'Robot Cars Can't Count on Us in an Emergency'


#1

'giving self-driving cars to Google employees for their work commutes
and recording what the passengers did while the autonomous system did
the driving. In-car cameras recorded employees climbing into the back
seat, climbing out of an open car window, and even smooching while the
car was in motion, according to two former Google engineers.

'“We saw stuff that made us a little nervous,” Chris Urmson, a
roboticist who was then head of the project, said at the time. He
later mentioned in a blog post that the company had spotted a number
of “silly” actions, including the driver turning around while the car
was moving.'

We don’t really want to drive, do we?


#2

“face slap combined with eye roll”

These academics don’t understand real people do they?? When you take over a task for humans, they will no longer pay attention AT ALL.

Automotive automation is a step function not a continuum. Levels 3 and 4 don’t really exist and the Stanford study confirms it. There is no “driver takes over in an emergency” or a driver takes over when the car is out of its operational window. They do not exist.

Note that in this article the expert from Ford already understands this;

There won’t BE Level 3 and 4 automation cars. We have 2+ now. The next commercial offering is Level 5


#3

And if a proximity sensor fails resulting in rear ending the car ahead slowing for a turn does Siemens pay for all the damages?

Can we really expect that self driving automobiles will be on expressways any time soon?


#4

I think we can, yes. Robot cars do not have to drive perfectly. They only have to drive better than humans in order to improve the system.

Well, I dunno how they drive where you live, but here in Minnesota they drive like morons. Prairie Home Companion even did a whole skit about adapting auto-drive cars to Minnesota rules, including things like swerving over 3 lanes of traffic at the last second to take an offramp, taking onramps as slow as possible, etc.

A robot car will never intentionally run a red light. That alone will improve safety around here, where red light running is an absolute epidemic.

That’s something for the lawyers to figure out. If I had to guess, once robot cars become more ubiquitous, few of us will own our own car. We’ll just schedule a robot car to pick us up, and whatever’s nearby and available will. It’ll probably be a subscription-based service. Pay $100 a month for access to the car network. At that point collisions will not only be rare, but they’ll also not be the general public’s problem from a fault perspective.


#5

Can I assume that the Robo-cars will be somewhat utilitarian and when operating independently offer very limited performance while able to telepathically link up and fall in line with other Robo-cars on expressways to become 70mph wagon trains? Won’t that be fun! But of course an automobile is just transportation and the public would likey prefer riding in their own private mini Pullman rather than on a bus filled with riff-raff and hoi-polloi like me.


#6

The technology is at it’s infancy. Fully automated cars for general public isn’t expected to be available until 2032 at the earliest. That means we’re not even half way through R&D yet.

BTW - Google already has had a car drive out the parking lot through the town and onto the highway then pull off the highway an into another parking lot and park itself. Total trip was over 100 miles. A human was behind the wheel but didn’t need to intervene at all.

I think the technology is going to be fine, but we have a long ways to go. And when it’s here…the roads will be much safer then they are now.


#7

Respectfully disagree. Yes, they must be better than today’s drivers - a challenge in itself. Even the most moronic driver is still pretty highly functioning compared to a machine system. But they must be as perfect as engineering can make them. Any less is unacceptable.

No automaker can afford reliability less than 99.99999… percent. If properly designed, the system will recognize IF and WHEN that Seimens sensor (or ANY mission critical sensor) fails and safe-stop the car. It must be done this way or any automaker will be sued out of business.

“Well, we are BETTER than the average driver” will NOT be an excuse a jury will accept.


#8

Many years ago I recall reading an “in the future” column somewhere that predicted expressways would have electronic ‘tracs’ that cars would follow and result in reduced accidents and less stress for the driver. It seems that technology has eliminated the need for those ‘tracs’ by way of GPS and recognition of lane striping. I guess I won’t live to see it and that’s just as well.


#9

Some of them will. I suspect it will depend on the tier you choose when you subscribe to the service. Lower tiers will be utilitarian, but you’ll have the option to spend more to get a nicer ride.

Kind of like now - you can take the bus, or a cab, or a limo.

Routine “driving” won’t be fun anymore, but then I can imagine a rancher in 1908 bemoaning the idea that he won’t be able to play on his horse anymore too. I suspect cars are going to be somewhat like horses. We still have horses around. We use them almost exclusively for enjoyment rather than necessary work. No one rides a horse to work anymore. In 50 years no one will drive a car to work anymore either, but cars will still be around as hobby items.

I would guess the speeds would be very much higher than that. Most cars today could comfortably do 100+ if they had proper maintenance and a competent driver, and the guarantee that all the drivers around them are competent. Well, all that’s gonna get solved by automating driving, especially when, as you correctly speculate, all cars are going to be part of a communication network so there won’t be any surprises at all. Wouldn’t surprise me to see 120mph and more on highways.

The disadvantage to buses is that they only come every so often, and they only go to certain places. That’s not such a big deal in places like Chicago where those places are every block, and they come every few minutes.

