Slashdot Mirror
Self-Driving Cars Should Be Legal Because They Pass Safety Tests, Argues Google (theverge.com)
An anonymous reader quotes an article on The Verge: Chris Urmson, director of Google's self-driving car project, has sent a letter to US Transportation Secretary Anthony Foxx today with a plan for selling autonomous vehicles that have no steering wheels or pedals. The plan appears to be pretty straightforward: Urmson argues that if a self-driving car can pass standardized federal safety tests, they should be road-legal. Urmson adds that regulators could 'set conditions that limit use based on safety concerns.'
About three years ago I accidentally let my license expire and thus had to re-take the driving component of the exam.
I am somewhat convinced you could pass it with a non-autonomous vehicle having no steering wheels or pedals.
We can't write a 100% working OS for a phone. Please trust our software with your life.
..and lots of them have been proven later to be unsafe anyway. The law cannot account for everything.
We should create a pool for when the first self driving car will be turned into a turtle.
Driving down the road... then suddenly, Skynet determines an alternate route for you
+1
As long as Google is willing to take responsibility for the damages, I'm fine with it. Basically, Google is the driver of all driverless cars they sell.
Any ordinary car driven by a raging retarded monkey would pass the safety tests as well.
BECAUSE THE SAFETY TESTS ON CARS DOESN'T TEST DRIVING OR COGNITIVE SKILLS!!!!
Here's my thing is that tons of people die from drunk drivers and random physical ailments which then turn to an accident. I would argue that self driving cars would allow those people who would drive drunk or those slow elderly to get around safer. The down side is that with the nice security features built(Not) into cars, hackers could always have a self driving car go off a cliff.
No steering or pedals? Talk about snatching defeat from the jaws of victory!
Urmson argues that if a self-driving car can pass standardized federal safety tests, they should be road-legal.
Umm, yes, seems reasonable...
Sorry, is there any actual story here? It's practically tautological. Of course there should be some kind of safety test for self-driving cars before they're allowed on the roads for any reason other than testing. Was anyone expecting anything else?
systemd is Roko's Basilisk.
There are plenty of computers in use (a lot of the better ones are running Linux or an RTOS and hell, even Windows NT/CE/XP) that people trust their lives to implicitly on a daily basis in a lot more delicate situations than driving a car. Commercial planes do most of the flying fully autonomous, most of both your debt and savings is being invested fully automated, any machine in a hospital parses a lot more data than a few dozen sensor and requires much more precision.
Custom electronics and digital signage for your business: www.evcircuits.com
Yes, and by "responsibility", that means that its executives and/or shareholders (fuck the anti-capitalist limited liability that shareholders enjoy) go to jail for dangerous driving / local equivalent if the software fucks up. None of your slap-on-the-wrist fines bullshit.
If those who own and run the company are cool with making themselves effectively the driver of hundreds of millions of cars every day, I'm cool with "driverless" cars.
As long as Google is willing to take responsibility for the damages, I'm fine with it. Basically, Google is the driver of all driverless cars they sell.
No, the owner of the car should get regular car insurance, and then the insurance company will take up the responsibility for the damages. Of course, if the insurance company can argue that google has been grossly negligent in some case, they can take them to court.
There are plenty of computers in use (a lot of the better ones are running Linux or an RTOS and hell, even Windows NT/CE/XP) that people trust their lives to implicitly on a daily basis in a lot more delicate situations than driving a car. Commercial planes do most of the flying fully autonomous, most of both your debt and savings is being invested fully automated, any machine in a hospital parses a lot more data than a few dozen sensor and requires much more precision.
Have you seen a commercial plane without human pilots ? I thought so.
Driverless cars will never ever come on the road. Unless you somehow want a driverless tram-like system.
But I don't see an AI coping with unexpected events in a timely fashion ever. Google and others like them are trying to sell a pie in the sky dream. Beware.
robot cars will take over the taxi driver profession, including uber, and if the price of a ride comes down low enough it will put a big dent in busses and subway customer base
Politics is Treachery, Religion is Brainwashing
Have you seen a commercial plane without human pilots ? I thought so.
That's not the point. The computers on board a commercial plane have the potential to cause major accidents that the pilots would be unable to prevent. And still we trust them.
Exactly. Let the people whose job it is to calculate risk do exactly that. If they determine that a self-driving car is more accident prone, your insurance will cost more than usual. If they determine a self-driving car is safer, it will cost less than normal.
My guess is that within 5 years of self-driving cars being on the road, there will be substantial insurance discounts for owning one.
flight is actually a much simpler problem to solve for an AI than ground travel. Planes don't typically have to avoid unexpected obstacles because their vectors are carefully monitored and controlled by human pilots in the air and on the ground. So while the speeds and distances are much greater, the path to destination is much simpler (even if elliptical).
Machines are quicker, yes, but a lot dumber and lack situational awareness. A medical machine monitoring vitals can notice changes a lot more quickly and attentively than a human, but you still want a dr to monitor the patient's overall state and administer drugs manually. Also, overengineering leads to increased failure rates and expense, and a lot of tech today suffers from this to the point of being consumer hostile.
There are plenty of computers in use (a lot of the better ones are running Linux or an RTOS and hell, even Windows NT/CE/XP) that people trust their lives to implicitly on a daily basis in a lot more delicate situations than driving a car. Commercial planes do most of the flying fully autonomous, most of both your debt and savings is being invested fully automated, any machine in a hospital parses a lot more data than a few dozen sensor and requires much more precision.
Driving is a far more difficult problem than auto landing, auto pilot and auto takeoff on an airplane.
So if one vendor's software passes a driving test let it also share all the driver's license "points" accumulated by all the autonomous vehicles. So if it makes too many mistakes or gets into too many accidents it looses its license. Again, not an individual car, all cars running the vendor's software.
Would you be happier in you car while surrounded by autonomous cars, or by cars driven by drunks tweeting while shaving? Because the latter is what you have now.
How will a computer respond to a tire blow out on the highway at 60mph plus? Some other emergency?
How will the car that gets slammed into by an autonomous vehicle with a blown tire respond?
How will the cars behind it react to the event in front? How fast?
What will happen to the hacker that intercepted and manipulated those signals the other cars are sending to each other? ( assumption made )
Has any of these scenarios been tested? I don't see any crushed google cars so I am going to guess NONE.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
- would I feel OK sitting in one of those and everything around me moves and I cannot act at all, sitting there passively?
I think I would worry myself to death.
Cruse Control fine, maybe automatic steering but steering wheel disappearing, seats swerving 90 degree away from driving direction towards each other, as it was written about the BMW "experiment" and semis overtaken by software passing by right before my eyes - no mf. way ...
Machines are quicker, yes, but a lot dumber and lack situational awareness
On the other hand, machines are capable of watching dozens of different sensors and cameras at once, in all directions around the car, with much higher precision, and without getting distracted or sleepy. What they are lacking right now is human-like interpretation of what they see, but that's a field that is rapidly improving.
it's very much the point, because goog's argument in the summary is that the us shoudl legalize automated cars with no steering wheels or brakes.
pilots land and take off manually.
That's not a very good strawman; it makes you look stupid.
Exactly. Let the people whose job it is to calculate risk do exactly that. If they determine that a self-driving car is more accident prone, your insurance will cost more than usual. If they determine a self-driving car is safer, it will cost less than normal.
My guess is that within 5 years of self-driving cars being on the road, there will be substantial insurance discounts for owning one.
Jesus, how deluded can one be ?
Now, I can just about grasp that a self driving car can be constructed that will navigate on the road, but that is not all that a car has to do. Let's look at a couple of examples:
1) Suppose I live on a small farm or ranch and you are coming to visit me in your car. I might say "When you get here, come up the drive, turn left at the old tractor and park behind the barn next to the chickens". With a conventional car this should be easy, but what if you have one of these Google cars with no controls. Presumably it will find my address and arrive at the end of the drive. Given that there are no manual controls, how would you tell it the bit about the tractor and chickens? Will you just be able to type that in and it will be clever enough to follow those instructions?
2) What about parking at work? I work on a big site with several car parks. How will I describe to the car which one I want to park in. They don't have separate Zip codes.
How will a computer respond to a tire blow out on the highway at 60mph plus?
Within one millisecond after the blow out, the computer will get data from a tire pressure sensor that indicates what happened, and which tire blew out, and take the appropriate action. They'll be in a much better position to handle this gracefully compared to a human driver that has never experienced this, and maybe requires a few seconds to realize what the hell is going on, and has no clue how to react safely.
Has any of these scenarios been tested? I don't see any crushed google cars so I am going to guess NONE.
Why would you assume Google would notify you of any experiments they have done with blown out tires ? And if they did it right, there wouldn't be any crushed cars.
Have you seen a commercial plane without human pilots ? I thought so.
That's not the point. The computers on board a commercial plane have the potential to cause major accidents that the pilots would be unable to prevent. And still we trust them.
THAT is exactly the point. Human pilots are needed on airplanes because they are the ultimate authority and can override the system if the situation calls for it. A diverless car without any kind of overiding system enabling a human to take control (for whatever reason) is like a fully autonomous airplane. You will never see them.
The only use cases where driverless cars make sense is :
- a system where these cars are constrained to a set of specific roads (with no other kind of cars, camions etc...).
- a system that has eliminated all normal cars/trucks etc.. from the streets (do you see that happening ?)
outside of these 2 use cases, we will have driverless cars that have to share the road with cars driven by humans. And in that case you fucking want the ability for a human driver to override an AI if the need arises. AI can't take into account all possible modes of failure or unexpected events.
Spoken like someone who hasn't the faintest clue about the differences between an autopilot and a self-driving car. In other words, an idiot (don't feel too bad; that's very much the norm these days).
The point of GP was that computers can't be trusted with life or death decisions, because even our phones have bugs. To which somebody else responded that we already trust computers with our life, such as those in airplanes or medical equipment. The fact a modern plane still has human pilots does not negate that, as the humans can't override every single computer action.
and take the appropriate action
And tell me how you would code that? Tell me how a coder that may or may not have experienced a blowout would approach that problem ?
I have, and it's immediately obvious why you're getting the snot shook out of you while you bring the car to a halt without making it worse. Think about the visual problems that would occur from the shaking and vibrations of that event?
and take the appropriate action
I am sorry, but I really think you don't have a clue.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Tell me how a coder that may or may not have experienced a blowout would approach that problem ?
