Pedestrian killed by self driving car.

Articles of interest to the AD community, currently in the news.
User avatar
Horse
Posts: 3558
Joined: Mon Sep 28, 2015 9:20 am

Re: Pedestrian killed by self driving car.

Postby Horse » Wed Mar 21, 2018 11:34 am

Triquet wrote:This is well and good when all vehicles are "connected", but consider the situation now where school buses are quite often the oldest vehicles in the fleet (due to general neglect and kiddy vandalism). They won't be "connected" for years.


Since many current drivers won't expect* to see pedestrians in the vicinity of buses, having an autonomous vehicles 'trained' to approach carefully could be an advantage.

* Similar theme: research on roadworker safety, over half the people interviewed wouldn't expect to see 'people' near to vehicles displaying amber beacons.
http://www.highways.gov.uk/knowledge/pu ... nspicuity/
http://assets.highways.gov.uk/specialis ... mplete.pdf

Drivers do not necessarily expect to see road workers
The participants in this study who were interviewed about this issue showed a varied set of expectancies regarding encountering road workers in everyday driving. More than half the participants reported that they would not expect to see people where a vehicle with flashing lights is stopped at the roadside.
Your 'standard' is how you drive alone, not how you drive during a test.

ancient
Posts: 179
Joined: Mon Dec 14, 2015 4:03 pm

Re: Pedestrian killed by self driving car.

Postby ancient » Wed Mar 21, 2018 12:20 pm

Strangely Brown wrote:
ancient wrote:Some interesting information about Uber's development strategy up to now: https://www.newscientist.com/article/2164297-self-driving-uber-death-should-halt-techs-race-to-the-bottom/


To continue reading this premium article, subscribe for unlimited access.Existing subscribers, please log in with your email address to link your account access.

Oops sorry! The paragraph I found particularly interesting was:
New Scientist technology writer Mark Harris wrote:Nevada was the first US state to authorise experimental autonomous vehicles on public roads, in 2011. Vehicles were treated like human learners, even undergoing a driving test with an examiner in the passenger seat. California followed shortly after, requiring companies to employ highly trained safety drivers, file reports detailing each bump and scrape, and note every time a vehicle’s systems failed.

States coming later to the game enticed companies with fewer regulations. In 2015, Arizona’s governor Doug Ducey issued an executive order opening the state’s road to autonomous vehicles with few licensing requirements and no California-style reporting necessary. The tactic worked. After Uber’s self-driving cars had their licences revoked in California for not complying with its regulatory regime, they fled across the border to Arizona.

My bold emphasis.
Also of interest is this: https://www.sfchronicle.com/business/article/Exclusive-Tempe-police-chief-says-early-probe-12765481.php
Apparently the Uber vehicle was
Traveling at 38 mph in a 35 mph zone

Yet was not at fault for not detecting and even attempting to slow by the time the object of the collision was right in front of it.

TheInsanity1234
Posts: 544
Joined: Fri Oct 02, 2015 8:03 pm

Re: Pedestrian killed by self driving car.

Postby TheInsanity1234 » Wed Mar 21, 2018 4:21 pm

Horse wrote:
Triquet wrote:This is well and good when all vehicles are "connected", but consider the situation now where school buses are quite often the oldest vehicles in the fleet (due to general neglect and kiddy vandalism). They won't be "connected" for years.


Since many current drivers won't expect* to see pedestrians in the vicinity of buses, having an autonomous vehicles 'trained' to approach carefully could be an advantage.

* Similar theme: research on roadworker safety, over half the people interviewed wouldn't expect to see 'people' near to vehicles displaying amber beacons.
http://www.highways.gov.uk/knowledge/pu ... nspicuity/
http://assets.highways.gov.uk/specialis ... mplete.pdf

Drivers do not necessarily expect to see road workers
The participants in this study who were interviewed about this issue showed a varied set of expectancies regarding encountering road workers in everyday driving. More than half the participants reported that they would not expect to see people where a vehicle with flashing lights is stopped at the roadside.

Well the road workers thing is hardly a surprise when you think about the fact I'm shocked every time I see a road worker somewhere in the victinity of any roadworks! :mrgreen:

I had often wondered, wouldn't adjusting speed limits through roadworks be a good idea? Like, have a system where you can travel at 60 through roadwork zones, unless there are workers in that zone, then the speed limit gets reduced to 40. It would probably boost compliance, as I see a fair few people who drive faster through roadworks than they should, because you rarely do see people working on them. This must lead to people thinking "Well there's no point going at 50, there's nobody working here, I'll just keep my foot in".

