Get Ready for Freeways That Ban Human Drivers

[Stylteralmaldo]

New rules of the road for robot cars coming out of Washington this week could lead to the eventual extinction of one of the defining archetypes of the past century: the human driver.

While banning people from driving may seem like something from a Kurt Vonnegut short story, it’s the logical endgame of a technology that could dramatically reduce – or even eliminate – the 1.25 million road deaths a year globally. Human error is the cause of 94 percent of roadway fatalities, U.S. safety regulators say, and robot drivers never get drunk, sleepy or distracted.

Autonomous cars already have “superhuman intelligence” that allows them to see around corners and avoid crashes, said Danny Shapiro, senior director of automotive at Nvidia Corp., a maker of high-speed processors for self-driving cars.

More:
bloomberg.com/news/articles/2016-09-22/robot-rides-may-force-error-prone-human-motorists-off-the-road

 

[ShrodingersCat]

Self-driving cars are much safer than human-driven cars.

What I find really interesting is the ethics behind it. Because you have to program in accident-preventative measures and there could conceivably be scenarios where you have to choose between two different terrible outcomes.

In a real-life scenario, it’s up to a split second decision from the driver. In this case though, the response to such an event would be predetermined.

For example, if you are about to hit two people, is it OK to swerve if you will only hit one person? A trivial enough example, but they can get complicated.

Not that that means I’m against self-driving cars. I just find them to be a fascinating source of ethics thought experiments.

 

[TCEL]

Sometimes, there’s something to be said for accidents and human error. The thought of having to quantify a human’s worth is deeply unsettling. What if the person is a child? Homeless? Elderly? Etc.

 

[Hatikvah]

Sometimes, there’s something to be said for accidents and human error. The thought of having to quantify a human’s worth is deeply unsettling. What if the person is a child? Homeless? Elderly? Etc.

Then it is often said: “what can you do?” Accidents are inevitable, unfortunately.

 

[Exiled_Child]

Thanks for posting this, Tyler. I’m really dreading the inevitability of this, despite all the supposed advantages laid out in the article. I’m planning on resisting this emerging technology for as long as is legally possible. I would prefer taking any alternative route, no matter how long and inconvenient, than to ride on an autonomous freeway.

 

[edwest2]

This is just totally nonsensical. I’m fine with surface streets now. If I’m forced - that’s right, forced - to get a driverless car, I will drive less. I will never use freeways. I write science-fiction and there is no possible scenario where this will work as planned. None.

The military autonomous vehicle testing I’ve read about for years still have issues. But some are being deployed anyway.

nbcnews.com/tech/security/future-tech-autonomous-killer-robots-are-already-here-n105656

How many accidents are caused by manufacturer defects in current human-driven cars? How many recalls have we seen? Sorry, this is a deadly joke to celebrate a very, very poorly thought out system that will begin to break down. There are other issues, but who cares? Some wealthy people want their vision to create a fictional future where all driverless cars will be perfect. Just like all roads are perfect

I love science, but I hate this. It is human guinea pigs in cars. And if a few of us die, “Oh well, they’re just peasants.” But if someone counted as really important dies, the tune will change.

Ed

 

[Exiled_Child]

This is just totally nonsensical. I’m fine with surface streets now. If I’m forced - that’s right, forced - to get a driverless car, I will drive less. I will never use freeways. I write science-fiction and there is no possible scenario where this will work as planned. None.

The military autonomous vehicle testing I’ve read about for years still have issues. But some are being deployed anyway.

nbcnews.com/tech/security/future-tech-autonomous-killer-robots-are-already-here-n105656

How many accidents are caused by manufacturer defects in current human-driven cars? How many recalls have we seen? Sorry, this is a deadly joke to celebrate a very, very poorly thought out system that will begin to break down. There are other issues, but who cares? Some wealthy people want their vision to create a fictional future where all driverless cars will be perfect. Just like all roads are perfect

I love science, but I hate this. It is human guinea pigs in cars. And if a few of us die, “Oh well, they’re just peasants.” But if someone counted as really important dies, the tune will change.

Ed

Agree 100%, Ed. I dislike even riding around in newer cars with all these gadgets like keyless ignition, remote control door locks, headlights that turn on automatically, etcetera, etcetera. Looking back, I think the best cars I’ve ever owned were Volkswagen Beetles. When I was a teenager I practically rebuilt an old 1967 VW engine with virtually no mechanical knowledge and only an amusing how-to book by an author named John Muir. That car was non-stop reliable, and I drove it from the West Coast to the East Coast and back, from the northern border to the southern border, and throughout Northern Mexico and Baja California. Never had any real problems and it was always a pleasure to drive. Didn’t even have a radiator, and had such a simple design that a clever person might be able to train a Border Collie to do basic repairs. I had a couple of them, and while you wouldn’t want to get in a head-on collision in one, they were great cars. I’ll never buy one of these newer cars until it’s legally mandatory.