Drive a car along an empty road, and blow out the tire. Collect data, code algorithm, run tests, and repeat until you get it right.
Should every self-driving car pass the same test humans do?
What will happen if everyone has self-driving cars, and our GPS system goes out?
Who will be the first one to die in a driverless car?
So you don't enjoy driving?
How do motorcycles fit in in a world of self-driving cars?
How many people will be made unemployed by self-driving cars, buses, and trucks?
https://www.youtube.com/c/BrendaEM
"So if it makes too many mistakes or gets into too many accidents it looses its license."
You now have one point on your spelling license.
BeauHD. Worst editor since kdawson.
Tell me how a coder that may or may not have experienced a blowout would approach that problem ?
Drive a car along an empty road, and blow out the tire. Collect data, code algorithm, run tests, and repeat until you get it right.
Lol, now put the driverless car on a highway with thousands of other cars, fog, rain, etc... and let the fun begin.
Think about the visual problems that would occur from the shaking and vibrations of that event?
Really, those are the easy problems. A much harder problem would be to recognize whether a pedestrian is signalling you to stop or urging you to move on, or anticipating what a cyclists is going to do based on where he's looking.
*Chuckle*
This right here. ;)
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
(don't feel too bad; that's very much the norm these days)
As you aptly demonstrate.
... until somebody loads up a few car bombs and sends them to town.
code algorithm
An algorithm? Just one? I can tell you have never experienced a blow out that's for sure.
Lets crank it up a notch. The tire didn't blow out, but the tread separated and wrapped itself round some structure of the undercarriage and locks up the rear end. On asphalt pavement with a ton of road grit and gravel.
Then there's this other post:
Lol, now put the driverless car on a highway with thousands of other cars, fog, rain, etc... and let the fun begin.
Put those cars on the rad and let the slaughter begin.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Assuming that they are safer than the average driver, it is almost a certainty. Most car insurance companies are mutual companies, which means they aren't trying to make a profit. If those vehicles are actually safer, those insurance companies will quickly add discounts to encourage further adoption so that everyone's rates will eventually go down. And any non-mutual insurance companies will be forced to follow suit if they want to remain competitive.
Check out my sci-fi/humor trilogy at PatriotsBooks.
An algorithm? Just one?
You must not know what an algorithm is. Hint: 1 algorithm + 1 algorithm = 1 algorithm.
I can tell you have never experienced a blow out that's for sure.
That's right. And I probably wouldn't handle it correctly, even though I'm human.
lol, now put the driverless car on a highway with thousands of other cars, fog, rain, etc... and let the fun begin.
Highway driving is the simplest of all. The car can see through fog and rain using radar and infrared, and can accurately measure the speed of all the cars using doppler, measure the slickness of the road, and accurately determine safe speed and driving distances. These extreme circumstances are actually much harder to handle for humans, as you can clearly see every time there's heavy snowfall, and cars just pile up left and right.
Oh, I love this game! The Luddites bring it up every time autonomous cars are discussed.
How will a computer respond to a tire blow out on the highway at 60mph plus? Some other emergency?
Within a few milliseconds of the emergency being detected by sensors, the computer will have fully assessed the situation and determined the safest course of action. A blown tire is simple, because it only really affects the vehicle handling parameters. At over 60MPH on a highway, the vehicle is going to have very minimal handling needs. The steering system can be told (within those milliseconds) that it will need to adjust, and in a few rotations of the tire, it can analyse the shape of effects of the new tire's shape. At minimum, it will know that it needs to steer a few degrees to the side of its intended course, allowing it to stay on course and maneuver safely to the shoulder.
How will the car that gets slammed into by an autonomous vehicle with a blown tire respond?
If by some absurd accident that does occur, it'd be treated like any other unavoided collision. As soon as the vehicle determines that a collision is unavoidable, it will attempt to minimize the damage. There has been research into having algorithms adjust the vehicle speed to change impact position, relying on a database of the vehicle's crush characteristics to reduce the chance of injury. Results show that the computer can do that faster and more successfully than any human driver.
How will the cars behind it react to the event in front? How fast?
When the front car detects the blowout, or when the cars behind it detect the debris or change in driving characteristics, they will consider the car to be a risk, and avoid it. They will start slowing down, changing lanes, and otherwise avoiding the affected vehicle. Again, this has all been tested.
As for the speed of this decision, everything happens in a few dozen milliseconds. At 60MPH highway speeds, that means that the computer will process and understand a situation before the vehicle has traveled a few feet. In comparison, a human brain reacts in about 150 to 250 ms, depending mostly on the type of stimulus. It doesn't matter how good of a driver you are, or how much attention you're paying to what's going on. An autonomous vehicle can observe, consider, and begin reacting to an emergency in front of it before your brain can even understand what your eyes are seeing.
What will happen to the hacker that intercepted and manipulated those signals the other cars are sending to each other? ( assumption made )
Oh, don't be so coy. You've made an awful lot of assumptions without actually understanding the current state of the art.
Legally, probably nothing will happen to the hacker, because it'd be difficult to find and catch such an attack, but that has nothing to do with the cars. It's equally hard to find someone using a cell phone jammer today. There have been a few cases, but they were caught due to prolonged or repeated use.
From the perspective of the cars on the road, losing communications with the other vehicles is a known and tested risk. Similarly, mismatched information is a risk condition. The easiest response is to slow down and try to move out of traffic.
Has any of these scenarios been tested? I don't see any crushed google cars so I am going to guess NONE.
Yes, these scenarios have been tested extensively, but you don't know about it, because you're not bothering to do research. One of the fascinating aspects of robotics research is that researchers can control all of the inputs to the algorithms. We can put a sensor in a tire, then drive it down a test track and make it blow. We can take that data, mix it with data from a dozen other test runs, and use that to build thousands of simulation input cases. Those inputs are run in simulated environments, with and without vehicle-to-vehicle
You do not have a moral or legal right to do absolutely anything you want.
Only when they can pass the written test and okay the driving instructors commands during the drivers test.
If a self driving car will have accident with ordinary driver car ... -what will be "initiator" of that accident? Of course that one who have cheaper advocate... -- But who has usually cheaper advocate ??? - Car owner person, - not a car owner multi billion owner firm! That's explain, why so much efforts are spent to advertise thing, that will destroy yet in another way people trust in anything beyond sacrifice...
The thing is that a third party would actually be responsible for the accident, and as such you would need additional coverage specifically for the case. Insurance policies spell out such things. They want to even know how much you drive it on average, where you drive and in what conditions. They take into account things like if it is a motor cycle all the way up to a tractor trailer. The price of insurance is based on these factors.
Now add the concept of hardware driving the car. Sure, we can demonstrate that this car with this software is this safe, but then comes a software update. This makes the insurance company's management of their risk much more difficult. All it takes is one bad update and lets say, hypothetically, a vehicle or two careens into the side of a bus. Some analysis and a patch later and all of a sudden we are at a new version without the bus smashing feature, but did that update add a toddler avoidance error?
AI "drivers" receiving software updates only add to the chaos that is our roads. No one has died yet, but someday someone will.
Sure, we can demonstrate that this car with this software is this safe, but then comes a software update. This makes the insurance company's management of their risk much more difficult. All it takes is one bad update and lets say, hypothetically, a vehicle or two careens into the side of a bus.
Insurance companies are experts at calculating risk. This type of risk won't be any different from the thousands of other types they already have to worry about. If there's X% additional insecurity, they just add Y% extra premium. Furthermore, software updates can be installed gradually over an extended period, and the insurance company can keep track of numbers and types of incidents for all the different software versions. As soon as there's a pattern that a certain version is less safe than others, the updates can be stopped and rolled back.
Look up TCAS sometime. The planes have sensors to detect each other. If the TCAS system detects a possible collision situation, the planes determine, all by themselves, the correct course of action, and then relay that information to the pilot. Commands like CLIMB or DESCEND or STAY LEVEL. In this situation, the pilot has absolutely no say in the matter. They are required to obey the computer because in the past, pilots ignoring this input have cause planes to crash into each other in mid-air because the pilot thought he knew better. The TCAS commands even override Air Traffic Control commands. How's that for trusting your life to a computer?
"Freedom in the USA is not the ability to do what you want. It is the ability to stop others from doing what THEY want"
Your comparison is a bit apples and oranges. How may people have stepped out in front of an air plane in flight? The systems for avoidance here are a bit more advanced. Aircraft typically know about collision threats miles away. Human traffic controllers manage this avoidance. Are we going to fit all the things that might get into the path of the car with transponders? Automating the piloting of a motor vehicle safely is far more complex than flight.
Unless the weather isn't conducive to VFR, in which case, guess what? The autopilot lands the plane. In situations where the pilot cannot be trusted to land due to poor visibility or other issues, the autopilot is king. I've taken an A320 class D flight simulator from ATL to JFK without touching the controls except to set the throttles to the climb position from TOGA, setting the flaps, and landing gear position.
"Freedom in the USA is not the ability to do what you want. It is the ability to stop others from doing what THEY want"
An algorithm? Just one?
Yes, just one algorithm. As slashping points out, it's the combination of other smaller algorithms:
All of those parts get bundled together into a single unit for designers to track, and it'll probably be documented somewhere as the "tire blowout algorithm".
Lets crank it up a notch. The tire didn't blow out, but the tread separated and wrapped itself round some structure of the undercarriage and locks up the rear end. On asphalt pavement with a ton of road grit and gravel.
Lol, now put the driverless car on a highway with thousands of other cars, fog, rain, etc... and let the fun begin.
Put those cars on the r[o]ad and let the slaughter begin.
Ok. Where's the hard part? What you're describing is a worst-case situation, and most of it doesn't matter. Fog and rain only affect one out of several sensors, and a road with "a ton of road grit and gravel" isn't unusual. That's expected for a road. Every time the software checks its own state, it considers the road condition. If it doesn't have "a ton of road grit and gravel", then the vehicle doesn't need to compensate for that.
What is being built is a perfect super-human driver. It has access to senses that humans don't, never gets distracted, never panics, and never deviates from best practices. It has spent several decades of (sim) time in simulations, dealing with hundreds of thousands of hazards, and now you're worried about rain?
You do not have a moral or legal right to do absolutely anything you want.
What, AC now looks stupid? What is this place coming to?
The argument was in fact more complicated than your simple "understanding" of it but don't feel bad; confusion often tend to be frustrating and I dont blame you for letting it get to you.
and now you're worried about rain?
