Driverless cars by 2025: will it happen?

[Joe Frickin' Friday]

My brother has made a bet with his friend:

 

Paul owes me dinner if, before August 20, 2025, there is a self-driving car in commercial production anywhere in the world and available for purchase in the US, said car able to autonomously and legally travel from one US address in an urban or suburban setting to another, while any seated driver (whether or not attentive to road conditions) is 100% passive beyond providing the destination address. Otherwise, I owe Paul dinner.

 

Do you think this is going to happen? Why/why not?

 

My position is that it's not going to happen.

 

Commercial airliners currently have an autoland feature, which will put the plane on the runway and brake to a stop without any intervention by the pilot (this feature came in real handy recently). However:

 

• the flight environment during approach/landing is far more simple and predictable than the driving environment;
  
• the task is very simple (keep the plane on this path at this speed, flare near ground, hit the brakes); and
  
• a professional pilot is present and ready to take over if the autoland starts behaving erratically.

In contrast, the driving environment requires consideration/judgment of a huge array of visual input, some of which is difficult for a computer comprehend the significance of; autonomous vehicles are turning in impressive performance during tests, but always with an aware, professional driver present to take over if/when the system chokes. If autonomous vehicles are to be truly useful and live up to their name, they will need to allow a driver to input a destination and then drift off to sleep for the entire journey; I don't see that being possible within the next ten years.

 

Apart from that, IMHO liability in the event of damage/injury/death is a difficult (even show-stopping) problem. If the software is driving the vehicle, who's responsible for compensating injured parties in a crash? Cars have assistive devices now, but they are explicitly understood to be just that: they are there to assist a driver during occasional lapses of attention, but the driver is expected to maintain attention to the best of his ability and will be held responsible for any crashes on his watch. If you're in a self-driving car and it plows into a street fair, who pays for damages? The driver? Ford? Nobody?????

 

If you were liable for damages caused by an autonomous vehicle in which you were an occupant, would you be willing to ride in it, even if (statistically) the car was a better driver than you were?

 

I don't think any car manufacturer would be willing to accept liability if their self-driving car crashes and hurts people, and I don't think drivers will be willing to accept liability if a self-driving car they're in crashes/hurts people either. With no market for autonomous vehicles and no one willing to serve it, I don't see it happening any time soon, certainly not in the next ten years.

 

What do you think?

[Jake]

- Autonomous car testing has been made legal in a number of states.

- Insurance companies will insure just about any liability if the actuaries ok it.

- A driverless car can be owned by a corporation and used by many monthly fee-paying urban commuters, amortizing the cost of liability insurance across many customers.

 

I think the liability obligation can be easily overcome, and this is an issue that has to have been well anticipated.

 

I would want to be on the side of it happening within 10 years.

[Danny caddyshack Noonan]

Based upon the number of texting and phoning drivers here, that's already a done deal. Good thing the RT mirrors don't incur over-use wear.

[smiller]

There will be a number of legal issues to work out but I don't see this as being a significant impediment overall. Liability/legal costs will, one way or another, be built into the price of the vehicle by manufacturers, and any driver culpability (limited though it will be) will be handled by insurance.

[Glenn Reed]

Possible unintended consequences have also been anticipated...

 

[John Ranalletta]

I can't see how can a driver-less car operate safely w/o "knowing" the location, trajectory and speed of every other vehicle in its vicinity. So, how can it function on open roadways until EVERY other vehicle, driver-less and otherwise are equipped with transponders, operating in a automated, GPS network? Is that possible by 2025? I don't think so. I'm not paying to put a transponder in my 1988 F150 so someone else can use a driverless car.

[Guest Kakugo]

I fully expect self-driving earth moving equipment to become available in a matter of a few years: Komatsu has already tested an autonomous excavator/dozer to use in landfills with customers in the US and test results were very satisfactory, with just the right amount of bugs you'd expect from such an advanced technology.