Here in the Minneapolis area, if I want to go from my outer-ring suburb to the next outer-ring suburb over, I have to take a bus into downtown Minneapolis, then transfer to another bus that goes to the other suburb. I have to spend more than an hour going 35 miles in order to go a net distance of 4 miles, and if I want to make the trip at any time other than rush hour, I can expect to be waiting for each bus for a good while to boot.

And then the bus might stop 6 blocks away from where I really want to go. I’m not anti-exercise, but 6 blocks on a -20 degree February day, or a June severe thunderstorm is pushing my limits.

That’s stupid, and no one in his right mind who can afford a car is going to put up with that. The nice thing about auto-pod-cars is that they will drive to you, when you want them, and they will take you exactly where you want to go rather than 6 blocks away.


#10

By that logic, Airbus should be out of business because its overly-automated systems led to pilot complacency on Air France 447 in 2009, which resulted in a stall and subsequent crash into the ocean, killing everyone on board. More than 200 people died because the machines didn’t work the way they should have.

We have to get away from this notion that 100% safety is the only acceptable outcome. It’s never going to happen. We can continue to work toward it, but it’s not an attainable goal.

I think that’s a given. Where the problems are going to happen is in unanticipated situations, and the designers can’t be expected to anticipate all situations. Or, as my dad used to say, “sometimes s$#@ just happens.”

The old engineering saw says that anytime you build something to be idiotproof, someone will invent a better idiot. People will find new and inventive ways to do something stupid that goes outside of the autopilot’s ability to cope, and bad things will happen as a result. But the overall safety factor will be greater than trusting a bunch of individual apes of varying intelligence and ability to run the system.


#11

By my thoughts, that reinforces my point. Full automation is the only path because the skills a driver (or pilot) has, will wither away without practice. That solution is Level 4 by SAE’s description.

And Airbus is currently being sued over that very crash;


#12

The one that needs suing in that case is Air France, because their pilots became complacent in relying on the autopilot to do their thinking for them and therefore did not have the training and experience necessary to realize that the autopilot had switched operating modes and therefore their inputs would be different.

Not to get too far in the weeds because this isn’t Planetalk, but basically Airbus planes are very fly-by-wire. The autopilot is always on, even when the pilot is flying manually at which point the autopilot judges the pilot’s inputs and makes sure the input won’t result in a dangerous situation. But when sensors fail, the autopilot switches off. The Air France pilots didn’t recognize that a sensor had failed (which they should have because there were 2 divergent airspeed indications that they failed to notice) and therefore assumed the plane would still keep them from doing something dangerous. It didn’t, they did, and it crashed. It was lack of training and improper reaction on the part of the pilots that crashed that plane, not the autopilot.

Full automation must have a backup because when sensors fail the human needs to take over. I agree that the car should pull over and stop, but what if the failed sensor is the obstacle detection system? How does it pull over without running into things? There’s always going to have to be manual override capability - and the tricky bit will be in making sure that people are capable of overriding when necessary.

Air France failed to guarantee that. Hopefully the DMV will learn from that lesson.


#13

But staying very much in business, as many companies whose products or services have resulted in damages: Airbus invented fly-by-wire; Boeing and others have adopted it.


#14

About those horses. We shouldn’t forget that Trigger wouldn’t run off into the Grand Canyon because Roy wasn’t paying attention.


#15

Well, if a snake scared him he just might do something unexpected like that. Similar to the car, the horse has senses(ors) that keep him from doing stupid things. Then life throws some oddball scenario at you…

BTW, why would Roy be distracted?

  1. Texting? Nope.
  2. Girl watching? Not likely in the desert southwest.
  3. Trail racing for horse titles? Umm, no.
  4. Eating take out burgers on the trail? The restaurant by the same name not invented yet.
  5. Trail rage? Sidekick getting on your nerves again?
  6. Saddle sore(s)? Bad guys getting away again? Others?

#16

You guys are hilarious. I can’t remember if it was Toyota or some other automaker that was now working on a flying car again. I hope those robot cars come with trailer hitches to pull the boats and snow mobiles to the cabins. In Minnesota that’s important.


#17

https://www.terrafugia.com/about-terrafugia/


#18

No it was one of the major players dusting off their plans again.


#19

I’m going to play devil’s advocate here. A couple of years ago CBS news had a news item about 2 guys who used a laptop computer to completely take over a 2015 Jeep Cherokee via it’s UConnect system (lucky for me, no one has decided to hack my 2015 Jeep Cherokee; but it’s only been 2 1/2 years so far…). No computer system is 100% secure. Any security software one genius can write can be hacked by another genius. My worst nightmare is the Los Angeles freeway at rush hour and all the cars get hacked. Someone hacks a tanker truck carrying gasoline or propane and turns it into the guardrail causing an explosion and all the following cars are hacked and their brakes disabled, causing a massive pileup.

Bottom line, I don’t EVER want to see self-driving cars become ubiquitous. In my opinion, with all the hackers/terrorists out there it’s only a matter of time before something like this happens. I know that’s a glass half-empty point of view but I think it’s a point of view that should be seriously considered.


#20

Trigger was an actor and knew who filled his oat bag. That’s why he wasn’t going anywhere.