 

[JimG]

I’m thinking of the massive hacking potential by foreign nations and criminals.

 

[ThinkingSapien]

While banning people from driving may seem like something from a Kurt Vonnegut short story,

California had some R&D lanes for automated vehicles years ago. I can understand the rule expanding. Human drivers are a radical unpredictable factor.

Autonomous cars already have “superhuman intelligence” that allows them to see around corners and avoid crashes

Isn’t that usually from radar, not intelligence?

 

[TCEL]

Agree 100%, Ed. I dislike even riding around in newer cars with all these gadgets like keyless ignition, remote control door locks, headlights that turn on automatically, etcetera, etcetera. Looking back, I think the best cars I’ve ever owned were Volkswagen Beetles. When I was a teenager I practically rebuilt an old 1967 VW engine with virtually no mechanical knowledge and only an amusing how-to book by an author named John Muir. That car was non-stop reliable, and I drove it from the West Coast to the East Coast and back, from the northern border to the southern border, and throughout Northern Mexico and Baja California. Never had any real problems and it was always a pleasure to drive. Didn’t even have a radiator, and had such a simple design that a clever person might be able to train a Border Collie to do basic repairs. I had a couple of them, and while you wouldn’t want to get in a head-on collision in one, they were great cars. I’ll never buy one of these newer cars until it’s legally mandatory.

This. I think there’s something to be said for the ability to fix one’s own method of transportation. The further along we go, the less ability I foresee of that potential. I don’t see how anyone could do that with a driverless car unless they can fix their own fried chips or whatever. It seems to me to create even more dependency.

 

[johnnyc176]

I’m thinking of the massive hacking potential by foreign nations and criminals.

I’m thinking of the legal killing of innocent human beings.

 

[Monte_RCMS]

Will it automatically steer around a deer crossing the road?

If it loses the signal, then what?

This coming winter should be interesting. Do they have branch offices in Alaska? … for testing under harsh conditions.

Nice for visiting grandma. Or for going to the beach.

 

[Monte_RCMS]

Agree 100%, Ed. I dislike even riding around in newer cars with all these gadgets like keyless ignition, remote control door locks, headlights that turn on automatically, etcetera, etcetera. Looking back, I think the best cars I’ve ever owned were Volkswagen Beetles. When I was a teenager I practically rebuilt an old 1967 VW engine with virtually no mechanical knowledge and only an amusing how-to book by an author named John Muir. That car was non-stop reliable, and I drove it from the West Coast to the East Coast and back, from the northern border to the southern border, and throughout Northern Mexico and Baja California. Never had any real problems and it was always a pleasure to drive. Didn’t even have a radiator, and had such a simple design that a clever person might be able to train a Border Collie to do basic repairs. I had a couple of them, and while you wouldn’t want to get in a head-on collision in one, they were great cars. I’ll never buy one of these newer cars until it’s legally mandatory.

Do you know the reason why VW gave up the original Beetle?

 

[Usige]

Having been a software engineer for 20+ years I find the idea both interesting and completely frightning. I’ve written software for small scale autonomous vehicles and there is alot that goes into making a vehicle run a course where you know what to expect. Now throw in other vehicles, weather, potholes, and hundreds of other unexpected event sources and the code can become exceedingly complex.

Creating software that can deal with the unexpected 100% safely is nearly impossible. Every software engineer is used to getting a bug report where you go “how in the heck did you do that?” I would expect autonomous cars to be the same. My real concern is that some bug will be found and then there is patch pushed to 100s of thousands of vehicles that fix an uncommon issue, but impacts a major component. I’ve seen hundreds of minor bug fixes that have had huge cascade failures. In business software a couple minutes to a couple hours might be very bad for the business, but can you imagine if a half million cars all the sudden have their accelerator or avoidance algorithms messed up by a “minor” bug fix?

 

[ShrodingersCat]

Having been a software engineer for 20+ years I find the idea both interesting and completely frightning. I’ve written software for small scale autonomous vehicles and there is alot that goes into making a vehicle run a course where you know what to expect. Now throw in other vehicles, weather, potholes, and hundreds of other unexpected event sources and the code can become exceedingly complex.

Creating software that can deal with the unexpected 100% safely is nearly impossible. Every software engineer is used to getting a bug report where you go “how in the heck did you do that?” I would expect autonomous cars to be the same. My real concern is that some bug will be found and then there is patch pushed to 100s of thousands of vehicles that fix an uncommon issue, but impacts a major component. I’ve seen hundreds of minor bug fixes that have had huge cascade failures. In business software a couple minutes to a couple hours might be very bad for the business, but can you imagine if a half million cars all the sudden have their accelerator or avoidance algorithms messed up by a “minor” bug fix?