There were also prototypes of heavy mining equipment in the works, chiefly aimed at customers in Australia and South America, but the commodity carnage of the past year has forced Komatsu to give these projects a lower priority.

 

But these machines are made to be used in a confined environment where installing additional beacons to help navigation and other tasks is economically feasible. Also, Komatsu-trained technicians and engineers will be at hand to deal with any problem and, at least initially, will have a direct line with the R&D center in Japan.

 

Cars face many more challenges. Insurance and liability can be probably sorted in a reasonable time, though lawsuits are to be expected at least for the first five-six years of operations.

But what about unexpected problems?

Say one programs his car to take him to a business meeting. While on the highway, signs indicating a detour ahead appear. How does the car deal with those? Will it shut down to recalculate a route (through GPS), be "smart" enough to follow road signs meant for human drivers or...?

Also, can driverless cars deal with poor road maintenance? Will they automatically slow down to avoid passengers been rattled to the bone? Will they avoid potholes without invading the other way to avoid bending a rim?

 

I am sure given time and resources these problems can be solved. But the car industry presently faces huge problems: massive overcapacity and too much reliance on low interest financing and overleveraged consumers to conitnue spending. Investing the massive sums needed does not rank very high, especially given lwgal issues haven't been worked out yet. Why bother with offering a self-driving car if you, and not the owner, will be sued for millions?

[smiller]

I can't see how can a driver-less car operate safely w/o "knowing" the location, trajectory and speed of every other vehicle in its vicinity. So, how can it function on open roadways until EVERY other vehicle, driver-less and otherwise are equipped with transponders, operating in a automated, GPS network? Is that possible by 2025? I don't think so. I'm not paying to put a transponder in my 1988 F150 so someone else can use a driverless car.

Driverless cars operate even now in normal traffic. Certainly the whole system would be much more effective if communication with other cars was common, but it's not necessarily a base requirement.

 

My guess is that one of the first real-world applications might be in large over-the-road trucks that typically leave a terminal located very close to an Interstate exit (as they typically are) and travels on Interstate highways to another terminal that is also located near an exit. This represents a relatively 'easy' driverless task and would also be more likely to provide some kind of direct cost benefit (saving the cost of a driver, being able to operate 24/hrs. day, etc.) and thus providing a commercial incentive for someone to do it.

 

But a private vehicle traveling from anywhere to anywhere in urban areas? Certainly that is coming but seeing it by 2025 may be a stretch.

[John Ranalletta]

Two driver less cars each unaware of the other meet at speed on a one-lane road. Which one is programmed to drive off the cliff?

 

If the question is "fully automated", driver less roads, not by 2025; but if it's "in controlled situations where variables are eliminated", certainly.

[smiller]

Two driver less cars each unaware of the other meet at speed on a one-lane road. Which one is programmed to drive off the cliff?

In mixed traffic an autonomous vehicle will look at the road and other vehicles and decide what to do, and if it is too confused to continue it ultimately will stop as safely as possible and let the human-driven cars make the decisions. So in the case above the driverless car would see the opposing traffic and if it had no way of exiting the road or pulling over it would likely just stop, and the human driver would handle the situation.

[roadscholar]

This article is a couple of years old.

 

http://www.autoevolution.com/news/a-short-history-of-mercedes-benz-autonomous-driving-technology-68148.html

[Bob Palin]

What will driverless cars do about deer (etc)? I regularly drive on rural deer infested roads and I know where they often are, I slow down for those areas, but I have still hit 4 and sometimes in poor visibility slow way down for what turns out to be a bush. Even when they are visible their actions are unpredictable, how slow must a driverless car go if it can detect one?

 

There are about 1.25 million deer-auto collisions in the USA each year costing about $4 billion dollars and killing 200 people (numbers vary).

[Glenn Reed]

Bob,

 

It's possible these cars will have FLIR (Forward Looking InfraRed) sensors in addition to the optical sensors we think of right now. ( Example Here)

 

The web page for Google Cars - How It Works specifically mentions lasers, radars and cameras in the Sensors section, so they might not be using FLIR now, but I wouldn't doubt it could be incorporated.

