Do you think self driving cars will become common in the future?

It would be nice if you had to actually demonstrate some level of competence in order to get a driver's license.

 

...but they don't have public transportation out in the county.

 

Yes it's inevitable and I CANNOT FREAKING WAIT. It will totally change our lives in huge ways that most people now simply lack the ability to understand. It will wildly increase efficiency of vehicles, increase safety, reduce congestion, decrease travel times. Absolutely every thing about driving that sucks will be rectified by self driving cars.

It is also totally inevitable. No doubt in mind about that.

 

I really think that self-driving cars are one of those things like household robots and flying cars that people are always predicting but never happen. The technology exists, sort of, but the practical issues will prevent it from being available to the public in my lifetime. There will be prototypes, just as there have been prototype "flying cars," but nothing you can buy and take out on the street as a regular citizen.

In particular, some degree of artificial intelligence would be needed for anything other than slotted freeways for big slot cars, and in spite of all the hype, nothing remotely resembling real A.I. is anywhere near development. Computers can play chess damn well, but they do it by brute force negotiation of move trees and sophisticated heuristics; they can recognize spoken words pretty well; they can search data bases. But anything like actual thinking is still totally beyond them. And a self-driving car that is not confined to a well-defined alley of some sort, that must deal with other unconfined cars, pedestrians, unmapped obstacles, etc., will need some intelligence.

The only way self-driving cars will work my my lifetime is within a structured construct where every car is physically confined to a defined lane and every lane and obstacle is mapped, and pedestrians are excluded, or are allowed to cross only under the control of signals which physically prevent them from crossing against the signal.

 

It's essential to look at what is happening in the commercial aviation industry to understand the technology path for Self Driving cars. Also the military expansion of UAVs makes it clear that there are more ROE and political hurdles than pure engineering hurdles for autonomous vehicles. The first self driving implementations would be long, straight lanes. Most likely present day HOV like lanes. Once those take root, it would be a slow sustained expansion to more and more roads. That will take a few decades, but that is a really short period of time, being less than one lifetime.

 

Best possible case, I have only a third of a lifetime left. More likely case, I have a sixth of a lifetime left. Very possible case, I have considerably less. I won't see self-driving cars except as concept prototypes, if that.

In commercial aviation, autonomous systems keep the plane flying a straight line, constant altitude, and can even land the plane, I think, but air traffic control keeps the flight path clear of other planes, and makes sure that only one plane at a time has permission to use the runway, and the runway is equipped with transponders so the instruments have a sort of electronic slot to follow. The air traffic controllers provide the intelligent input for the guidance. And there are no pedestrians or puppy dogs jumping out in front of the plane. Occasionally birds get in the way and the results can be disastrous.

The analog for cars would be electronically delimited lanes where nobody is permitted to enter without permission from human controllers watching everything and tracking every car.

 

There are real problems, but I don't think we need AI. For starters, these cars could be limited legally and by code to function only at certain times in certain ways. For example, driving on a highway. That takes very little brain power almost all of the time and a computer is well suited to it. There won't be pedestrians, etc.

This is also why google has a car now with many tens of thousands of miles on it, no accidents (of course it's diligently watched by the driver to make sure it doesn't do something dumb).

I originally thought we'd need set roads, or things installed in the road, but with good sensors and code we won't. It will begin on highways. People commuting to work will be able to surf the net while the car does the stop and go. At least one state (I think Nebraska) has even passed legislation to legalize self-driving cars.

Once this hits the market and a few people start raving about how their 1 hour commute has now been time they catch up on a movie in the car everybody will want it. Not only that, but when it's discovered that fuel economy improves massively (multiple cars could tailgate each other even at very high speeds, hugely decreasing drag), accident rates plummet (no more rear-end collisions for example), and eventually congestion goes way down (e.g. when a light goes green all cars accelerate in unison instead of a delayed one by one thing), the benefits are huge.

The internet was the last great technical marvel that has changed lives and I think automated vehicles will possibly be the next. I do (hopefully) have more than half my life left and expect to see it rolled out at some point.

 

In virtually all cases it could be trumped by a good software and hardware package. We can identify all day ways in which such a system could fail, but we already have many ways in which people do fail. It doesn't have to be perfect to be better. Traffic accidents are a substantial cause of fatalities in the country and I would have no problem putting my life behind the wheel of a computer than some people I know who suck at driving

 

^ Good points. I'm still skeptical.

 

I'll take what ever they have right now.

 

Not on present roadways. The risk is too great. Just imagine a crowded interstate and something goes wrong with one car with hundreds of cars in the immediate vicinity with no one paying attention. Kinda like it is nowadays at rush hour and iphones. We already know that that won't work.

 

All the "insolvable" problems with self driving cars have been, or are being, solved in the aviation realm. A great many problems with emergency conditions (e.g. partial loss of a wing) were thought to be unsolvable, but turn out to be solvable with 100s of man-years of hardware and software engineering. For example, driving safely after a tire blowing out is just a much more complicated control problem, not an unsolvable problem. The difference here is autonomous vehicles would not enter an "unrecoverable" state whereas humans do this all the time. That's one big aspect overlooked.

In the driving world, would the standard being sought be absolutely zero deaths or vastly less than 40,000 a year?

 

I would point out that having a self driving car without a corresponding self-check and monitoring system validating vehicle safety for operation is not happening. Planes don't leave the gate to fly if anything flight critical is not 100%.....and those systems check everything possible. Same would have to be true of self driving cars. They would be vastly better instrumented and have very stringent criteria for operation at all.

 

Daniel:

A self driving car already exists. As other have said, Google has been road testing prototypes for several years now. My understanding is that the car will go from one address to another without input from the drive. The route is controlled by GPS, while lane positioning and avoidance of other vehicles is accomplished with cameras and radar.

You may also be interested to know that airliners are moving away from a ground based traffic control system to a GPS based system. The appropriations bill just passed the US Senate to go live with the system at 35 airports in 2015. The switch to GPS will not only allow more planes to take of and land in a day but also reduce emissions.

 

I didn't say there would not be a self-driving car. I said it would not be available to the public for general use.

 

A little precision is needed here. The move is not away from a ground based air traffic control system. All plans have ATC remaining ground based. Also, navigation has been GPS based for quite some time, so there is no "switch" to GPS. What is changing is that plane location tracking is shifting from ground radar based to aircraft beacon (ADS-B ) based reporting. Specifically, aircraft must be equipped with radios that constantly report not only position, but eventually will report future maneuvers.

Expect the same for self driving cars. GPS and local car-to-car coordination will be integral to this change. Right now, most folks think that self driving means totally self contained. That will not be the case at some transition point in the technology.

 

The GPS used for self-driving cars will have to be better than what's currently used in navigation systems, as will the voice recognition system. If it's no better or cheaper than taking a cab, it's not going to be useful.

It's pretty cool technology, and I'm not saying it's impossible, but we've got a long ways to go before it's practical and commonplace.

 

The GPS used in cars is $5 chip. More of a feature than a precision navigation component. However, the prime use of GPS would be for route planing and destination determination, not "station keeping". What would be part of a self navigating car would be a very low cost Inertial Measurement Unit/GPS. These would be in the hundreds of dollars.

And yes, I agree we have a way to go.

 

They really would have to keep the GPS data file up to date every single day. We've had a bridge out for several years near here and the GPS in my Lincoln will route me right down that road. I'm not up for watching a self driving Toyota Prius try and do a Dukes of Hazard style leap.