idea, especially when one of the jobs (sit and do nothing at all until an emergency happens and you have to identify it and solve the problem in under two seconds) is one of the hardest things for a person to do.
Let's be honest
"There was a person sitting in the driver's seat who was supposed to take action if there was a problem" applies to most traffic accident as it currently stands. Humankind hasn't yet evolved to the point where we can handle a cell phone, a large iced coffee and a steering wheel all at the same time.
Yes, but...
the "driver" was texting at the time.
Screw this. These companies
Screw this. These companies shouldn't be using our public streets as their "testing ground". They can feel free to build a test city somewhere with all that sweet capital and come back once they're ready for primetime. Who will be Boston's first Elaine Herzberg?
Ever heard of
"real world conditions ." Hard to get those in a fake "city."
Spend a few hundred million
Spend a few hundred million of VC money on a replica city and staff it with people. Easy
nuTonomy?
I knew we were living in a William Gibson novel, but I was hoping it could at least be one of the earlier, better ones.
Fighting words
They're all great.
But yeah, we're heading for the Jackpot featured in the last one unfortunately.
We've had brainless drivers
We've had brainless drivers for years.
What could possibly go wrong?
There are all kinds of things that can, will, and have gone wrong. Just search the internet and you will see millions of articles.
But one specific question for all you advocates. How do these cars handle pedestrians who cross the street any time they want? On Dot Ave for example there are people crossing the street all the time. They dont care if there is traffic or a green light. They dart out from behind a parked car or a snow bank. They often dont look before they cross. As a driver, even if I am travelling at or below the speed limit, I have to be prepared to slam on the breaks or honk my horn to avoid hitting a person.
They hit them
"How do these cars handle pedestrians who cross the street any time they want?"
They just hit them, like that unfortunately lady in AZ who got hit by I think it was Uber's driverless car. I think driverless cars are a terrible idea on every level from public safety to the Uber drivers who will lose out on the extra income. In fact we should just boycott any company that insists on driverless cars.
Alternative?
It's not like human drivers have a good track record. Robot drivers killed 1 pedestrian, humans killed 6,000 last year alone.
I for one welcome our new robot overlords and their cars
https://www.npr.org/sections/thetwo-way/2018/02/28/589453431/pedestrian-...
Again
Humans drive so many more miles than robots. This statistic is either meaningless or stands in support of the idea that humans are currently safer drivers than robots.
Keep in mind
The autonomous car detected the pedestrian 6 seconds prior to impact. A human had decided it would be wise to turn off the emergency braking system that would have prevented the accident because it would generate too many false positives, braking frequently when no emergency existed.
At least according to the articles I read, a conscious human decision allowed this to happen (plus a distracted emergency driver).
Speaking of Dot Ave
One of the problems entering Dot Ave is that often the drivers view is blocked by parked cars or trucks at the corner. The only way to enter traffic is to blindly stick the front of the car out into traffic, and hope that someone will stop and let you enter. How does a driverless car , either on the main street, or trying to enter from a side street , handle that?
Just like we do
only with a faster reaction time when obstacles appear if the system is running as designed.
First Fatality Pool
Put me down for September 3rd
Before or after
Somebody Storrows a truck?
In a twist
the truck was driverless
Two intersections I'd love to see it test.
5-7pm.
Congress St. and Atlantic Ave.
Seaport Blvd. and Atlantic Ave.
They have been in the South Boston Waterfront
all this time so hopefully they have a handle on that -especially as they are expanding to other parts of the city.
Driverless cars aren't coming one way or the other, they are already here! Audi's Traffic Jam Pilot and GM Super Cruise will only get better in time. Hopefully, nuTomony can really contribute something special in the near future to the driverless experience in Boston.
What about cell phone restrictions?
As I recall with the automated Uber car that murdered that woman in Arizona, the human "fail-safe" was supposed to be paying attention if there was a need to take control from the computer but instead was looking at a cell phone, which was clearly shown in the dashcam video.
Edit: I've been corrected, seems it was a tablet connected for data collection. Still, the point remains, distracted driving kills.
Also they disabled the
Also they disabled the automatic emergency stop system, since it was triggering too often.