 

I've totaled one car in a deer strike, and had numerous near misses, so I understand your concern.

[Bob Palin]

Even with FLIR it almost certainly won't be able to tell the difference between a deer on the side of the road, which it needs to be very cautious about, and a cow which it can pretty much ignore if it is not moving. Will people tolerate slowing to 10mph to pass a cow? Presumably if an animal is actually in the road it will stop and disengage, waking the sleeping 'driver'. Hopefully the car behind will notice (self driving or not), could be a nice source of road rage when a driverless car stops for a deer that scampers away and is followed by a human driven vehicle that doesn't even see the animal.

 

I just read another article about the subject which brought up the issue of new drivers. Somebody who spends the majority of their driving life in auto mode might suddenly find themselves in control when the situation is too complex for the computer to figure out, and they would have limited prior input and little experience to deal with the situation.

 

(http://www.bbc.com/future/story/20150824-what-may-be-self-driving-cars-biggest-problem)

 

I really want a driverless car, even one that only drove itself on the freeway, it would improve my life quite a lot. (Back and forward between Torrey and Cedar City is getting old)

[Joe Frickin' Friday]

What will driverless cars do about deer (etc)?

 

Don't know, but they seem to handle

 

I expect some of their ability to handle the random behavior of animals will come down to the designers making tradeoffs between the probability of arriving safely versus the probability of arriving in a reasonable amount of time. In other words, you'd be relying on the subjective valuations of time and safety as dictated by the designers of the cars - and those subjective valuations may not match your own.

[Sonor]

Interesting topic and lots of good thoughts posted. I believe we will see driver less cars in controlled areas as a matter of public transportation. By what year, who can say. The technology may be available but there is the very real cost of implementing. As with all things cost vs. return must be analyzed before investment takes place. One of those factors is public perception as that will directly affect the use of the vehicles. If the public has little regard or (to borrow a hunting term) is spooked of the idea, then use of these vehicles will be small which means little pay back which translates into no investment.

 

I remember back in college I was bar tending in a restaurant/bar for the happy hour crowd. A guy was sitting at the bar and I got talking with him. He told me that he was from IBM engineering. Long story short he said, "You would not believe the products that we have waiting to be released and they are scheduled out for the next fifteen years." I asked why they were not released now to which he responded, "Because the public is not ready for them. They would not even touch them." This brings into light the fact that there is a mass education that comes as a form of peer pressure and without it, technology would be still with the green screens.

[John Ranalletta]

More insight from an technology insider: http://www.autoline.tv/journal/?p=39128

 

2025-2035

 

[Joe Frickin' Friday]

One of those factors is public perception as that will directly affect the use of the vehicles. If the public has little regard or (to borrow a hunting term) is spooked of the idea, then use of these vehicles will be small which means little pay back which translates into no investment.

 

This was a concern I expressed to my brother: people are irrational. You can tell them that this car is a far better driver than they are, you can even back up your assertion with a million miles of crash-free testing, but a lot of folks will still be weirded out by the idea of ceding control of their immediate physical safety to an unseen, unknowable entity.

 

The interview that John linked to points out an additional problem associated with the intermediate step. Imagine it's 2025, and you're shopping for a new Honda Civic. There's an auto-driving version of it that sells for an additional $2500, but the salesperson tells you "this car can take you to your destination all by itself, but you still have to pay attention, you know, in case it screws up." Would you pay for that upgrade? Would you use it? Would you trust it with your life, given that little seed of doubt? I think that before this technology will be widely accepted by the market, it will have to be so good that the salespeople will be instructed to say "you can take a nap while this thing drives you to your destination." I doubt that will happen by 2025.