User avatar
Strangely Brown
Posts: 1018
Joined: Sun Sep 27, 2015 8:06 pm
Location: Sussex

Re: Pedestrian killed by self driving car.

Postby Strangely Brown » Thu Mar 22, 2018 7:09 am

Back on topic...

AZ police have released footage from the car.
https://news.sky.com/video/fatal-autonomous-uber-crash-on-camera-11299527

The thing that is most obvious to me is the appalling use of the headlights. For a human driver, using the lights properly (main beam) would have enabled them to see the pedestrian. For the autonomous driver... are they *really* only using the visible spectrum? Surely the car would have picked up the pedestrian on IR, RADAR or LIDAR?

Not encouraging either way.

ETA: Oh, and it /looks like/ the "safety driver" was more interested in his phone than supervising the car.

User avatar
dvenman
Posts: 261
Joined: Sun Oct 18, 2015 5:28 am

Re: Pedestrian killed by self driving car.

Postby dvenman » Thu Mar 22, 2018 8:43 am

Strangely Brown wrote:The thing that is most obvious to me is the appalling use of the headlights


I wouldn't have had full beam on because of the vehicles in front. And I only have visible light-capable vision systems, so LIDAR / radar / sonar wouldn't have been much use to me either.

And she really does just pop out in front of the vehicle...

User avatar
Strangely Brown
Posts: 1018
Joined: Sun Sep 27, 2015 8:06 pm
Location: Sussex

Re: Pedestrian killed by self driving car.

Postby Strangely Brown » Thu Mar 22, 2018 9:09 am

I agree completely that she just "pops out" of the darkness. From a visible light perspective, she appears to have been in the shadow between two brighter light sources and had very little/zero reflected light from her clothing.

I would have used main beam there because the vehicles ahead look far enough away to not be adversely affected[*], and they can always dip their mirrors if they don't do it automatically.

Apart from the "safety driver" being apparently disengaged from the driving task (a problem that will become endemic as self-driving modes become more mainstream) my concern is that the technology did not spot her. Are manufacturers really relying on visible light only? Madness!


[*] As best I can tell from a grainy dashcam video with necessarily distorted perspective.

User avatar
GTR1400MAN
Posts: 2209
Joined: Fri Apr 29, 2016 12:23 pm

Re: Pedestrian killed by self driving car.

Postby GTR1400MAN » Thu Mar 22, 2018 9:40 am

Massive 'hole' in the street lighting there. With hindsight you can see her shows to the left before anything else.

Doesn't bode well for over here where a huge number of cyclists look just like that from a side view. I'm surprised as in America they are pretty hot on side on reflectors for cars and motorcycles.
Mike Roberts - Now riding a Triumph Explorer XRT. My username comes from my 50K miles on a Kawasaki 1400GTR, after many years on Hondas of various shapes and styles. - https://tinyurl.com/mikerobertsonyoutube

Triquet
Posts: 526
Joined: Tue Sep 29, 2015 7:32 pm
Location: Occupied North Berkshire

Re: Pedestrian killed by self driving car.

Postby Triquet » Thu Mar 22, 2018 10:54 am

It looked just like driving round Oxford.

waremark
Posts: 898
Joined: Fri Oct 02, 2015 9:23 am

Re: Pedestrian killed by self driving car.

Postby waremark » Thu Mar 22, 2018 11:21 am

Uselessly short throw for the dipped headlights. The pedestrian didn't move suddenly nor start to cross just in front of the Uber. Therefore, if able to stop within the distance seen to be clear, it should have been possible to stop. I think even with those lights an alert driver would at least have started to brake. The lady at the wheel was a passenger - at the current state of the autonomous art I think that should be an offence. I am disappointed that the Uber's systems did not detect the crossing lady. But this incident in no way weakens the case for progressing towards autonomous vehicles.

User avatar
Horse
Posts: 3558
Joined: Mon Sep 28, 2015 9:20 am

Re: Pedestrian killed by self driving car.

Postby Horse » Thu Mar 22, 2018 12:21 pm

Interesting that the pedestrian decided to cross 'then' (ie in front of an approaching vehicle) - I wonder whether she was somehow 'impaired'?
Your 'standard' is how you drive alone, not how you drive during a test.


Return to “In The News”

Who is online

Users browsing this forum: No registered users and 19 guests