Worse per NTSB report
The NTSB determined that the "safety" driver was entering required data on an Uber tablet at the time of the crash. It was her job to focus on the road and also enter data on a computer. What could go wrong?
https://www.ntsb.gov/news/press-releases/Pages/NR20180524.aspx
Not a cellphone
She was looking at the console for the automated driving system, since Uber thought it was a great idea to get rid of the dedicated passenger who had been performing that job.
My opinion has evolved
First of all, the woman who was killed in Arizona was high on crystal meth and walked in front of the car in the dark. Horrible accident? Yes. Murder? Hardly.
I read a piece recently that said if all cars were driver-less, that auto-related deaths would drop 90%. In that scenario, 10,000 people a year would still be killed by auto accidents. The author was concerned that society would consider 10,000 deaths - caused by self-driving cars- unacceptable, even though the status quo is demonstrably far worse.
Whoa whoa whoa
Now this is victim blaming!
A human driver (not looking down at their cell phone) would have had a good shot at stopping in time. Not to mention Uber lied in their initial story.
Hey, you're starting to get it!
I guess a broken clock is right 2 times a day or something like that.
Did you watch that video?
Yes, I am blaming the victim. If you walk into traffic in the dark and are hit, it is your fault. You watch the video and you tell me if you would have been able to stop.
https://www.youtube.com/watch?v=XTXd5bfX_GI
Yeah, I have watched the video
The video is crappy, like the actual film quality is bad. Even good video has trouble "seeing" in both shadows and lit areas, compared to humans. She was visible to humans, and would have been seen by a driver paying attention. If you actually watch the video, you'll see that she was in the right side of the rightmost lane when she was hit, crossing from the left - that means she was in plain view for a decent amount of time. The driver wasn't paying attention, that's his fault. The car was equipped with LIDAR, which doesn't depend on visible light. The LIDAR saw her. It thought she was trash and ignored her. That's not her fault, that's Uber's fault. There's no designated or signaled place to cross the road there, in front of a popular concert venue. That's the government's fault.
A reasonable assessment
Could/should the Uber do better than a human? Absolutely. Will the systems get better from (terrible) accidents like this? They better and I trust that they will.
I would question that in this case she was "in plain view for a decent amount of time." Maybe the video isn't as clear as a human's vision would be, but to me when I see her, even if I slam on the breaks, she's done for.
If they insist on marketing this as a safety feature
They need to demonstrate that it will be safer than meatbag drivers.
Have you ever been outside at night?
Take a look and see if it looks anything like what that car could "see"
As a daily motorcycle rider...
... I’m torn. I hate cars, and hate drivers, but I also hate the thought of flipping off a driverless car. Where’s the fleeting satisfaction in that?
Good luck with all …
that
Even under good conditions -
Even under good conditions - all movements & markings around them clear & conventional - I have substantial doubts about whether these "autonomous" vehicles monitor all the import they need, can process it fast enough, and can make good decisions.
Add bad conditions to the mix - nonstandard movements, work zones, random & arbitrary BS, etc.... I have NO confidence in the ability of machines to extrapolate or imagine, which is a necessary part of a driver's skill set
Another concern would be
Another concern would be scope of input/programming/decisions - and how that can cascade. I'm thinking of situations where even good programming reacts to data/situations and initiates responses that are good for that narrow focus, but the programmers didn't think about the wider system or other users - with catastrophic results. Think of when you've heard stock market reports where programmed orders for trades, sell-offs, etc... when certain criteria were met triggered free-fall.
That being said, I have to be fair and say there are some good elements, and potential for some things. I'm fortunate to be able to drive a recent-model car some of the time. Some of the modern features (backup sensor alarms) are a bit erratic. Others, however, have been pretty good - lane departure alert, radar/dynamic cruise, emergency braking/alarm, fatigue detection.
Too funny. What are
Too funny. What are driverless cars used for? Deliveries? How much do they cost? I better start Googling.
Has your GPS ever made a
Has your GPS ever made a mistake?
Would you trust it with your life?
Regarding the Tempe, Arizona fatality
A June 22 article on ARStechnica says that the Tempe police have determined that the safety driver in that case was streaming video to her personal device during the time leading up to the crash, which contradicts both her prior statements and the findings of the NTSB investigation.