[smiller]

This was a concern I expressed to my brother: people are irrational. You can tell them that this car is a far better driver than they are, you can even back up your assertion with a million miles of crash-free testing, but a lot of folks will still be weirded out by the idea of ceding control of their immediate physical safety to an unseen, unknowable entity.

Yep, you can bet that when a driverless car has its first major or fatal accident it will be a cause célèbre... nevermind that the accident rate might still be a fraction of what human drivers would experience, the mere fact that it's automated will negate all that in the public's mind.

 

I think that before this technology will be widely accepted by the market, it will have to be so good that the salespeople will be instructed to say "you can take a nap while this thing drives you to your destination." I doubt that will happen by 2025.

Yes, but there might still be a significant amount of value for some people even with a 'semi-automated' system, for example you run into traffic gridlock (or even during many people's normal commute if they live in problem urban areas) it would sure be nice to let the car handle the beep-and-creep so you could read a newspaper or get some work done, even if you are required to remain in the driver's seat and keep an eye on things. If I commuted in that kind of environment I would pay a lot for that option.

[Rocer]

After listening to Urmson's (head of Google's driverless car prog.) talk in which he "hopes" driverless cars will be in place before his 11 1/2 year old son gets his driver's license in 4 1/2 years I'm placing my money on the less than 10 years side of the bet.

 

Manufacturers, vehicle owners/drivers and all parties related to the transpiration industry are already assuming liability as soon as their vehicles are rolling. Heck, they don't have to even roll before there could be a liability scenario such as a parked car catching fire from something like an electrical short. So I think that the question of liability is really a non-issue. We're already sharing it now.

 

In Urmson's talk he mentions that 34 people will have died on American roads during his 15 minute talk. All the driverless vehicle industry has to do is significantly lower the death rate/injuries by vehicle accidents to get both the insurance industry and the driving public to accept this technology. I assume (a dangerous term I know) that significant reductions in accident rates will result in significant reductions in insurance rates. The market will follow those reductions, not all of the market but a large enough segment to make acceptance of the driverless technology feasible.

 

The question remains: What will we have for dinner?

 

[John Ranalletta]

On a related note from autoextremist.com:

J.D. Power says that 20 percent of new vehicle owners have never used sixteen of the 33 technology features in their cars. The 2015 J.D. Power Driver Interactive Vehicle Experience Report evaluates the typical owner's experience in the first 90 days of ownership of a new car or truck. What does this survey really say? That the auto manufacturers are loading up vehicles with expensive technology in a technological arms race that adds cost upon cost and to what end? Bragging rights at the auto shows? We're talking billions of dollars here in the most egregious case of technology for technology's sake that this industry has ever seen. But then again this business has been burdened with rational thought only for fleeting moments in time throughout its history. And this isn't an "age" thing either in case you're wondering, because Millennials don't bother - or look to - the center stacks of their cars for technological sustenance. They have their smart phones and tablets for that. When will this absurdity stop? Not anytime soon, you can count on that. -PMD

[EffBee]

Mitch, do not underestimate the willingness of governments to legislate liability out of certain activities and/or to pass legislation forcing the citizenry to purchase something it does not want.

[Guest Kakugo]

On a related note from autoextremist.com:

J.D. Power says that 20 percent of new vehicle owners have never used sixteen of the 33 technology features in their cars. The 2015 J.D. Power Driver Interactive Vehicle Experience Report evaluates the typical owner's experience in the first 90 days of ownership of a new car or truck. What does this survey really say? That the auto manufacturers are loading up vehicles with expensive technology in a technological arms race that adds cost upon cost and to what end? Bragging rights at the auto shows? We're talking billions of dollars here in the most egregious case of technology for technology's sake that this industry has ever seen. But then again this business has been burdened with rational thought only for fleeting moments in time throughout its history. And this isn't an "age" thing either in case you're wondering, because Millennials don't bother - or look to - the center stacks of their cars for technological sustenance. They have their smart phones and tablets for that. When will this absurdity stop? Not anytime soon, you can count on that. -PMD

 

It's actually worse than that.