Nice try, that method of argument will not be accepted.
Decades of road simulation? Perhaps with functional programming regarding rules and 'best practices' according to the rule of law and making sure the car operates on the road as expected in ideal conditions. But there's no way to simulate the real world failures like you're asserting has been done. And certainly not for decades.
Pardon the pun, but the real work will be where the rubber meets the road.
The good news is natural selection will get it's chance for anyone that wants to trust their lives to what is essentially a robot and we won't have to put up with their kind too much longer.
By all means engage in self destructive behavior, but don't take out the rest of us in the process please.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
All true, but how much damage can be done before a pattern is established? I think that such insurance will be expensive at the outset. I think having manual controls in the event of a problem will reduce the cost of that insurance. I also hope that they avoid costly mistakes that could make this technology common in the distant future.
Still, while there are techniques to make releasing software as safe as it can be companies still find the need every year to perform a live network change. The issue might not exhibit itself in a small initial release, but be devastating (read kills people) once it reaches a larger audience. For example, the software works great in Sunnyvale, but in icy conditions in northern states cars spin out their traction control failing to account for it.
Furthermore, what about some malware attack that exploits a newly added hole. The insidious rogue code waits for a period of time to propagate using vehicle to vehicle communication then one day goes on a murderous rampage at noon on a Tuesday?
Everyone was out to lunch to get some wonderful tacos, but instead... vehicular armageddon ensues!
Well, that is a bit far out there, but I still don't envy the position of the insurance companies. Having a licensed driver at the wheel and override controls at the ready can mitigate much risk. I wouldn't feel comfortable in a car that lacked manual controls. I would also prefer that early adopters of this new technology be made to remain at attention and ready to take over or face a hefty citation. At least for five years or so until the technology had proven itself. If no death or serious damage to property occurs within such a time we could re-evaluate the idea that someone in the driver's seat must remain at attention. Removing override controls should take much longer.
If I get hit by an automated car, I sure hope my insurance agency sues Google. Furthermore, if I have an automated car and it gets into an accident it's not my fault so I'm not paying insurance for it, Google can pay the bill.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Yes, this is real life, not your if-it-ain't-broke-don't-fix-it fantasy.
You can choose to ignore what I've written, as you've chosen to ignore the other research and progress in the field. Looking through your comment history, you seem to do that a lot. You also seem to have a bias against modern automotive safety, based on myths that have been continually disproven.
Yes, it's true that there are faults to modern technology, especially with regards to safety equipment. However, you don't seem to understand that the risk of new technology is far less than the risk presented by older technology. A driver in a collision who has an airbag may break a collarbone, but he won't have frontal-lobe trauma as he would if he were allowed to hit the steering wheel.
These things can (and are) tested. What's important is that the circumstances of the test be identical, so the difference in effectiveness can be measured. One good way to do that for vehicles is to take an old vehicle, and crash it into a new vehicle. That way, thanks to Newton's third law, we know the impact force is exactly identical. In fact, here's a good demonstration.
You do not have a moral or legal right to do absolutely anything you want.
No, the owner of the car should get regular car insurance, and then the insurance company will take up the responsibility for the damages. Of course, if the insurance company can argue that google has been grossly negligent in some case, they can take them to court.
If these cars are as good as they are claimed to be, no liability insurance is needed.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Decades of road simulation? Perhaps with functional programming regarding rules and 'best practices' according to the rule of law and making sure the car operates on the road as expected in ideal conditions. But there's no way to simulate the real world failures like you're asserting has been done. And certainly not for decades.
You clearly have no idea what the state of the art actually is, where I've actually built such simulators. We've been building them for several decades (of real time) already. Yes, some algorithms (mostly for communications and basic sensors, including tire pressure) have had simulated time exceeding a hundred thousand hours, which is over a decade of continuous real testing.
You'd be amazed what can be simulated these days. One of my colleagues works on simulators for a tire company, modeling the wear pattern and friction for their tires, to maximize the tire's handling and hydroplaning resistance throughout its life. His simulator evaluates different tread patterns and different rubber formulations against different driving styles and conditions. Fascinating stuff, and the only physical result is the confirmation tests, when the real thing is actually manufactured and driven for a few months.
You do not have a moral or legal right to do absolutely anything you want.
Don't forget to ask 'em about privacy, about ubiquity - not all of us live in areas where the maps are accurate - or even remotely accurate. (Remember my story about the bus(es) full of Canadian foliage tourists that show up in my driveway every year?) I modeled traffic for a living - well enough so that my business has long since been sold and I'm happily retired. I'm a driving enthusiast - someone who's taken many, many advanced driving courses and even done things like take special classes, rent exotics, and then spend a week or so doing laps at Nurburgring. Yeah... I'd like to think I've got a qualified opinion.
What's awesome is when they say how many miles the cars have driven without accidents. Sure, how many times have humans intervened and how much of that was on roads that were unfamiliar? How much of that was in less than optimal situations?
However, my big concern is privacy but nobody gets that part. Ah well... It's amusing that they rate you as the one making assumptions about the state of the art. I've followed the state of the art and it'll get there, someday. Yay for having no more autonomy of self and having limits to where you can go as well as being monitored for going there. Enjoy your hike if you want privacy.
I don't mind. I'll be dead and gone before it's ubiquitous or mandated. Even if it gets more expensive to drive, I've got a few bucks. Sorry for the rest of you but I can do no more than what I'm doing. This can not be done, nor will it, without huge privacy concerns. Fuck that. And before someone says that I can't drive without being seen, I'd like to point out that I can go from my home to Canada without even crossing a routed highway. Err... It's not speedy or legal - but I can.
"So long and thanks for all the fish."
We have our suicide hill --- on the map a direct and plausible route south, but with a steep S curve which even the locals occasionally fail to navigate.
Come Labor Day and the VFW and our rural volunteer firemen have beer tents set up in a small town parks, with temporary parking on the grass, while minor road work elsewhere have traffic being routed off onto the shoulder. In theory, every hazard should be properly signed and flagged and posted to the web. In practice, it doesn't always work out that way. The point being that your self-driving car can't be reliably programmed to deal with every possible contingency.
They try to cover a representative subsample of ever possible situation. They're based on the premise that if you're cognizant enough to figure out what to do in those situations, you're cognizant enough to figure out the common sense thing to do in other situations which aren't being tested.
A machine built to only pass the situations being tested should automatically fail. The actual "rules" of the road is the State driving code, which is usually several hundred pages long and probably filled with situationally contradictory requirements (e.g. if stopped at a 4-way stop sign and you're on the right, you have the right of way. Unless the other guy goes first in which case he has the right of way).
LOL I'm glad I don't drink alcohol any more. I didn't quite spit it up but I *did* have to swallow awkwardly.
Hmm... That sentence should not be taken out of context. It was just coffee. There's a whole lot of really strange optimism with this. People seem to think they're going to be able to get drunk, hop in the back, and say, "Home, James." They seem to think this is right around the corner - and that it will be done while respecting privacy. Oh my... They've never seen the big picture that is traffic and, trust me, one can not really model it (well) in their head - there are too many variables.
What they're getting is advanced lane maintenance and adaptive cruise control with automated braking and stability control. Half of 'em seem to think they're getting Star Trek teleporting machines. It takes longer to model your commute, on some really big iron, than it does for you to make that commute. I know... Obviously we don't need to go to those extremes but the closer one can get to that then the closer one can get to having a utopia (sans privacy). Some people seem to believe we're almost there.
I envision kids in the back my car. "Are we there yet, Dad?" My standard reply was, "Has the car stopped and I said it's time to get out?" Fortunately, my children are grown, health, and intelligent. As I look around, I feel like I've been blessed with good fortune. Thanks, FSM. I'm glad my children aren't retarded.
"So long and thanks for all the fish."
How will a computer respond to a tire blow out on the highway at 60mph plus?
Faster and safer than a human. In fact why isn't this test part of getting a drivers licence. After all How do humans respond to this?
How will the car that gets slammed into by an autonomous vehicle with a blown tire respond?
Faster and safer than a human. In fact why isn't this test part of getting a drivers licence. After all How do humans respond to this?
How will the cars behind it react to the event in front? How fast?
Faster and safer than a human. In fact why isn't this test part of getting a drivers licence. After all How do humans respond to this?
/. , home of the "tec geeks" with a massive Luddite complex. Fact is autonomous cars will be a WAY better driver than you will EVER be.
You get the idea. And as for a hacker. Well how many ABS systems have been hacked? Yea you can ALREADY hack a modern car. Yea it is not killing people.
Ahh
Have they tested these cars in situations that were not the most ideal?
- going around traffic cones
- driving on the shoulder when the road is obstructed
- heavy rains/winds
- snow
Unless it has mastered all of those things, then it has no business being on the road. That is, unless Google wants to be sued out of existence.
not all of us live in areas where the maps are accurate - or even remotely accurate.
Maps are just one sensor. The algorithms work well enough without them, though you may need to tell the car exactly what route to follow.
I'm a driving enthusiast... I'd like to think I've got a qualified opinion.
Your opinion may indeed be qualified, or it may not be. Experience and understanding are related, but distinct.
What's awesome is when they say how many miles the cars have driven without accidents. Sure, how many times have humans intervened and how much of that was on roads that were unfamiliar? How much of that was in less than optimal situations?
Actually, a lot of it is in less-than-optimal situations, with very little human intervention. There are controlled and uncontrolled tests. Google and the media make a big deal about the uncontrolled tests, where a vehicle is out on the roads in public areas, but those aren't really the most useful for refining the software. In controlled tests, engineers throw everything at the vehicle, and any time the human has to assist, the software goes back for redesign.
However, my big concern is privacy but nobody gets that part.
Actually, there is research being done into privacy aspects of automated cars, but it is still a minority. Sadly, concern about privacy is a minority concern pretty much everywhere outside of echo-chambers like Slashdot.
One of the promising areas for vehicular privacy is not really privacy, but anonymity. It's not the idea of hiding what you're doing, but rather hiding who's doing it. Your car may announce its presence with a unique identifier, but that identifier can (in research tests, though I don't know of any real-world implementation) change at every turn. Someone intending to follow your travels might know that your car left your house and entered a particular road, but then a completely different vehicle announced it was on that road a few seconds later. For a few seconds (where nothing of interest happened, as far as your vehicle's concerned), there is uncertainty about your car's identifier, and that makes tracking your movement far more difficult.