True, but there are plenty of examples of systems that cannot afford major or even minor bugs already in the real world. The level of testing you do before uploading a patch should be directly proportional to the level of disaster potential if it all goes belly up.

I really think that the reason people are afraid of this is the lack of control they will have. But human control of cars is far more dangerous than a computer. This would result in far less road deaths.

The hacking potential does exist, but it depends on whether the self-driving cars operate autonomously or as part of a network. The network could get hacked, potentially. That would be bad. But if each car is autonomous it’d be like hacking the PCs of everyone in a city simultaneously. If someone hacks the occasional car it’d still be safer than humans driving.

 

[TCEL]

This probably seems like a ridiculous question, but in this case, is safety worth the price of independence?

 

[Usige]

True, but there are plenty of examples of systems that cannot afford major or even minor bugs already in the real world. The level of testing you do before uploading a patch should be directly proportional to the level of disaster potential if it all goes belly up.

Those systems are also in imited deployments and run into the millions of dollars. When most products are commercialized, the efforts to cut costs are generally done in areas that have the lowest chance of happening. Adaptive systems get replaced with more static systems at significantly lower costs.

My point is not that it’s not possible, but rather their are tradeoffs that are made in order to bring products to market at reasonable prices. The multimillion dollar test platforms will have components stripped or lower cost components substituted to get them to market. I can grantee you that someone will do a risk analysis and start requiring things to be stripped to keep costs down.

I really think that the reason people are afraid of this is the lack of control they will have. But human control of cars is far more dangerous than a computer. This would result in far less road deaths.

Loss of control is likely part of it, but I think for many it is because they have had enough technology failures that they are leary of anything they have no chance of correcting. Technology is not perfect. A few years ago I was going down the interstate at 75 (yes that’s the speed limit in that area). All the sudden a warning light on my dash turned on and the throttle became non responsive. I now had to figure out how to get to a safe spot when I had no way to control my speed. Once I had it towed to the shop and diagnostics ran on it they said it was a 4 dollar sensor that went bad. Since the sensor went out the engine control unit decided the safest thing was to stop accepting throttle (name removed by moderator)ut. Total cost to replace a faulty sensor was around $700 between the tow, diagnostics, and labor.

Sounds reasonable right, but I use a similar rotation sensor in some home projects. Sometimes they fail completely as with my car, but other times they simply send incorrect data. Software that can make prudential judgements between something failed versus just feels wrong require significantly faster processeors, more memory and more complex software. That means increased cost and greater chance for unexpected interactions.

Having worked with hardware software interfaces and various types of automation, I have seen enough failures to be leary of giving complete control without the ability for human correction.

The hacking potential does exist, but it depends on whether the self-driving cars operate autonomously or as part of a network. The network could get hacked, potentially. That would be bad. But if each car is autonomous it’d be like hacking the PCs of everyone in a city simultaneously. If someone hacks the occasional car it’d still be safer than humans driving.

Hacking the system is fairly low on my list of concerns. The chance that they are completely autonomous is fairly low anyway. Most are going to have GPS, (name removed by moderator)ut from traffic monitoring and management systems, etc. For what people will be paying they will want it to be able to route around traffic slowdowns and chose alternative routes. That will require (name removed by moderator)ut that doesn’t simply replicate the mark 1 human eye.

 

[Tarpeian_Rock]

I used to see drawings of this on the covers of “Popular Science” and “Popular Mechanics” back in the 1950’s and 60’s. Don’t think it’s any closer now than it was then, despite the advances in technology. Something like 60-70% of our bridges are deficient in some way. The Interstate Highway System, dating from the Eisenhower Administration, is ageing rapidly. Hacking. Unbelievable traffic congestion. Liability and litigation issues beyond belief. Remember all those “personal helicopters” that were coming too?

 

[adowcday]

I used to see drawings of this on the covers of “Popular Science” and “Popular Mechanics” back in the 1950’s and 60’s. Don’t think it’s any closer now than it was then, despite the advances in technology. Something like 60-70% of our bridges are deficient in some way. The Interstate Highway System, dating from the Eisenhower Administration, is ageing rapidly. Hacking. Unbelievable traffic congestion. Liability and litigation issues beyond belief. Remember all those “personal helicopters” that were coming too?

I work a block from Google headquarters and we see driverless cars every day and encounter them on our roadways. Not some futuristic dream at all.

 

[Tarpeian_Rock]

I work a block from Google headquarters and we see driverless cars every day and encounter them on our roadways. Not some futuristic dream at all.

I’d argue that Google headquarters is neither the world the rest of us live in, nor represents the real-world driving challenges the rest of us face on the way to work.