Most cars are "engineered to fail". Manufacturers know exactly how long components will last and use this to build cars that after a given number of years/miles (usually just when warranty runs out) become extremely expensive to keep on the road, hence incentivate owners to get a new one.

My mother's Mini is one of the most egregious examples I know of. Not only it's becoming really expensive to keep on the road (especially considering it's a basic model and hence not worth much), but finish is starting to look bad as well. The interior unpainted plastics are starting to fade, it uses a lot of oil, it eats sensors for lunch etc. Luckily my brother and I can help keep repair bills down, but we are not tooled to take a transmission off...

Mind it's your normal "Old Lady's Car", not a hot rod owned by a boy racer.

 

All this unnecessary technology adds parts that may and will go wrong, requiring expensive repairs down the road and hence driving people to keep changing cars far more often than they would.

 

There's a reason why Toyota and Mitsubishi trucks have become so widespread here despite being notorious gas guzzlers: being engineered for Third World markets they are built to last and without frills.

[greiffster]

Most cars are "engineered to fail". Manufacturers know exactly how long components will last and use this to build cars that after a given number of years/miles (usually just when warranty runs out) become extremely expensive to keep on the road, hence incentivate owners to get a new one.

 

I think you are giving the engineers way too much credit. Sure, they have plenty of historical data on which/what materials or components fail, break, fade, whatever. And maybe they don't have incentive to make them last longer, never mind the cost. But the idea that the engineers are designing a part to fail 4 days after the warranty expires or at 36,004 miles, on purpose, well I guess I'm just not buying it.

[Dennis Andress]

I like driving and really don't look forward to this. But it sure would be nice when I push the little button on the remote if the dang car would back out of the parking space and pick me up.

[John Ranalletta]

I like driving and really don't look forward to this. But it sure would be nice when I push the little button on the remote if the dang car would back out of the parking space and pick me up.

..or, send the car to the grocery store to p/u my online order.

 

OTOH, wouldn't fully driverless cars shift ownership from individuals to suppliers? Why wouldn't stores, Amazon, etc. have fleets of self-drivers to driver products?

[OoPEZoO]

I like driving and really don't look forward to this. But it sure would be nice when I push the little button on the remote if the dang car would back out of the parking space and pick me up.

 

I like driving too.....except for the 10-12 hours of commuting hell I deal with every week. I'd buy a self driver in a second if I could just space out or take a nap during that time. When I get home, I would still have something fun in the garage with a 3rd pedal for my evenings/weekends.

 

Besides, think about how much fun you could have with a bunch of self drivers around. I'm sure they would drive super conservative. I'll bet you could easily weave your way through traffic at a high rate of speed while laughing every time one of of them slams on the brakes and disturbs the occupants. Not exactly the "right" thing to do, but I'll bet it would be amusing.

[Glenn Reed]

Besides, think about how much fun you could have with a bunch of self drivers around. I'm sure they would drive super conservative. I'll bet you could easily weave your way through traffic at a high rate of speed while laughing every time one of of them slams on the brakes and disturbs the occupants. Not exactly the "right" thing to do, but I'll bet it would be amusing.

 

There you go thinking like a hooligan again, eh Keith?

[John Ranalletta]

Imagine the fun kids will have darting or throwing a paper cup into the path of driver-less cars just to see them slam on the brakes.

[Bill_Walker]

Dexter Ford, in the "Megaphone" column in the October 2015 issue of Motorcyclist (not yet available online, as far as I can tell) argues that self-driving cars will make the roads safer for motorcyclists, because the cars will not be susceptible to inattention and distraction, and have many more sensors than humans have eyes. I think he's got a point.

 

Just as a data point, Google has over a million miles of accumulated testing of driverless cars. They've been involved in 11 accidents, IIRC, ALL of which have been the fault of human drivers. I suspect at least some of these have been due to what I'd call the "Google Gawker" effect, since Google's test cars are very conspicuous.