Movement can be extrapolated, but there's no requirement (yet) that a vehicle announce every move. Such announcements are a courtesy to other vehicles, but those vehicles have their own sensors, and don't really need the announcement.
You do not have a moral or legal right to do absolutely anything you want.
Folks don't normally go to jail for auto-accidents, unless they were intentional or they were drunk or something. Since none of these applies to autonomous cars, why would anyone go to jail for a legitimate accident? (e.g. ambiguous situation where the autonomous car couldn't stop in time due to weird road condition or something, etc.).
Chances are, you're way safer in one of these cars than you are behind the wheel.
I completely agree that all of the questions posed by "luddites" are fairly inconsequential; every one of them has a specific, detailed answer that can be found within 30 seconds of googling (heh!) by someone without any inside knowledge of the state of the art. Your answers cover most of them.
Much more interesting to me is that Google is developing the two types of vehicle, a modified sedan (which one assume could be extrapolated to SUVs, minivans, etc.) with full regular controls, and the "bubble" car that has none (except perhaps an emergency stop/pullover button?)
The Bubbles would be purchased by customers with large fleets: Uber, taxi companies, bus/transit authorities, or Google itself, and these cars wouldn't NEED parking. They drop you and move along to the next customer. They could even avoid highways if that's preferred, meaning serious injury at high speeds is already extremely unlikely, even though it's the easiest 'mode' of driving.
The "regular" autonomous cars would/could be purchased by consumers, who could still be taking over in situations like backroads, unmarked areas, etc.
But the whole POINT of autonomous vehicles in the long run is to not require a driver to be present or aware. We should be able to read in the vehicle, work, sleep, talk, etc, without any requirement to pay attention to the road.
What I'd like to know, is more on the human side: what level of integration is going to be required on a legal/city planning front? Will there be pilot cities with useable google fleets? Subscription model? Will autonomous vehicles need or prefer "AV lanes" in the city? Should we change other traffic laws for cyclists & pedestrians, and in what ways? Is a federal legal framework even required to do this on a city-wide or provincial or territorial level? (you Southerners can keep you States for yourselves!)
What will the pushback be from the transport unions (both goods transport & public sector workers?) Will "driver" realistically not be a job any more at all? Or will that sector be reduced by 80%-90%, with some jobs being kept? How do we encourage development while transitioning jobs? What new jobs will it develop? etc.
I think having manual controls will increase the cost of insurance. You simply can't have an "autonomous" car that has the capability to hand-off to a real driver if it "gets tough". There's no point in it. It's hard enough paying attention when you really need to, but when 98% of the time it doesn't matter, but those 2% remaining you have to be at full attention? Good luck with that.
Can you be Even More Awesome?!
So tell me how does a human driver that never experienced a blowout (most drivers) respond? Most drivers are very bad at realizing what happened in the first 1-3 seconds and simply panic by slamming the brakes, swirving etc.
In a car you can at least code an appropriate response, cars don't care if it's foggy and already have countermeasures for wet roads or loss of traction.
Custom electronics and digital signage for your business: www.evcircuits.com
Everyone keeps saying you CAN do this or that. And nobody has linked to any real research, done in the real world. Specifically related to vehicle failures.
Also, if the car encounters a condition it's not programmed to deal with, what will be it's default response?
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
This is not true, according to the Wikipedia article. The only thing the pilot is absolutely required to do is not manouver in the direction opposite to the one indicated by the TCAS. As for actually following what the TCAS says, there is some leeway.
entropy happens
Rolls eyes...self-driving coffins...thanks but no thanks.
These self driving cars will die by themselves. Lack of interest of the general public, plus real world conditions will ensure nobody (sane, fanboys don't count) would want those near death experience on tires.
You can't see, can you? this super human driver that has none of the shortcomings of the humans, has all the shortcomings of the computer (actually it'll have several of the humans since a meatpuppet will do the coding, pray it isn't outsourced to save some bucks)... I trust computers for many things but this is not one of 'em, and never will. The human brain is better suited for this kind of problem than a cpu, someday maybe but we're still far, very far away from it.
TCAS is a last ditch system to avoid a collision by proposing resolutions to situations that shouldn't happen in the first place but sometimes do. The biggest feature of TCAS is that the devices talk to each other and so if you have two planes on a collision course one will get a climb and the other a descend (or whatever) and they both won't take the exact same evasive action at the same time, which is kind of useless.
TCAS is an engineered system designed by engineers using purpose built equipment and it does one thing and does that one thing rather well. The reason ignoring TCAS can be deadly isn't because TCAS is so perfect, it's because you have a pretty good idea that the pilot of the other plane is doing something opposite of what you're about to do, which increases your odds of survival. It's not magic, it's not a full air traffic control system or even close to one, and above all it's reliable because it was engineered to be, unlike smartphone apps developed by young twerp "professional developers".
I'm never going to allow one of these things to drive me, as I will not let Google make life-and-death choices for me. Let me illustrate with a story that happened to me a couple of years ago in Morocco. I was driving from Marrakesh to Ouarzazate (beautiful trip, btw) in a narrow road when a truck coming at high speed from the opposite direction invades my lane. I thought quickly of two ways to escape it:
1 - Swerve to the right, crashing into the mountain. Likely to destroy my car and cause me minor injuries.
2 - Slam the brakes, hoping that the trucker will get back to his lane fast enough, and the car behind me will not crash into my rear. Likely to avoid any trouble for everyone, but has the potential to cause a major accident if the truck hits me anyway.
What would the AI do? Decide it's worth it to damage me and my car to avoid any risk of a major accident? Or decide that the likelihood of a major accident in option 2 is low enough to be worth taking it?
Well, I don't know and I don't care; this is my life and my car and no AI has the right to decide this for me. If anyone cares, I chose option 2 and indeed nothing happened.
entropy happens
So I'm curious, besides the ad hominem attack on people who disagree with your position ("luddites"), do you work in research or in marketing? Maybe marketing research. You write like a total BS artist using 100% positive words discussing self driving cars and 100% negative terms discussing human drivers and those who are critical of self driving cars. They teach that stuff in school, and you write like a PR textbook, so who exactly are you shilling for? Do you get paid by the word or by the paragraph?
echo-chambers like Slashdot
That's the most frustrating thing hindering good debate on this forum any more. Dissenting opinions are shouted or modded down no matter how valid the concern is of the dissenter.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
They actually might get the same instruction. In this scenario the spoken content is a little different so that both pilots understand that the system knows what it's doing.
For example, suppose plane A is heading towards a mountain, and meanwhile plane B is passing over the mountain. Left to their own devices they may collide (or at least come much too close for comfort). TCAS will intervene, but it cannot tell either A or B to descend, since they'll hit the mountain.
Instead both A and B will receive the TCAS resolution "Climb. Crossing Climb". Which means yes, you're going to be climbing even though the other guy is climbing too, but I've done the maths and if you both obey nobody dies.
Further TCAS is live, so if one pilot unaccountably disobeys (maybe they can't obey due to technical problems) or circumstances change drastically, the other TCAS system will overrule its original resolution. For example having told you to descend, expecting your counterpart to ascend, the TCAS notices they have instead begun to descend, closing the gap instead of widening it. TCAS will change its mind, "Climb! Climb NOW. Climb"
Having owned a few big boats and seen them in accidents, and been Tboned in one. I call BS.
That old car has to have had some structural tinkering to fail to a tin can like that.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
And tell me how you would code that? Tell me how a coder that may or may not have experienced a blowout would approach that problem ?
/quote? Simulations, evolutionary algorithms, and the like. The beauty of software control is that software can have the experience of thousands of blow-out variations. Blowout of the front, right tire. Blowout of the front, left tire. Blowout of the rear, right tire. Blowout on dry pavement. Blowout on wet pavement. Blowout on gravel. Blowout on ice. Blowout on a flat road. Blowout on a hill. Blowout of the front, right tire on a winding mountain road in the dark on a stormy night. Blowout of the left rear tire on slush covered pavement while towing a trailer. Variation upon variation can all be gamed out in simulations before the code ever takes a real car out on the road.
When our name is on the back of your car, we're behind you all the way!
That's still not an answer.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Why do you move to the things that humans are worst at? Everything you're describing is *super easy* for a computer to do better than humans. You've picked scenarios with an incredibly low bar for software to cross.
What you should be picking are things humans find easy and safe, but which are nonetheless unusual. Like "the traffic lights have failed and a private citizen has taken it upon himself to direct traffic" or "driving into a restricted off-road area which is closed to the general public and blacked out by satellites etc.". Until that's solved (and I have little doubt that they will be solved), self-driving cars will likely need a human backup driver.
You're kind of like the person from the 80s who thinks a computer may recognize writing someday, but can never beat a world champion human at chess or go due to a lack of creativity. Turns out, beating humans at games is relatively easy for computers. Solving captchas is hard. In your "self-driving cars will never work" rants, you should look for scenarios that are like captchas, not scenarios that are like chess.
I am saying history has proven mankind if far too over-optimistic when it comes to what we think we can achieve and the reality check has often come at the expense of human lives lost.
At some point people will die and whether or not it's been determined if the vehicle was responsible there will be great controversy on either side. Nobody will take the issue *seriously* until after those lives have been lost. ( As with most every tragedy unfortunately ) So far a great overwhelming portion of the conversation here has been 'It's NEW it's going to be AWESOME! The software will make everything PERFECT. I am sorry that punch-bowl just looks like it's WAY too full of koolaid.
*Real world tests* Not simulations, as much as I love playing in them, will be what determines fault.
Have you seen those real world tests? Yes or no? If yes, please cite your references.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
A diverless car without any kind of overiding system enabling a human to take control (for whatever reason) is like a fully autonomous airplane. You will never see them.
Seriously? I'll never see a fully autonomous airplane or driverless car? You see these all the time, especially in rural areas.
http://www.uavcropdusterspraye...
http://www.enterprisetech.com/...
So obviously these cases do fit into your "constrained set of specific roads".
Anyway, I do suspect there will be a period of time where there are fully driverless roads, fully human-driver-only roads, and mixed roads. I suspect the major highways will be among the first to be driverless only (you already aren't supposed to be a pedestrial or a cyclist on those roads).
And in that case you fucking want the ability for a human driver to override an AI if the need arises.
