Will self driving cars actually work? Really?

You also must consider the viewpoint of the total effect. If 10 children are saved while one is lost due to a big technology shift, does this not matter more than the individual case you are talking about? Every technology shift has advances and regressions. Keep in mind a lot of the avoidance issues you are mentioning have always been present with trains, so we actually have accepted (or been forced to accept) technology advances that create new hazards while reducing previous hazards. With a horrible loss of life at about 40,000 per year in the US, the ethics of manually driven cars needs to be addressed in the same manner.

 

All of those are valid questions and yes is appropriate for the last two. But the opening post was on the "ethics" or workability of driverless cars. That was what I was addressing. I would point out that significant introduction of driverless cars requires huge changes in roads and other infrastructure as well as the cars. Something that will take many decades and will be extremely incremental, but ultimately be safer than what we have now. If it is not safer, it's not going to happen.

 

Cargo airlines can already run without pilots aboard, and will be doing that very shortly. Self-driving cars are almost on the market now. The technology has been around for some time. Making it safer when we are killing 40,000 each year on the road is not too hard to do.

 

Mode S TPDR, TCAS, RA, GPWS, and CAT IIIc won't help when kids and animals suddenly jump out from behind parked cars though. I see A/P for cars limited "mainly" to highways use.

 

I have mixed feelings about the safety, although I've seen so many drivers doing so many stupid things, that a computer would likely be an overall improvement. Breeding the ego and the competitiveness out of drivers sounds like a step in the right direction.

 

Possibly on dedicated "UOV" interstate lanes.

 

This right here is why Americans may invent* the technology, but we will be the last ones to use it. The lawyers will bankrupt anyone who tries to let them loose in the US. There may be exceptions for restricted access roads or important groups, but the general public is not going to have them here, it won't be allowed. You can see the effect in all the various nag screens and nuisance beeps and fancy floor mats that infest our Priuses.

*Or at least help invent

 

Agreed.

 

The car computer can conceivably react faster and make better decisions than the driver, most certainly an impaired driver. I think this is something that we will see come of age in the next ten years or so. Automation if this sort is already in place for a lot more aviation than most people realize.

 

This ultimately means that you will end up with "dumber" drivers. Since people won't actually have to know how to drive, you will have people "in command" of a vehicle who don't' know how to operate it...

how many times has your laptop, smart phone, ipod, ect "glitched out" on you...

I'm a firm believer that the barriers to becoming a legal driver in europe, is the reason they have such better safety record than the US. Heck in the US show up with 20-30 bucks, hop in do a driver test, spend a few hundred on driver education, Bam heres your card go and drive a killing machine...

I think its neat and would have its place on highways but in high density traffic and surface streets, not so much...

 

These are computers we are talking about. Any decision that you can think clearly enough about to write down the way the decision *should* be made, the computer is capable of executing.

My concern is not balancing hitting a animal versus hitting a child, versus being hit from behind. My concern is things that we AREN'T good at describing, like how we see, and differentiate objects on insufficient data. For example driving in snow. I lost a cousin because she made the choice to miss a deer, and ended up hitting a tree instead. No computer would make that mistake.

But here's the thing, we don't have to switch over all at once. We can switch just the worst drivers first. Let's say we get a computer driven car which is only better than 10% of drivers. Switching those drivers over to computer driven cars would be a net safety win. I imagine the first ones deployed will be as taxis, either conventional or specifically for elderly, handicapped or bad drivers (we have many people here driving (illegally) on suspended licenses just because they need to get around).

 

And I missed the really big safety increase. EMPTY cars. Let's say you are your buddies are watching the Superbowl, drinking beer, and you run out. Do you drive to the store to get more (and miss some of the game)? Or do you send the computer-driven car to pick it up, and stay safe at home?

 

50 years is an eternity when it comes to computers. State of the art 50 years ago was piloting a craft to the moon. With memory (including all data) of about 84k bytes, not gigabytes, speeds of 1 Megahertz, not gigahertz.

But let's look at the advantages that computers have: 1) Multiple sensors all inputting in real time (radar, cameras (looking in all directions with NO blindspots), IR sensors, GPS sensors, input from smart roads, traffic reports, etc. 2) The ability to model all objects within range simultaneously, computing velocity, acceleration, etc. 3) Complete knowledge of the cars current operating conditions. 4) Extremely fast control over all aspects of the car. Braking, steering, engine speed, all happen at the speed of electronics, not neurons, and muscles. Humans take up to 1/2 second to start braking from the time they decide to brake. 5) Communication with nearby cars, for example, once a computer has determined that an emergency stop is necessary, it can alert the car(s) behind, so that it can brake at exactly the same time. 6)No distractions. No pretty girls to watch, no texting, no phones. 7) No ego. No need to beat that other car to the next light. No need to impress that guy by 'peeling out'. No peer pressure to drive faster.

Narrow roads with 'change in priority' are a piece of cake, especially when you can call that car at the other end and negotiate who goes first. Pedestrians jaywalking are actually EASIER for computers than humans, since computers don't start with an assumption about what pedestrians are going to do. All those other things are just obstacles that the computer can notice faster, and react to faster, the slower (perhaps) part is identification (i.e. seeing)

 

Neither. It is a waste of both time and fuel to send out your empty car.

Amazon's little flying drone will deliver it faster and with less energy.

 

Nonsense. The liability issue is already solved. If you run a red light and hit someone, you are liable. If you claim that you accelerator stuck, the car company is liable. If the car company can prove you lied, you are liable again. If the car company can prove a component manufacturer is at fault, they are liable. We have seen all this before.

 

Very thought out comments Corwyn. This technology does exist, however have you considered the costs. Just the boxes alone used on commercial aircraft, with less than this capability, cost anywhere from $80,000 to over $300,000 for each different type unit. Now you have to integrate multiple units with everything in a car, have redundant systems in case of failure, then repair and maintain each car...and we have not even begun to talk about the roads sensors, ect.. and other navigational aids required to support such a massive system, all maintained and/or controlled by other people. You just brought up the liability issue too. I do not see it being as cut and dry as you predict with all this automation. Even in aviation there is always the blame game, where everyone points fingers to someone or something else and it usually takes years to determine, even with an Army of personnel and with recovered voice and flight recorders. That $5,000 repair to our Prius will look like pocket change then and lest we forget future road taxes. I love and embrace technology, but predict very limited use of "Johnny Cars" owned by the masses within any of our lifetimes at best...but what do I know?

 

No, but Google has ($75,000 for the current development version). But if it costs $300k for a $300m airplane, that represents 1/1000 of the total cost. I suspect that Google could build a computer driven *only* car for *less* than a comparable human driven only version. Interfacing with humans is tricky and expensive, computer control of machines is cheap. A top of the line Prius already has radar, GPS, and a camera. 360° radar and cameras all around are only going to be an incremental increase in that cost (x4?). Budisteanu has one for $4,000!

Killing hundreds of airline passengers is going to cause a lot of liability dodging, less so for a car accident. But, the airline *passengers* are almost never involved in that game of liability tag. So too will it be for the passengers of computer driven cars. The big companies will fight it out betwixt themselves, you won't care. And they will only do so to until things settle out. Million dollar law suits are only worth it when millions of dollars are on the line.