 

I think they have the potential to revolutionize our transportation system. Why even own a car, if you can use your smartphone to summon one to pick you up and take you to your destination at any time? And if, therefore, cars are more "shared", we'll need to devote a lot less space to parking, since autonomous cars can be driving somebody else around instead of sitting in a parking lot for hours. Imagine all the space that will free up for other uses! There's lots of work going on in this area.

 

I do have some concerns:

 

How will autonomous cars handle driving in snow, when lane markings may not be visible and traction may be tenuous?

 

If there are conditions where humans must take over, how, in the future, will they learn how to do so? They'll have virtually no experience operating a car.

 

What will we do with all the unemployed drivers of trucks, taxis, Uber, Lyft, livery services, airport shuttles, etc.?

[Mike]

Just as a data point, Google has over a million miles of accumulated testing of driverless cars. They've been involved in 11 accidents, IIRC, ALL of which have been the fault of human drivers. I suspect at least some of these have been due to what I'd call the "Google Gawker" effect.

 

I know that the technology is evolving, but that seems like a lot of collisions. In 45 years of driving I've had two collisions. One was a guy turning in front of me--not my fault and almost impossible to avoid--and another that was my fault (turned into a driveway not knowing it was iced over, thereby whacking into a tree). I'd guess I've averaged 20,000 miles a year.

 

I have, however, made a ton of mistakes...being distracted, misjudging speed, getting angry, etc. Somehow, that gray computer inside my skull has been able to compensate and react. At times it's been catching the eyes of another motorist, other times just some innate sense of the need to compensate. I wonder how long it'll be until the computers get to that point.

 

[Joe Frickin' Friday]

I know that the technology is evolving, but that seems like a lot of collisions. In 45 years of driving I've had two collisions. One was a guy turning in front of me--not my fault and almost impossible to avoid--and another that was my fault (turned into a driveway not knowing it was iced over, thereby whacking into a tree). I'd guess I've averaged 20,000 miles a year.

 

I have, however, made a ton of mistakes...being distracted, misjudging speed, getting angry, etc. Somehow, that gray computer inside my skull has been able to compensate and react. At times it's been catching the eyes of another motorist, other times just some innate sense of the need to compensate. I wonder how long it'll be until the computers get to that point.

 

I read a book a few years ago: Traffic: Why We Drive the Way We Do (and What It Says About Us). In it, the author suggests that collisions are a statistical rarity, which makes them a rather unreliable indicator of a driver's skill level. After all, a driver may go 45 years with no collisions at all, but that doesn't mean he's infallible; luck has at least a small part in everyone's fate. Instead, he suggested that the number of close calls was a better metric; as you've observed, close calls tend to happen far more often than collisions, and each close call represents a roll of the dice that could have, but for a tiny bit of luck, gone the other way.

[Joe Frickin' Friday]

Elon Musk thinks driverless cars are just months away.

[tallman]

Mitch,

I think there are a lot of drivers who have few collisions, yet cause numerous situations for other drivers to deal with on a continual basis.

People react to the poor (or deliberate) driving choice these folks make and avoid them and/or cause a ripple effect for others depending on traffic conditions.

Driver one goes blithely (or malevolently) on their way with

a host of other problems in their wake.

We have a couple intersections on our "truck route" bypass around the city.

I'm commuting in the dark, early, and often see big rigs/logging trucks who deliberately run red lights because they have a head of steam up, there is no enforcement (live or camera) and they are bigger.

This is 1-2-3 seconds, or more, after light has turned green for cross traffic.

I wonder how driverless vehicles will deal with a situation like this because without prior knowledge and intentional waiting period before moving into intersection all seems ok and you expect those trucks to stop, not continue.

However, there are many places I can see a driverless vehicle operating safely and efficiently.

Look forward to that, I think.

[Joe Frickin' Friday]

I'm commuting in the dark, early, and often see big rigs/logging trucks who deliberately run red lights because they have a head of steam up, there is no enforcement (live or camera) and they are bigger.