Only in the same sense that I want to remotely slam the brakes on asshole's car on the road if the need arises. Otherwise, at a certain point, I fucking *don't* want a human yanking the wheel from the machine. Yes, we aren't there yet, but you're ridiculous if you don't think that's happening.
AI can't take into account all possible modes of failure or unexpected events.
Literally nothing can, including humans. This is a meaningless goalpost. If you take it seriously, then cars should not be allowed, whether or not there is a human driver.
An automobile not only has many, many more variables but it has variables that apply to it as well as variables that apply to other vehicles and terrain types. Call me back when your AV can get me to my home, in a snowstorm, without having been there before. Then, it needs to be safer than I am.
My home is in NW Maine, near the edge of cellular connection but there are actually two towers to work with, it's on the side of a mountain, the driveway is paved, it's about a half-mile long. It's steep but not too steep. The road to get there varies, it's in the mountains and foothills. It's not unknown to have 12" of snow. Your best bet would be something in a ruggedized SUV if you want to make this work.
I need to be private, completely private. The vehicle has some serious drawbacks due to the required size and resistance so it's inherently less efficient than it can be so that's a serious concern. The house is 24 miles from the village, the village is Rangeley, Maine if you want to see it on a map. I've actually needed to put a vehicle into reverse to keep from continuing forward - it was ice on the side of a mountain and hadn't been sanded yet.
I need to be able to operate this vehicle in the city as well. It has to meld with and ensure it does not interfere with local traffic, shipping traffic, and conditional traffic for those who might drive for the purpose of enjoying the trip and not just getting from one location to another. I want to do this privately. I can travel, with relative anonymity, currently - I do not live in an area where this is tracking - I do not want to.
When I am at my home, I can go through the woods and be in Canada. It'd be against the law but I will not cross a routed highway to get there. I have vehicles that will accomplish this task. It is a mere 40 miles, it's on logging roads, a old dirt road (without power lines or anything - some dude does live up there though and he's gonna need one too), and then I'm on snowmobile and ATV trails all the way into Canada. It crosses the border just above the actual border crossing. You can actually see it on the map - just use the address I mentioned and Google.
I don't want to be tracked. That's a requirement. I'd go so far as to say it is a right. More so considering I'd be needing to use it in areas where I have ever expectation of privacy - I'm quite frequently on private property. I should not be mandated to use it to the exclusion of all other vehicles. Access to the federal highway system should not be restricted for me - nor should state highway systems - even if tracking occurs on those sections.
There are many others who are in more interesting situations than I. I've not even begun to get into some of the many, many difficulties that need to be overcome in order to make this happen and happen in a reasonable manner. Hand-waving it away with airplanes, things that land under limited conditions, with limited surfaces, in limited places, and with limited locations, doesn't cut it. Let me know when the plane picks up passengers, departs, taxis, and makes it over to the mechanic's shop on its own - without human intervention. 'Cause having humans drive at the beginning and at the end is not autonomous, it's partially autonomous and that's not what we're discussing here.
Full autonomy is going to be needed, otherwise you end up with idiots who don't know how to drive and they're operating it only under conditions where computer is able to deal with it. Do you know where the largest number of accidents occur? In parking lots. They're not serious but, as a percentile, they're something crazy like more than twice as many - and that's not even extending it out to roads that are less than 10 miles from the destination. That's where, that and parking lots, crash - and those are often the areas where people live - or not far from them. You can't just go, "Your turn now!" and expect it to work out well in the end. That's just retarded.
So, sure... Make me one. Make me one that's affordable. Hop right on it Go
"So long and thanks for all the fish."
Of course it is. You want to know where all the crashed google cars are? They're in the same place a file goes when you empty your computer's recycle bin.
When our name is on the back of your car, we're behind you all the way!
Have you seen a commercial plane without human pilots ?
Yes. They're called drones. They're used all the time.
When our name is on the back of your car, we're behind you all the way!
Too much Koolaid in here man.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Yes, I agree with you. The actual accident scenarios are fairly logically covered. However, and here is the thing that HASN'T been convered:
The aftermath.
the accident has happened. Its a divided highway. The affected cars in the accident are strewn across the road in one direction. The autonomous cars all safely stopped by the side of road.
Emergency services show. Close the road, the police block a lane in the oncoming traffic. Then half a mile back they open one of the service roads connecting the divided highway and start directing traffic through it into the closed oncoming lane past the accident, and then accross another service lane back.
New arriving traffic is directed through the detour. Traffic caught ahead of service entrance, backs up to the opening and are directed through the detour as well.
Meanwhile the fully autonomous cars... sit patiently by the side of the road. waiting for it to clear.... for hours. and hours. and hours.
I gave a detailed starting list. It's in this thread - it's from a guy saying stuff about airplanes. Rather than type it out, I encourage you to give it a look. I am distinctly not the echo chamber - I've been expressing this for a very, very long time. If anything, I'm the one with the odd-ball view. See my other post for details. It's a bit long but you might like it.
"So long and thanks for all the fish."
D, I always enjoy your posts even if I don't always reply!
"So if it makes too many mistakes or gets into too many accidents it looses its license."
You now have one point on your spelling license.
Nope, I am an unlicensed poster. :-)
Google's argument fails if you add the following to the safety standards, which previously had all largely gone without saying:
Any vehicle operated on a public thoroughfare is required to have a minimum of one, live, actual, natural, biological, real, flesh-and-blood human person onboard, at any time while the vehicle is in motion with respect to the surface of the Earth; that human must be duly trained, licensed, in such condition as to be able safely and competently to operate the vehicle, and be either insured or able to present evidence of financial responsibility if self-insured, and the vehicle must have a full set of controls, immediately accessible to the driver, operable and in good repair, allowing said human to control the direction and speed of travel, to signal via sight and sound to other HTS users the presence of the vehicle, and the intentions of the operator as conditions may dictate at any given moment.
Moreover, any device that augments the driver's ability to see, signal, and control the direction and speed of the vehicle must be constantly and instantly overrideable by the operator, who must be ready, and the vehicle must be designed to grant the ability as well, to override ANY AND ALL such things, including but by no means limited to--automatic light control, speed (or 'cruise') control, and/or any DIRECTIONAL control, continuously while the vehicle is in motion.
FTFY. YVW.
Are you kidding me? Your argument boils down to "I don't know how to program this, therefore it can't be done".
"We" do this all the time. Often on dSpace HIL benches to simulate the dynamics of a failure and in the real world. What happens if we lose this sensor, what happens if we lose that sensor, what happens if this sensor.
A tire blow out is an absolutely trivial problem to program around.
Think about the visual problems that would occur from the shaking and vibrations of that event?
Google for 'low pass filter'. Hell you could probably run to stop open loop based on last good data.
Reading these stories you can tell that most Slashdotters aren't controls engineers. Stick to coding websites and what ever else you do.
But there's no way to simulate the real world failures like you're asserting has been done.
https://www.dspace.com/en/pub/...
We do it every single day.
Good god, if people tried to reply to 'em all they'd get nothing done - nor would I. I've actually had that problem - I spent almost all my posts answering replies from the day before. I had to stop answering 'em all. Sorry but, alas, it is so.
I don't think they're adoring fans, by the way. No, no... I get some interesting replies at times. Fans is definitely not the word I'd use. *sighs* I've really gotta start putting some of this word salad online somewhere at some point. I've multiple reasons to do so but I'm really picky and I've got to prove a point (and win a bet) with it. The good news is that the other party in this bet keeps putting it off and I keep letting him. I've changed my mind several times. Unfortunately, I can not disclose the details of this bet, per our agreement. Yes, yes it is funny.
"So long and thanks for all the fish."
I would assume letting an automated car on the road would need to be more like a series of drug trials. First test them in controlled situations, and then graduate them to situations where you are assured of meeting every kind of real world traffic and whether situation imaginable. Once they can get through all of that, then they are safe for public use.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
pilots land and take off manually.
An F/A-18 launches from aircraft carriers on auto (hands free). Various commercial jets can land automatically.
... well then its somewhat comparable to the F/A-18 catapult launch.
Takeoff is a "simple" thing compared to driving.
The difficult thing is taxiing, where its more a problem of getting verbal clearance and instructions from the tower. But once your on the runway with clearance
Because a serious failure in judgment of an AI car would be vehicular manslaughter if it killed someone.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Why would the owner of the car need to have car insurance? The owner of the car isn't the one driving. Perhaps you would need property insurance to protect the replacement of the car, as people have house insurance. But you wouldn't need to protect yourself from what that car might do.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
It doesn't really matter how safe they are. A bus is safer, but I don't have to pay for insurance when I ride a bus because I'm not the one driving it.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Yup I agree. Furthermore, if Google insists they are safe enough for people then no liability insurance needed.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Getting back to the original article, how many of these situations will come up on a standard driving test?
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Lets see how that compares to real world testing.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
You think that we use them because they don't correlate to real world testing? We just have an extra $100k laying around and go "Hey, lets buy one of those expensive dSpace things. It doesn't correlate at all to real world data but it'd be a cool space heater."
If you're indicative of the average Slashdot user that's scared of technology it's no wonder you're being replaced by H1Bs.
No, actually, the autopilot does not land the plane in bad weather, or in any weather for that matter, except in very rare situations involving a very few specific aircraft. We pilots take over when the aircraft gets within a few hundred feet of the runway and do the hard part of landing in over 99.9% of all landings.
I am extremely interested to see how a simulation reconciles to reality.
The fastest way to start a fight in nearly any community is to declare:
- Flight simulation or driving simulation is nothing close to the real thing.
- Simulation is as good as the real thing.
Yes I am sure there are some simulations that reconcile to reality better than others but the real point of all this will be determined by lawyers when 'edge cases' start taking lives.
- Was simulated emergencies adequate testing and who is liable?
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Wow Bin that was an amazingly low shot.
I am not afraid of technology, but lets do some *real* due diligence before we so blindly strap ourselves in.
Just because we can doesn't mean you should, and you don't grow old by being stupid.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
It'll be a cold day in all nine Hells before I get in some box on wheels with no controls for a human driver, and I know for a FACT that I am far from alone. Google can fuck off.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
We've spent millions and millions of dollars on this 'self-driving car' project, and we need to see some ROI soon or we're screwed! The only way that's going to happen is if the tech we've half-created can be made legal for sale RIGHT NOW so we can start making some of that money back, before the shareholders/investors/whoever comes down on us like a ton of bricks!
