Sounds like the lady is more at fault than the computer. The article says she wasn't at the crosswalk
That sucks, a guy from my driving club got deported to Arizona from california to work on the self driving cars for Uber so this might come down on him.
He can't come back to California now, he's had a taste of freedom and can't go back to a state that won't let him bring all the guns he has now and he definitely can't take all of them back to England
People jaywalk all the time, self-driving cars are going to have to be able to adapt.
I'm sure whether or not it's found that the woman was the primary cause, however, there's going to be backlash against self-driving cars.
On one hand yeah, on the other if self-driving cars are designed to always stop no matter where they are to objects in the road it would probably end up causing a lot of issues if idiots decided to start walking on the highway or across larger roads on the pretense that the cars have to stop.
It's something that needs extra safeties for sure but the lady is still at fault for breaking the rules.
While Jaywalking might be illegal in the US, these issues need to be sorted out for other countries where it isn't a thing, which I think is most countries.
It's a shame somebody died, but the technology is still in early stages and issues will come up. What concerns me is that the car had a driver behind the wheel which begs the question of whether or not they were paying attention.
She's still at fault to a degree but honestly I'm surprised that they don't stop whenever there's something in the road. That doesn't make sense to me, it's better to have a traffic jam than put the lives of not only pedestrians, but passengers as well at risk.
If I were driving along and hit a jaywalker without even making an attempt to slow down people would very reasonably wonder whether I was alright in the head, and I don't know about the legality of the situation but I wouldn't be surprised if there were some charges to be filed as well.
I still think that self driving cars will be safer than those driven by humans in the long run but if they get the stigma of being unfeeling juggernauts ready to drive down anyone who gets in their way you're going to be facing at best social backlash and at worst actual bans.
There's plenty of situations where you cannot stop safely while driving, that's partly why crosswalks exist in the first place.
Lets be real here, Who is going to walk an extra half mile to one mile just to use a crosswalk? Everyone is going to take the shortest direct route across that road.
Might be shocking now but honestly self-driving cars will never be 100% safe. Now or in 10 years we will probably have occasional deaths like this.
Although in the future there should be way less vehicle accidents overall due to self-driving cars and less human error.
We don't know details but I wonder if this accident could have been avoid outside of the pedestrian not crossing the road where they did.
Self-driving cars have better detection but physics still apply to a moving car, a human driver in the same place would have probably hit her as well.
Then again, there are no details.
Details are released. It was a cyclist.
https://www.abc15.com/news/region-southeast-valley/tempe/tempe-police-investigating-self-driving-uber-car-involved-in-crash-overnight
Yeah, so much for the jaywalking theory. This was the software's fault 100%.
https://files.facepunch.com/forum/upload/774/ac5dfa76-e814-4285-bc09-a87adedcffdb/1521481105053.png
I don't think a human driver would just hit a person walking across the highway. Self driving cars will need to handle situations like that eventually. Even Tesla's autopilot system (not self driving) will attempt to stop or reduce the speed of an impact with a pedestrian. Probably better to have a car go into the back of you than hit a person anyway (which again would be the car behind you breaking the rules by not having a proper following distance).
if there's absolutely no cars in sight then sure maybe, otherwise i'd gladly take half a mile over risking my life.
lady probably jaywalked assuming incoming cars would have to stop for her, and what a shitty assumption that would be to make.
I live in Tempe, and I'm not saying the cyclist was at fault because I don't know the details obviously, but we have a serious problem with people on bikes not following traffic laws.
We have a very large college here with everyone and their mom riding bikes to avoid traffic. I've had to slam my brakes on several occasions because some dumbass rides their bike across the crosswalk when they have a red light
Here's where the impact happened, a bike lane:
https://files.facepunch.com/forum/upload/774/0a6da817-3175-462e-8ced-29b60a187714/1521484343838.png
Looks to me like the software was at fault in this case.
Damn yeah that's probably the fault of the car then (and the operator who failed to stop). This is going to be a huge setback for autonomous vehicles which fucking sucks because they had a huge presence in this city
The interesting thing is that if this is the first fatality it shows how safe self-driving technology is, since the amount of miles in total driven by autonomous vehicles would have resulted in far more fatalities compared to manually driven miles. The people who won't accept something that's clearly better than the alternative unless it's perfect will have a field day with this.
Sucks for the woman though, hopefully if it's a software problem they know what went wrong.
The article says that there was a safety driver behind the wheel so whatever happened occurred too quickly for either the car or the software to react.
This is why we can’t have nice things. But seriously, how many incidents have there been with self driving cars compared to human operated ones? I doubt the rate of accidents by self driving cars is anywhere close to accidents caused by human error if they were compared per capita.
Granted it may be a little early to even have a study like that due to the small size of driverless vehicles currently in use?
The sensors should have a 360 degree view and q decent range so not sure how.
Yeah lets just make all cars stop on every idiot on road, that will insure they dont keep breeding more idiots causing more idiots to walk on road.
i didnt realize some people were accepting of jaywalking... the amount of pedestrians ive come across that just walk right across a 2-3 lane road with traffic RIGHT THERE is seriously dangerous. why is it so difficult to just walk to the crosswalk and wait for it to actually be safe?
What if it's not a jaywalker and instead it's a fallen tree or some shit? If the software can tell the difference then by all means fair game but if I were writing software that could literally kill people I'd probably play it safe when it comes to obstructions detected in the road.
The scariest thing is that this seems like a thing that would of been tested before this car went live. How was something so simple like a bike in a bike lane over looked?
To me, uber seems like a company that would cut corners trying to be the first on the road. Remember they had one of their cars scream through a red light last year (I think it was) as well.
I am not sure who is at fault here in the accident, but one thing is for sure I bet she would've lived if she had been wearing a helmet. Looks like she hit her head on the concrete when the bike got hit, if she was hit by the car at extremely high speeds I don't think she would be directly next to it and still on the bicycle. This is a reminder that if you use bicycles or other forms of transportation where you are exposed it is essential you wear a helmet for reasons such as this as well.
They're up to ~0.4 fatalities per million miles on Uber's car (somewhere around 2-3 million miles travelled). Manual driving clocks in at... ~0.01 per million miles. Statistics for non-fatal accidents probably look different, but you're dead wrong in this case.
How many more fatalities is it going to take before people let go of the kool-aid and realize software is 10 years behind from being able to make a self-driving car work (or if things keep going this way, 30 years)? The hardware is there, but the software end of things is currently implemented in a way which is not only unverifiable and impossible to audit even after an accident, but also fundamentally unstable (LIDAR sensor shortage is partially to blame for this).
yeah that's neat and all but a) I specifically referred to autonomous vehicles as a whole, meaning b) if I chose another company I could get a nice fat 0 fatalities per million miles for fully autonomous vehicles using the same reasoning, and c) comparing statistics when one consists of a single fatality is pointless anyway since that doesn't show a rate, even ignoring precision errors for lower values.
I think it would be best to reserve speculation and rampant opinion-flinging until the facts of the situation are properly understood.
How fast was the car going? How fast was she going? Did either the computer or human attempt to intervene? Was it even possible to avoid her? Was she completely or partially occluded?
These are all deciding factors and we don't know any of them.
Sorry, you need to Log In to post a reply to this thread.
