Robocars - Autonomous Drive, Self-driving, Driver-less cars

Plenty of people have been killed by robots - but this is probably the first case of a random death of a random person in public by a semi autonomous robot, as opposed to death of an employee in a workplace. Or a combatant. Etc.

Many people have been killed by semi-autonomous drones. This has often been the intent-- although all too often targets are misread. Real automonomus killer robots, however, are slowing entering the market. Kalashnikov Group, for example, has revealed it has developed a set of fully-automated combat robots. They are not alone.

That said.. I think most of the deaths we will come to associate with self-driving cars, trucks and other autonomous vehicles (such as delivery drones) will be from depression, despair. The current wave of taxi driver suicides in NYC is just the start.. With millions of people employed as drivers....

Yeah hense combatants.

Is there a source / list of all known ‘random people killed by semi autonomous non military robots in public where failure of the robot to act as intended was the primary contributor to the death ’?

Maybe auto pilots. Hmmm this shit is complicated. Where does programming stop and AI / robots begin...

(Although I guess you could probably argue that the deaths, such as they are are, are intentional) (actually thinking about it, does anyone actually get explicitly killed in the Terminator films?)

does anyone actually get explicitly killed in the Terminator films?

Yeah loads. Lots of cops in the first one. And people in the future. At least the foster parent and the bike cop in the second (although the grisly bit is often cut out).

Yeah loads. Lots of cops in the first one. And people in the future. At least the foster parent and the bike cop in the second (although the grisly bit is often cut out).

Don't forget Deckard and his infamous monologue

Ash is a right dickhead though

Oh yeah, I'd forgotten about the parents and the bike cop, still as @Pifko says at least the cop got a pretty memorable monologue to be remembered by

I'd say so

https://youtu.be/qKLTbJMJOSI?t=40s

Totally agree - I find it highly disturbing that the Tempe Police Chief described the accident as 'unavoidable' based on their preliminary investigation and the video. Even back of fag packet maths suggest the ped would have been visible for around 6 seconds - way more than enough time need to react and stop from 40mph, let alone just slow down or change direction. Poor quality dashcam footage released to exculpate Uber by muddying the waters as to the accidents avoidability is highly disingenuous.

This is a monumental fuck up by Uber and no doubt the beginning of some serious arse covering by both them and the Arizona governor for failing their duty of care to citizens for the regulation of self driving testing.

Can't believe this isn't regulated at a federal level. I'll take my chances on drunk, stoned CA pickup drivers over this shit.

NHTSA stats show 1.25 deaths per 100 million vehicle miles in the US.
UBER were trumpeting 1 million autonomous miles driven by September 2017.

Phew.....

That said.. I think most of the deaths we will come to associate with self-driving cars, trucks and other autonomous vehicles (such as delivery drones) will be from depression, despair. The current wave of taxi driver suicides in NYC is just the start.. With millions of people employed as drivers....

How so, through automation? I'm assuming you mean that initially most of these vehicles will need to be 'supervised' by a present driver as such. You see an oncoming wave or a gradual progression?

How so, through automation? I'm assuming you mean that initially most of these vehicles will need to be 'supervised' by a present driver as such. You see an oncoming wave or a gradual progression?

I see the occupation of lorry driver as the first to go into fast decline. The industry employs alone in the UK 2.2 million people - the UK's fifth largest employer. That is a significant labour pool about to slam into a wall. How many of these people have skills or abilities that will allow them to transfer to other forms of similarly renumerated employment? Demand for people to handle the "last mile" will continue for awhile but those jobs are increasingly as sub-contracted "self-employed" with debt as entry ticket (and declining earnings)..

Right now the industry is lacking in drivers. Demand is high but as soon as automation-- and the shortage is just more fuel to the adoption-- takes hold these jobs will go the way of the dodo. If this is not the stuff for a mental health crisis in the making...

And when will the Sword of Damocles fall? I think within the next 5 years.