That's what I think is going on. They think they'll fix the monumental problems 'in beta' or some such nonsense.
Go to hell, Google. Your half-baked monstrosity isn't safe and doesn't belong on public roads.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
How do we move the car to safety if it dies on the middle of the road, if we can't steer it? Also can make problems for tow-trucks and mechanics who need to move cars around parking lot and shop floor if the machine is dead and there is no steering wheel..
I am sorry, but I really think you don't have a clue.
Neither have you since all your concerns appears to be about situations that are easier for computers to handle than humans.
If the autonomous driver is even half as competent as your average CNC mill it will be able to handle a blown tire without skipping a beat.
To the car it's no different than adapting to varying surfaces.
The programmer will typically not have experience with handling a blown tire and that is a good thing because experienced drivers doesn't take the optimal approach to it. An experienced driver will try to get the vehicle in a state more manageable. The computer doesn't need to.
The difference is that if something unexpected happens after the tire was blown the computer will have a lot better control over the car and can still react while the human driver is busy trying to manage a vehicle that he/she is in no way used to.
People seem to thing that being able to get the car to go exactly where you want to makes a good driver. That particular part is a standard regulator system and machines have been better than humans there for close to half of a century.
If you want to find arguments to why they can't replace human drivers you should stop looking for properties that you associate with good drivers. The things that computers might get problems with aren't as obvious.
Dissenting opinions are shouted or modded down no matter how valid the concern is of the dissenter.
Well, that can certainly happen, but it is completely unrelated to you invalid concerns.
A CPU would be better suited if it where only other self-driving cars on the road and the human factor was removed all together..
Imagine a car that not only see's what's around itself but also what all other cars around it sees.
"Oh there is a person running behind that structure and will be in my path"..
"The cars in front of me just noticed that there are ice-patches on the road in front of me.."
Look at how airplanes work.. A standard practice is when running into problematic things in the air is to actually turn on the auto-pilot because it knows the plane better than any human can.. Sure it's a lot less crowded in the air but handling situations with fast changing circumstances is something that a computer is actually good at..
Main problem for self-driving cars is non-logical humans.
1. Simulate the full car and introduce randomized failures / situations.
2. Evolve the code until it can handle everything you throw at it.
3. Load the code onto a real car.
4. Do tests on a test-track and introduce failures / situations for the real car. If failure go to step 1 and improve the simulation with more failures / situations.
5. Do small real-world testing. If failing go to step 1.
6. Increase to a larger set of cars. If failing go to step 1.
7. Increase to a even larger set of cars. If failing go to step 1.
7. Scale out.
It's not like it's exclusive with simulation or real-world.. It's incremental where the simulation can test for loads of things cheaply, including things that would be almost impossible to test in the real-world.
Imagine something that has been put thru *all* accidents in the world that occurred during the last 10 years, and introduce random failures like blown out tires etc, and where able to handle them all with the minimum amount of injury possible...
Even if all of those where simulated, and only 0.0001% was tested in the real world, it would still be many times better than even the safest human driver in the world.
All new accidents would also be fully recorded and introduced as new test-cases. This is how it works with airplanes.. For every accident learn what went wrong and try to make sure the same thing does not happen again...
[quote]Meanwhile the fully autonomous cars... sit patiently by the side of the road. waiting for it to clear.... for hours. and hours. and hours.[/quote]
Or the police sends out a message "back up until this exit-point" and suddenly all autonomous cars start backing up, in a safe manner, to the exit and they continue from there... Or the police, since they know the autonomous cars with drive in a safe manner, will open up a tiny lane on the side of the accident and allow the autonomous cars to pass by slowly and safely..
But to start with... when you have that many autonomous cars on the road the biggest suspect for the actual accident is that a human was driving..
Heck, a human would not be able to operate a modern airplane without the assistance from computers.. All the humans can do is say "I want to do this, please help me"
Why do I care about what someone's opinion on simulation is? I'll look at the data.
- Simulation is as good as the real thing.
This is how I can tell you're not an engineer. In engineering nothing is perfect. It's always 'good enough'.
After almost two decades on Slashdot and a decade in industry it's dead accurate. You guys whine constantly about new technology. I can almost guarantee most of you are the guys in the office that insist on doing things 'the way you learned them' 20 or 30 years ago. Where I work you're the barriers to progress. Up to a point it's just cheaper to keep you around until it's easier to start over with a fresh grad or a H1B and train them on new tools.
This isn't a CS problem, this is a mechanical engineering / dynamics problem. Sometime along the way they figured out it would be easier to teach engineers to code than it was to teach coders to engineer. We use tools that most of Slashdot thinks don't even exist, like Simulink (a drag and drop coding language of sorts).
but lets do some *real* due diligence before we so blindly strap ourselves in.
We are. But the test cases that you've come up with are quite literally the lowest hanging fruit there is. A flat tire? Most cars auto driving is going to come with first come with run flats. (BMW, Benz). For those that don't it's still one of the simplest FMEAs to possibly do. You put a car in the test cell and blow the tires over and over and over, collect data, toss it in the HIL and test your controller. Toss the controller on a car. Repeat.
And I can guarantee that the self driving car is going to handle a blown tire better than you ever will. It'll handle it better than race car drivers will.
and you don't grow old by being stupid.
And you don't stay gainfully employed by doing things the old way.
I'd be interested in seeing them take the test here in Winnipeg in January during a snow storm. My expectation is that they cherry picked the location, weather, etc, for the test. Given that it's a totally new paradigm, they need to be tested under the worst, most challenging conditions available.
linquendum tondere
The point isn't whether an autonomous vehicle can be made totally safe, the only thing that matters is whether it's safer than a human behind the wheel. And that's a much easier bar to pass.
Uncomfortable as it might be, autonomous cars causing a few deaths per year isn't a problem. Not as long as it's insurable (which it will only be if it doesn't happen too often, both compared to the current situation and in absolute numbers).
And besides ... reliability of service is not a design objective in a phone OS. It's designed to serve as a platform to gain market share at the lowest possible cost.
File systems, networking stacks, and process swapping on the other hand are designed with reliability in mind. And Novell servers were. And mainframe OS'es are. And avionics are. How often do any of those crash? Compared to the failure rates you'd get if those tasks were carried out by hand?.
The point of GP was that computers can't be trusted with life or death decisions, because even our phones have bugs. To which somebody else responded that we already trust computers with our life, such as those in airplanes or medical equipment. The fact a modern plane still has human pilots does not negate that, as the humans can't override every single computer action.
There have been a few air plane crashes which occurred when the pilot overrode the auto-pilot. The one that comes to mind was in the USA where the plane on approach to landing was struck by a wind phenomenon that is common to the area. The auto pilot attempted to correct but the pilot overrode it and promptly crashed the plane. The FTC (or whoever it is that post-mortems plane crashes) determined that if the pilot did nothing then the plane would have landed fine...
100% working os is no problem, in fact i would say all op sys-es can run forever without crashing. The condition being that there is no user doing who knows what all the time. Take a computer, give it a job and leave it alone, it will work until the hardware gives out the magic smoke.
LOL I'm glad I don't drink alcohol any more. I didn't quite spit it up but I *did* have to swallow awkwardly.
Hmm... That sentence should not be taken out of context. It was just coffee. There's a whole lot of really strange optimism with this. People seem to think they're going to be able to get drunk, hop in the back, and say, "Home, James." They seem to think this is right around the corner - and that it will be done while respecting privacy. Oh my... They've never seen the big picture that is traffic and, trust me, one can not really model it (well) in their head - there are too many variables.
What they're getting is advanced lane maintenance and adaptive cruise control with automated braking and stability control. Half of 'em seem to think they're getting Star Trek teleporting machines. It takes longer to model your commute, on some really big iron, than it does for you to make that commute. I know... Obviously we don't need to go to those extremes but the closer one can get to that then the closer one can get to having a utopia (sans privacy). Some people seem to believe we're almost there.
I envision kids in the back my car. "Are we there yet, Dad?" My standard reply was, "Has the car stopped and I said it's time to get out?" Fortunately, my children are grown, health, and intelligent. As I look around, I feel like I've been blessed with good fortune. Thanks, FSM. I'm glad my children aren't retarded.
An automated car does not need to model traffic. Do you need to understand "the big picture" of traffic on your commute while you are driving? Or do you drive your vehicle, navigating by sight (lane keeping, turning where necessary and speed control) while keeping an eye out for any potential hazards and being prepared for any "oh shit' moments? Heck the only real information that is nice to have (regarding "the big picture") is where the traffic is starting to get congested so you can route around it if required...
I'm getting really tired of their implication that people are software bugs that must be eradicated. Maybe these cars are safer than the average schmo, but some of us drive quite well and can make split second decisions the software can't, that makes *US* 'safer' by magnitudes. It's faux altruism anyway, Google has its eyes squarely on the $$$.
1. It's not what you say, it's how you say it.
2. Your concerns are also not shared by all, and certainly not to the degree that you seem to be displaying.
Right now you're playing the equivalent of the anti-nuke NIMBY. We aren't really interested in whether or not this technology is perfect. If it kills fewer people than the current technology it's a win. If it only does that by a small margin, so what? If it kills people by the millions, we'll probably stop using it, but if you think that's remotely likely, well, you can probably expect more downmods in the future. You're not as bright as you think you are.
if I were a commercial pilot, I would poke fun a bit, and when the plane was taking off I would poke my head out of the cockpit and say hey does anybody know how to use this thing?
You've got it all twisted... TCAS is NOT perfect, it's just that both pilots doing a sub-optimal avoidance maneuver is better than one pilot disagreeing with TCAS, while the other obeys... If one pilot obeys, while the other chooses a better maneuver, it's TCAS that CAUSED the crash. Without TCAS, those collisions would likely not have happened. Of course, blaming the pilot for every crash caused by faulty equipment, where only an omnipotent pilot could potentially have avoided it, has been a long tradition in the airline industry!
A superior system to TCAS would be one that reads the control inputs of each pilot, and uses that as the basis of how to instruct the other pilot.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
I can't imagine the level of stupidity required for Google to make such a request. Google's own reports (which they resist being required to provide) show quite clearly that human drivers are frequently a very necessary supplement to their autonomous systems:
"Between September 2014 and November 2015, Google said there were 272 occasions when a technology failure forced the test driver to re-take control."
https://ia.acs.org.au/news/goo...