This is 1-2-3 seconds, or more, after light has turned green for cross traffic.

I wonder how driverless vehicles will deal with a situation like this because without prior knowledge and intentional waiting period before moving into intersection all seems ok and you expect those trucks to stop, not continue.

 

Watch at 12:12 as a bicyclist blows a red light; the autonomous car safely waits for the bike to cross in front of it. As long as the car has a clear view to see/identify moving objects and can assess their speed/direction - whether it's a bicyclist jumping across against the light, or a logging truck approaching from farther back at 50 MPH - it seems like it has a good shot at responding safely.

[tallman]

Impressive technology.

Hopefully all will work as intended.

Wonder if speeder technology (radar jammers) can blind it?

[Joe Frickin' Friday]

Impressive technology.

Hopefully all will work as intended.

Wonder if speeder technology (radar jammers) can blind it?

 

From the video, it sounded like they were using some kind of laser scanning instrument; I think that's the little round dark thing often seen on the roof of Google's autonomous vehicles. Don't know what wavelength of laser they're using.

[smiller]

Wonder if speeder technology (radar jammers) can blind it?

There is a lot of redundancy and cross checking in the visual reference systems, I would doubt that they could be blinded by any system they are likely to experience in the real world. What if a LEO hits it with a lidar gun? Etc. Maybe if someone with malicious intent had technical details about how the system worked they could come up with a jamming device, but autonomous vehicles use laser measurement and/or RF and visual references so you'd have to jam a lot. And even at that the worst you'd probably be able to do is cause the vehicle to come to a stop.

[Osoyoos]

I doubt in my life time there will ever be a computer that is so reliable it can safely drive us around.

To errr is human....to really screw up requires a computer.

 

Ps. My first post and did not know where to put it, so here seemed fine.

[Glenn Reed]

Ps. My first post and did not know where to put it, so here seemed fine.

 

Osoyoos - Welcome!

 

You jumped right in to Other Topics, and that's fine.

 

If you still want to make an intro post and tell us a little bit about yourself and your bike(s), that's done in the Motorcycle Talk forum.

 

Again, welcome!

[John Ranalletta]

Google's self-driving cars can't handle bicycle track stands. Here.

One Austin-based cyclist reports an encounter where one of the autonomous cars was comically unsure of what to do when it spotted him doing a track stand at an intersection. Every time his bike moved even slightly, the car would lurch forward and promptly hit the brakes. Nothing happened beyond some good laughs, but it was clear that Google's self-driving code didn't know how to handle a not-quite-stationary bike.

 

Stop-Go-Stop-Go-Stop.....

[Antimatter]

Google's self-driving cars can't handle bicycle track stands. Here.

One Austin-based cyclist reports an encounter where one of the autonomous cars was comically unsure of what to do when it spotted him doing a track stand at an intersection. Every time his bike moved even slightly, the car would lurch forward and promptly hit the brakes. Nothing happened beyond some good laughs, but it was clear that Google's self-driving code didn't know how to handle a not-quite-stationary bike.

 

Stop-Go-Stop-Go-Stop.....

 

I see a lot of human drivers do this, too.

[Joe Frickin' Friday]

Google's self-driving cars can't handle bicycle track stands. Here.

 

Funny. The primary source has more entertaining detail.

 

That's the kind of situation - like driving behind a car with a mattress lashed to its roof with kite string, or being next to a truck whose tire sounds like its about to disintegrate - where responding appropriately depends on a higher level of cognition than I think computers and software are currently able to deliver. A meat-based driver would see the trackstanding cyclist, comprehend his essence/intent, and know it was safe to proceed through the intersection. OTOH, if the computer makes up for this shortcoming by tirelessly keeping track of the closest two dozen moving objects and managing to not hit any of them, then maybe it's a net positive.

[szurszewski]

Google's self-driving cars can't handle bicycle track stands. Here.