It's cool, Damocles got out the way ages ago.

https://www.youtube.com/watch?v=iMOD9TYRtr8

I have come to the conclusion that the current model of driver as control backup is only appropriate when the robotic vehicles perform poorly. As the vehicles get better and better at their task it is increasinly infeasible to expect the human driver to take control when needed. Last week's fatal Tesla crash just as the Uber accident are. as I'm increasingly convinced, sympton not of problems with the AI or sensor but of UI/HCI. The human interface is fundamentally flawed as it demands that the driver be alert and cognitively able to take command. We need, I think, to rethink our current approach and develop a new set of rewards for the driver as long as the state of the robotic control is insufficient (up to level 5) to work without one.
Until this is solved or adaquately addressed I argue it is negligent to put out cars with increasingly adavanced ADAS (as the case of the Tesla) or continue to conduct self-driving car tests. As this becomes more and more widespead, without the needed paradigm change, we'll see more and more nasty problems. Will the rate be higher than without AI? I don't know. Even if it is not.. it would be clearly significantly higher than need be...

Agree with this; I'd say dealing with automation is nowadays the biggest single human factors problem in aviation, which is slightly worrying. If highly trained commercial pilots suffer from mode confusion and have trouble responding quickly to automation failures, the chances of your average driver managing it seem slim. If the trade-off is lower KSIs overall, this may be a risk worth taking, but it doesn't seem like on-road performance is anywhere near that yet.

gree with this; I'd say dealing with automation is nowadays the biggest single human factors problem in aviation,

My take is given right now what we need is a lot more data I'd leave off the "comfort" and give the impression that drivers and pilots alike are in full control. If a driver does not really know when they are in control they must assume that they are always in control and must be alert, use all their senses (including tracking all the vehicles on the road, making a number of predictions on the characters driving those vehicles and what their potential next moves might possibily be to put one in harm's way etc.) all all times. The question, of course, arises: "how does one know when in robotic mode to suspend autonomy and handover control to the driver-- who at the instant might have thought they had control but did not?" Since we can easily track the drivers pose and head position as well as their affective state, heart and respiratory rate, I conjuecture that we could probably find some appropriate conditions where autonomy is distrupted and control is passed fully to the driver.
The control driver in this model has nothing to get bored of. In fact, I might even suggest that they are kept under higher cognitive demand than a driver in a normal sedan. As some pilot, resp. cruise, modus proves itself "better than humans" one can then enable them to be deployed without the "guessing game".. so over time more and more features could be "enabled"-- resp. the control game disabled.

I was on holiday last week so missed this. It's looking like the problem with the Uber self-driving car involved in the fatal incident - and in fact all Uber self-driving cars - is that they don't have enough sensors and have significant blind spots:
https://www.reuters.com/article/us-uber-selfdriving-sensors-insight/ubers-use-of-fewer-safety-sensors-prompts-questions-after-arizona-crash-idUSKBN1H337Q

So the whole 'were the sensors working properly' conversation is a bit of a red herring if however well they were working they wouldn't have spotted the pedestrian/cyclist anyway.

Uber cutting costs? No way...

that they don't have enough sensors and have significant blind spots:

Cow droppings.. I would argue that one does not need LIDAR.. and there us a lot of work on new approaches to LIDAR...the article clearly has no idea of the current state of R&D .. there might have been some serious problems with the Uber car but having only one LIDAR on the roof is probably not one of them...

and yet...

Did you actually read the article?

Uber referred questions on the blind spot to Velodyne. Velodyne acknowledged that with the rooftop lidar there is a roughly three meter blind spot around a vehicle, saying that more sensors are necessary.

“If you’re going to avoid pedestrians, you’re going to need to have a side lidar to see those pedestrians and avoid them, especially at night,” Marta Hall, president and chief business development officer at Velodyne, told Reuters.

Velodyne made the sensor on top of the Uber Volvo and are saying themselves that to avoid pedestrians at night you need side lidar.

To risk mixing metaphors if you think it's bullshit it's bullshit straight from the horses mouth.

Rats.

Although it's barely squeaked into the news, they've already settled with some of the victim's family.

https://www.npr.org/sections/thetwo-way/2018/03/29/597850303/uber-reaches-settlement-with-family-of-arizona-woman-killed-by-driverless-car

Irks me to see even NPR are still mis-reporting that dashcam crap "It shows her walking a bicycle across a darkened road moments before the fatal impact."

The quote doesn't seem very relevant to this case. Failure of the lidar to detect someone ahead when travelling at almost 40mph has nothing to do with a 3m blind spot around the car.