And this request comes shortly after a Google car was found fully-responsible for crashing into a bus:
http://www.bbc.com/news/techno...
And that's not a one-off... Google's small fleet of self-driving cars are getting in numerous accidents. 8% is the last figure I saw. Google spins it as the fault of everyone else except its own vehicles, but that claim is specious at best:
http://gizmodo.com/self-drivin...
There's ample evidence that self-driving cars do several things which (while they MIGHT be safe if all cars were autonomous) cause clashes with existing human drivers on the road:
http://pipedot.org/story/2015-...
Even the much-simpler task, of drive-by-wire in existing automobiles has proven too unreliable to trust human lives with. Toyota screwed this up badly, and it has cost them dearly:
http://www.eetimes.com/documen...
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
Your assumptions about me couldn't be more incorrect but I really don't care. Nobody ITT has done anything to communicate safety in terms the general populace will accept.
The other thing is how groups have made grand claims on what the future will hold. Flying cars and what not while some modest claims had been greatly exceeded. /. has become an echo chamber and to hell with anyone that says 'wait, lets analyze this'.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
I can see Google's point since a lot of people took their driving test years ago and some barely passed. Some states require a written exam after a certain amount of time and some states even allow people to take this exam open book if they are renewing their drivers license. This written test can also include a bunch of unrelated questions such as how high of a blood alcohol level can you legally have before you're legally drunk or how many months is your license suspended on a DUI conviction. These question have nothing to do with safely operating a motor vehicle and most people realize alcohol does impair their ability to drive and avoid driving if they have been drinking. As long as you have a valid drivers license, your car is insured, and your car registration is current (you passed the required safety and emissions inspections) you can legally drive. If someone currently is an unsafe driver it would take years before they would eventually lose their drivers license. For example, if people report the person as an unsafe driver to police they really cannot do anything unless they actually observe this driver driving unsafely or this driver gets into a lot of accidents. In this case all the driver has to do is retake the written test with a passing score and they may even be able to take the test open book to retain their drivers license. Even if you are required to take a behind the wheel test I know some people who took the test on a driving course and some have missed a red light or stop sign and still passed the driving part of the test. The driving course doesn't test for conditions such as pedestrians crossing the street or the ability of the driver to share the load with bicycles and being able to see pedestrians in a crosswalk, motorcycles, and bicycles when making turns. The test also doesn't test if the person has bad habits such as following others too closely to the point that they would be unable to stop if the person suddenly slows or stops, a person who makes dangerous lane changes, a person who doesn't pay attention to their surroundings, etc.
I think the system should have a way to operate a self-driving car manually in the event there is a problem with the system. I'd hate to be stuck in a car going 65 MPH and find out that there is a glitch that can cause it to crash into another car and you cannot temporary disable the self-driving system and "go manual" until this is fixed. Some issues may take a long time to figure out and it could be when a driver relies on this system to operate safely that they realize there is a glitch in the system. For example, I had truck with an ABS sensor that went bad which resulted in my ABS system not being able to stop at all as it continuously detected a brake lock condition when there wasn't a brake lock condition. The ABS testing system didn't detect this problem but I was able to pull out the fuse to the ABS system to temporary drive my vehicle without crashing to get it fixed. Some cars and trucks may not even have a way to disable ABS and in such a case the person would be screwed and would either crash or have to have their vehicle towed to a mechanic when the car would be fully operational (minus the ABS system and only conventional brakes of course) with the ABS system disabled. A self-driving car should have a steering wheel and other equipment in the event the self-driving system fails so that a driver isn't stranded and can simply "go manual" an manually drive the car to a mechanic to make repairs to the self-driving system.
Moreover, the TCAS command outputs are not hooked up to the autopilot inputs, for good reason. The human pilot still has to disconnect the autopilot and decide whether to obey TCAS or not.
Nobody ITT has done anything to communicate safety in terms the general populace will accept.
And they've been conveyed in terms they will understand how the RTOS on their current car runs? Or how open heart surgery works? Or how any number of things they use daily work? Or do they leave it up to people that do that for a living.
'wait, lets analyze this'.
We ARE analyzing it. We have been for decades. You refuse to listen to engineers that have actually done stuff similar. You've come up with the easiest and lowest hanging fruit as 'problems' and then refuse to listen to any solution anyone comes up with. It's fairly evident that no one else coming up with these 'crazy impossible problems' is an engineer. I haven't seen anything ITT that is non-trivial.
I even posted the tools we use to simulate real world scenarios and you refused to accept that it could possibly be correct. As if companies are spending millions+ on these devices because they don't simulate things accurately.
Or the police sends out a message "back up until this exit-point"
Ah, but now you are just speculating on future technology and systems which DO NOT CURRENTLY EXIST.
Sure such systems will surely eventually exist, but they want to push the cars out now.
But to start with... when you have that many autonomous cars on the road the biggest suspect for the actual accident is that a human was driving..
Or an act of god. A sink hole. A mudslide. A downed tree or power line.
And 15 years out, those autonomous cars are going to be riddled with dirty and failing sensors and dodgy electronics built by low bidders.
The autonomous cars on the road right now are state of the art being baby sat by teams of dedicated engineers who test and monitor everything continually.
Next time you're out driving take a look at all the poorly maintained clunkers out there. With a broken headlight, and the air conditioning doesn't work, the sunroof doesn't open, and one of the doors only opens from the inside. Tires are 2nd hand and don't match. Cruise control hasn't worked in 5 years. Filters are probably overdue. Timing belt is on borrowed time. Now imagine it's also autonomous ...
Google does not have a valid argument here. The current driving tests build on the assumption of a somewhat functional/thinking human being at the controls. The tests are certainly not comprehensive for all situations that might be encountered on the road.
So evidence that runs contrary to your prejudice must clearly have been fabricated.
I'm not sure whether that falls under a conspiracy theory, or just plain ignorance.
You do not have a moral or legal right to do absolutely anything you want.
A 100% discount is still a discount. Besides, car insurance pays for more than just wrecks. It also pays for things like a tree falling on your car. So even if the automakers cover the cost of wrecks, most people will still carry insurance, just at a deep discount.
Check out my sci-fi/humor trilogy at PatriotsBooks.
Right, but insuring property against damage is far, far cheaper than insuring a vehicle that I may use improperly and cause damage. Current vehicle insurance factors in the risk of me being a bad driver, but with automated cars if that risk is there than I am not the risk Google is. Think about how cheap house insurance is as opposed to driving insurance, at least it is a lot cheaper where I am. In fact, the automated car could just be something that is tacked onto house insurance for an extra 50/year or whatever. Furthermore, driving insurance is attached to my driving record which shouldn't be the case either with automated cars.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
There are plenty of computers in use (a lot of the better ones are running Linux or an RTOS and hell, even Windows NT/CE/XP) that people trust their lives to implicitly on a daily basis in a lot more delicate situations than driving a car. Commercial planes do most of the flying fully autonomous, most of both your debt and savings is being invested fully automated, any machine in a hospital parses a lot more data than a few dozen sensor and requires much more precision.
Have you ever seen the types and numbers of computers used for commercial aircraft? An Airbus has 4 flight computers, 3 primary and one backup. If it cant get a consensus on all three computers it goes back to the pilot for input.
You also know that the reason we keep two pilots in the cockpit is because sometimes these systems fail when given bad input. It may be rare, but it's not unheard of for a plane to start increasing the AoA (Angle of Attack) when it receives bad data. Generally a pilot picks up on this before it becomes an issue. AF447 is an example of both the autopilot failing (Pitot tube iced over provided bad input) and then the pilots failing (switching off the stall alarm, inconsistent input from the pilot and co-pilot).
Software that has the potential to take lives not only has thousands of hours of testing put into it, humans who interact with these devices are given intense training. How long does it take to become a pilot or a nurse who uses the machines you spoke of (and how often does someone's savings get wiped away because of incompetent traders). You're pretty quick to put your life in the hands of incredibly complex software when engineers aren't.
Lets face it, auto manufacturers cant even get the software for an automatic transmission right and that has one job (moving cogs about at the right time). Would you really trust it to make all the driving decisions for you?
Calling someone a "hater" only means you can not rationally rebut their argument.
The crux of your argument is "when the car detects".
This is also the fatal flaw of your argument as computers aren't actually very good at this. In order to process a lot of data in near real time (these milliseconds you keep mentioning) then you need a lot of computing power. Right now the car operates mostly on LIDAR which is good because it limits the scope of input making it easy to process but it also means that a lot of data is missed.
You haven't asked "how a car will detect a blowout" you've said "when a car detects a blowout". That is why your argument is bunk. You've started off on a false assumption. The assumption that makes the rest of your lengthy post pointless.
Now I have thought about it, the way a human detects a blowout in another car is visually. Sound is useless on highways, so we detect it by cars that lurch in a particular direction. I dont think you quite understand just how difficult it is to do this in real time with modern computers. It will take a fast server (such as an IBM X series, feel free to swap for your favourite brand, this is just an example) at least several seconds to find a specific landmark in an image and here we are talking about software that knows exactly what it is looking for, with a blowout you're looking for a set of conditions that will change as the degree and direction of a car lurching will be different each time. You're asking a computer to do this within a number of milliseconds at a minimum of 24 times a second.
We're not Luddites for questioning this, it just demonstrates we have some idea of the complexity of the problem.
The solution actually is that an autonomous car will keep a distance large enough that it can detect a change in velocity, which will piss off most drivers causing them to take manual control because the car is too slow. Yes, humans are selfish creatures. It also means that the pipe dream of hundreds of autonomous cars riding bumper to bumper is nothing but a pipe dream. In fact due to the inefficiency of sensors, reaction times of on-board computers (which will be made to a budget) and understandably cautious engineers they'll be keeping far larger gaps than will be needed for human drivers.
Finally, Google's track record comes in no small part to the fact that a human driver has been in the car at all times. One thing they dont like publishing is the number of times that the human has had to intervene in traffic situations. The at fault crash by an autonomous car in February demonstrates this. The car made a bad decision by pulling out in front of a bus, the human driver also made a bad call by assuming the bus would let them in. When autonomous cars are finally put into the hands of the average car buyer, we'll see accident rates increase significantly.
Calling someone a "hater" only means you can not rationally rebut their argument.