 

Funny. The primary source has more entertaining detail.

 

That's the kind of situation - like driving behind a car with a mattress lashed to its roof with kite string, or being next to a truck whose tire sounds like its about to disintegrate - where responding appropriately depends on a higher level of cognition than I think computers and software are currently able to deliver. A meat-based driver would see the trackstanding cyclist, comprehend his essence intent, and know it was safe to proceed through the intersection. OTOH, if the computer makes up for this shortcoming by tirelessly keeping track of the closest two dozen moving objects and managing to not hit any of them, then maybe it's a net positive.

 

Are you really asserting that most meat based driving systems will respond appropriately to your three situations? I think most drivers can comprehend the not quite stationary track stand, though here in Portland I often see drivers confused by it (especially when the cyclists is doing a crappy job of staying put), and I suppose a lot of people would give more following distance to the mattress (though I don't think MOST would), but I bet few would notice the sound of the tire and fewer would both recognize what it was and take the appropriate steps.

[Joe Frickin' Friday]

Are you really asserting that most meat based driving systems will respond appropriately to your three situations?

 

Perhaps not most, and that's a point my brother made: to justify their existence, self-driving cars don't need to better than most meat-based drivers, they just need to be better than average.

 

I expect that eventually the car's software could be updated to recognize:

 

A) this is a four-way stop,

B0 that thing arrived at its stop after I arrived at mine (so I have right-of-way), and

C) that thing is a bicycle/bicyclist and can't possibly accelerate fast enough to get in my way.

[Harry_Wilshusen]

Mitch, do not underestimate the willingness of governments to legislate liability out of certain activities and/or to pass legislation forcing the citizenry to purchase something it does not want.

 

Amen to that.

[Bill_Walker]

Just as a data point, Google has over a million miles of accumulated testing of driverless cars. They've been involved in 11 accidents, IIRC, ALL of which have been the fault of human drivers. I suspect at least some of these have been due to what I'd call the "Google Gawker" effect.

 

I know that the technology is evolving, but that seems like a lot of collisions.

 

Google's contention on this (at least re collision statistics, not your personal experience) is that non-injury collisions are generally under-reported, whereas Google is reporting ALL of their collisions. I think that's a reasonable contention.

 

I'd also contend that as motorcyclists, we tend to be much more aware of other traffic than your average car driver, and are therefore likely to have lower CAR accident rates (seems to me I've seen a study to that effect). Your personal accident rate is therefore likely lower than average.

[Bill_Walker]

My data was out of date. It's two million miles, and 14 accidents caused by human drivers (one of whom plowed into the back of a stopped Google car at 17 mph without braking).

 

Also, there's this:

"Their algorithms don't get bored, tired, or angry, and their 360-degree laser sensors mean they don't have blind spots. Just as importantly, they're seemingly programmed to always err on the side of excessive caution.

 

As a Mountain View, California, resident — who frequently interacts with these cars on the roads near Google's headquarters — wrote in June, "Google cars drive like your grandma — they're never the first off the line at a stop light, they don't accelerate quickly, they don't speed, and they never take any chances with lane changes (cut people off, etc.)." The resident described how the cars wait a few seconds after a pedestrian has completely cleared a crosswalk before beginning to turn through it.

 

And Google is clearly taking cyclists and pedestrians into account in the design of their cars' algorithms. It's specifically upgraded its software to navigate chaotic city streets, and earlier this year it patented a way for its cars to interpret bikers' hand signals."

 

See Why Google's self-driving cars will be great for cyclists and pedestrians

[Harry_Wilshusen]

The google car is controlled by artificial intelligence. It's logical in a machine way. Humans are for the part are logical in a human way, (which some will say is not logical). Machines see black and white, humans see shades of gray.

 

 

http://www.nytimes.com/2015/09/02/technology/personaltech/google-says-its-not-the-driverless-cars-fault-its-other-drivers.html?_r=0