Because the owner owns and directs a potentially very dangerous machine. Legally, people are responsible for things, and in particular owners of autonomous cars will be liable for what their cars do. The liability insurance may become very cheap, or the manufacturer may indemnify owners of its cars (and in that case will probably self-insure), but liability insurance of some sort is going to be necessary.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
If they are potentially dangerous then they are not ready for market, period. Imagine if people had to buy insurance before boarding a bus or a train. An automated car is nothing more than a personal bus driven by a black box. I should be able to trust that black box like a bus driver or train engineer, not play some sort of Russian roulette with my insurance policy.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
When speculating on technology that doesn't currently exist, it's normal to assume that it's been done competently, so that any remaining issues are the fault of the technology. Rejecting future technology on the grounds that, if the engineers are idiots, it could potentially have bad consequences is stupid.
For example, an autonomous car could do a systems check and refuse to go anywhere but a repair shop if it detects that too many systems are unreliable. That's one reasonable thing that could be implemented.
There are some of us who do maintain our cars properly. Once more, you're generalizing from the worst case and assuming the computer won't know what to do about it.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
You may not have realized this, but modern tires come with pressure sensors that communicate with the car. Unfortunately, the cars I've seen like to digest all of this information and distill it into the message "at least one of your tires is a few pounds low, or flat, or somewhere in between" to inform the driver, but the car will know of a sudden tire pressure drop. It will also note undue movement instability as it proceeds.
As far as a blowout in another car goes, the computer will be monitoring the movement of the surrounding vehicles, and will note if one starts behaving erratically. The computer HAS to keep a model of surrounding traffic, like you and I do.
You also seem very confident that computers aren't powerful enough to handle rich sensory inputs, without anything to back it up. The problems are formidable, and will continue to be challenging, but there's a lot we can do to offload processing. For example, your retinas do a tremendous amount of processing, and what comes out of the optic nerves is the result of sophisticated edge detection with color information.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Dunno about you, but I look about as far forward as I can manage, to see what I might be getting into. There are automated signs on the freeway I spend the most time on giving information about stuff going on beyond what I can see.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
No, I am an unlicensed poster!
Exactly. Situations humans find easy will sometimes be hard for software to recognize -- a human directing traffic is a great example, or when a construction crew puts up cones but not a clear indication of which side of the cones are going to be worked on, as well as other similar temporary markings. Traffic lights with a burnt-out bulb, or a stuck sensor that won't turn your lane green, old lane markings never properly cleared off before new ones were in place, or even just parking in the right place when a friend is having a party and all the regular spaces in their driveway are taken up. Driving off-road where there are no roads to follow (Sure, you'd drive yourself for fun if we're talking duen-buggies and desert sands, but not so much when trying to get to a remote herd on a large ranch.) Even things like snow, ice, and really hard rain shouldn't faze a robust driving system, but a poorly-marked detour through a parking lot like I had to take this morning? Those will probably need human intervention or similar for a while, though I doubt it will take long for even the extreme edge cases to be covered.
Rejecting future technology on the grounds that, if the engineers are idiots, it could potentially have bad consequences is stupid.
I am not rejecting future technology; I am simply arguing that the future technology does not exist yet. Therefore calls to put autonomous cars on the road today is premature.
If you read the thread, for example, you'll see that I argued that if a road is closed and a detour is established by police using service roads and oncoming lanes the autonomous cars would not be able to use it. The person responding to me said the police would simply "broadcast a message" to advise the cars stuck at the road closure to get them into the detour. The police do not have this capability. Surely one day they will, but if google puts an autonomous car on the road this year, this detour broadcasting system simply doesn't exist.
There are some of us who do maintain our cars properly.
Of course there are. So what?
Once more, you're generalizing from the worst case and assuming the computer won't know what to do about it.
I am not generalizing from the worst case. Even assuming many cars are fine, it is the remainder that will cause problems.
I am not against autonomous cars, hell I think they will improve the world considerable. But I simply don't think we are remotely ready for fully autonomous cars that never require a driver on the road today. There are too many scenarios they aren't able to cope with.
Hybrids that can be fully autonomous, or driven, sure we can do that -- because a driver can deal with the extra scenarios. But that's not ideal...we want JohnnyCab ... a car that can drop the kids off at hockey practice or piano lessons without having to go ourselves, a car that can pick us up and take us home after a night drinking, etc. But the cars aren't ready for that ... yet.
Your reading comprehension skills need some work.
My personal experience with large old steel cars has always favored the big car in conflicts with smaller vehicles made with more squishy materials. In fact I would be short a couple kids if we were in a minivan instead of a full size passenger van when we were hit. I will not put the lives of my children at the mercy of 'simulations'. End of discussion there.
What drives me batty is that people don't question anything anymore. Especially when said material is generated by people with a financial incentive to achieve a certain set of results. Yes, engineers, researchers and scientists all need a paycheck too I get it.
Bottom line is my personal experience trumps anything an 'industry' cobbles together to prove it's point. The drawback to all this information is the multitude of mis-information and the only thing I can trust anymore is personal experience or conversations with people I know in real life and trust to be reputable.
The rest is just the entertainment / advertisement industry. Those are the facts and I don't care if you approve or not. There's always going to be some shill crying foul somewhere.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Wake me up when those autonomous cars can handle these simple situations that can occurs in real life (not those edge cases with blowing tires and running kids):
1- Heavy rain ;-)
2- Unmarked lanes (because they never had markings, or because they fade away every winter like here in Canada)
3- Snow. Lots of snow. Like 1 foot. With ice under it.
4- Getting stuck in snow, and having to do a pendulum maneuver to get out (regular occurence for me... much easier with a manual car)
5- Freezing rain. Making sure that all those sensors are given a good 1/2" of ice on their surfaces.The car can drive itself, but it sure can't scrape the ice... you'll have to do it.
6- Gravel roads.
7- Parking. Not parallel parking. Parking at the mall.
8- Camping roads.Getting to an from your camping spot, in a national park for example.
9- Hand signaling from a police officer, or from a benevolent stranger in an emergency situation.
10- Construction zones, with temporarily driving in the wrong direction, against traffic.Or just dodging cones and sandbags without hitting a bus
These are rather simple things that, for me, the Google car can't cope with. But I might be a little behind the news so correct me if I'm wrong.
Try it! Library of Babel
Every so often, there is a bus or train accident. The reason passengers don't buy insurance is that it's covered by whoever's running the operation. If you're running the operation by having a car drive around, you're the one causing it to be on the road.
There is no such thing as perfect safety. If you're waiting for a perfectly safe car, you'll never get one.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
You're assuming that, in a world where autonomous cars have been around long enough to be common, that there would be no way for the police to communicate with them. Otherwise your argument makes no sense. Police do not currently have the ability to communicate with autonomous cars, because there are none. Hypothesizing autonomous cars with absolutely no change to deal with them is ridiculous. Either we will get them, and the police will be able to communicate with them, or we won't get them. Right now, we've got experimental vehicles with drivers ready to take over, and the police can communicate with the drivers. Before the first fully autonomous vehicle is allowed on public roads, there will be a way for the police to direct it.
An autonomous vehicle that relies on the human to take over in emergencies is really, really dangerous, because in the second or two for the human to realize what's going on the vehicle will crash. That's not a realistic option.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Then I don't see the point of putting AI in a car. If I'm driving the car and I want to be safe, I drive slower and/or more carefully. If I have AI in a car, I have no choice I just have to accept what happens and that's pretty ridiculous unless they can make it 100% safe.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
You're assuming that, in a world where autonomous cars have been around long enough to be common, that there would be no way for the police to communicate with them
I am making no such assumption. I am *merely* pointing out that if google puts those cars on the road *this* year, the police do not have the ability to "message them".
Naturally it WILL happen eventually.
Before the first fully autonomous vehicle is allowed on public roads, there will be a way for the police to direct it.
Did you read the full article. Google is asking for expedited permission to bring to market a car that has no driver controls.
To wit: "Under Google's proposed framework, a company that could show its vehicles passed federal safety standards could receive permission from transportation regulators to sell them. "
So, no, you are mistaken. The notion that the police would be able to send messages to these vehicles to direct them into detours is not part of this proposal.
And never mind the police, what about normal people... if your google car is driving along and you come across a downed power line or tree or flooding that has submerged the road, and your car comes to a safe stop before it reaches it, then what?
Are you going to be able to order it to make an otherwise illegal U-turn?? Will it drive on the grass or sidewalk around an obstacle at my prodding? Will it pull into a service lane?
None of that has been tested or even defined. Right now, when an autonomous car runs into trouble it safely pulls over. If it doesn't have a steering wheel or other controls... the question becomes: Now what?! Do I get out and walk?
An autonomous vehicle that relies on the human to take over in emergencies is really, really dangerous, because in the second or two for the human to realize what's going on the vehicle will crash. That's not a realistic option.
Of course. That would be ridiculous. I am not advocating that. However, after the car has stopped safely, there is no reason the human can't take control.
Well going by that logic, the people who write them are human and the humans that are driving on the roads consist of the lowest common denominator, ragers, texters and drunkers, I think self driving cars are still miles better
The AI will (at some point) be a considerably safer driver than you or I. There's lots of things that don't work 100% but are still worth having.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Well, ultimately we will have to see what insurance companies think of that and how affordable it is. Insurance companies don't like risk they can't calculate with a high degree of accuracy and they won't be able to do that without seeing a great deal of use under local conditions. Since people can't drive without insurance here either, it will ultimately have to be Google that provides those real world examples and testing. I live in a place where there are a lot of blizzards and snow covering everything. They don't plow down to pavement and the ice ruts that remain don't line up with lanes and can spin a car sideways and have it sliding down the road in an instant. If the insurance companies have the attitude users of the vehicle must pay for the unknown liability of a black box being able to accurately map the contour of the ice on the road and drive on it properly then they will be unaffordable anyway, even if the damage they may cause is within a reasonable limit.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Large corporation says it's products are safe because it's their product! No surprise there.
the only "surprise" is the lack of critical thinking on those who correctly have a healthy skepticism on anything with large amounts of money involved and profit to be made, because it's a "tech company".
Letting the insurance companies deal with these things generally works well. There's enough competition to keep companies fairly honest, and they employ large numbers of people who are very good at estimating risks.
I suspect that the AIs we have in ten or twenty years will be better at driving on ice ruts than I am (even disregarding the fact that my driving ability is probably in decline now).
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Ok.. in ten or twenty years probably, but it seems like Google is looking at releasing these imminently.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.