Have you read anything about the Google car? I think you will find it won’t need these restrictions:
http://www.autonews.com/article/20150103/OEM06/301059928/where-is-googles-car-going?
I don’t know about your car, but when I drive my Prius I’m driving tons of software already. Two firmware updates so far.
As far as sensors go, we’ll have them running on cars with drivers long before driverless cars are ready. The same collision avoidance software a self-driving car needs is going to be real useful for all cars, and is happening already. So that technology will be proved in.
Engineers know all about fault tolerance and ways of failing over to a safe configuration already, and stopping in the middle of the lane is not something high on the list of actions to take upon a software failure. If it were, we’d be seeing cars do this now due to a bug in the software that runs your transmission say. Give them some credit, please.
Do you have any clue what the cloud actually is?
He will if he mugs you when you take your key out near your car. Especially if it says “Ford” on it. My key never has to leave my pocket when I open my door. No fumbling with the key in the lock. I put my hand on the handle and poof, it is open. I can get in and lock the door a lot faster than you can.
Ever try to find your car this way? Not easy. Ain’t no way a mugger who steals your watch on the street or in the mall is going to do it - he’ll look real suspicious on the security cam.
I’ve got no ignition key, just a button. Which only works if the fob is in the car, of course.
I do reliability. Which do you think is more reliable - a wire sitting in a wiring harness somewhere or mechanical components in the engine subject to vibration, oil spray, heat, and temperature extremes? There is already tons of wiring in cars, and cars are getting more reliable, not less. I wonder why?
You can do a lot of component and software upgrades using standard car bus architectures already in place, just like you don’t have to touch the inside of your PC to use a new usb device. I’m sure sensors will be reasonably modular, or as modular as they build stuff today, some of which is a real pain.
Electronics helps. Electronics these days is incredibly reliable - I have the raw data. I can say this - your PC has a processor with a billion transistors, and a fan with a few moving parts. Guess which is going to fail first? It’s the fan.
I trust a computer to react much more quickly than I can or you can.
Consider airbags. They respond at the time of impact, but inflate quickly enough to keep you from hitting the steering wheel. The total time is about 27 milliseconds.
Yes, computers have faster reaction time. But there are other situations I wonder about. For example a person is walking along the side of the road. I am going to respond quite differently if this is a toddler vs if this is an adult. How will your automated car respond?
Remember the days when cars were all driven by meat robots? Meat robots controlled by squishy meat computers?
Sometimes these meat computers would just go into standby mode on their own if they hadn’t been shut down in a while. They were so slow that the car would travel several times its own length before the meat reacted to new conditions. They only had two optical sensors, which were only forward-pointing and took many milliseconds to respond to input. They didn’t do well under dark conditions, couldn’t see infrared, and didn’t focus very fast because they weren’t much more than warm, goo-filled balls. There were no radio-frequency sensors whatsoever.
The meat computers would sometimes glitch and just not see some inputs. In fact, they hardly responded at all to unexpected inputs, which you would think would be the most important ones. Anything out of sensor range was wholly unregistered, and as mentioned the sensor range was very narrow.
The meat robots did especially badly if there was more than a fraction of a percent of ethanol contamination, and no one ever figured out how to prevent this. Some of the robots deliberately sought out ethanol mixtures as some kind of fuel. Meat was already really slow and buggy, but this made it worse.
I don’t miss the meat robot days at all. Some of the meat still makes disturbing sounds by squirting air through its meat flaps, but if you lock the doors and disable the oxygen ventilation it stops doing that eventually.
A person on the sidewalk is tracked just like another moving vehicle. It has a location, a direction, and a velocity. Yes small children move quickly, but compared to a car or a bicycle they move much slower.
skip forward to about 6:30 they show you how self driving cars see the world around them and explain how the system accounts for a variety of challenges.This is not just some random computer guy writing this, there are a bunch of Really Smart People ™ working on this.
Actually the systems are watching for obstacles much smaller than cats. A piece of debris in the roadway that could cripple a car can easily be smaller than a cat. A small child will not be “under the radar” Watch the video I linked, they show it dealing with high volume pedestrians including someone dashing out at the last second.
Will there be people who can out drive the computers, probably a few, perhaps 20% of all drivers. That number will creep down as they get more experience writing these systems.
Computers don’t get bored, they do not become accustomed to a certain route, they dont check out the hot chick on the side of the road, they dont get tired, distracted by a kid in the back seat, or check their cell phones. They dont speed, follow too close, or accelerate at yellow lights hoping to make it through. The prototypes are already statistically better than average drivers.
Driving down a 2-lane scenic road with the windows down and the stereo cranked is one of the purest joys of my life. No robot is going to take that away from me!
I’m confident they’ll be made to work well. And some enthusiasts will resist them for a long time while many commuters will embrace them quickly. I like operating machinery of all kinds. I have a so-called “ultimate driving machine.” I take the train to work when I can because I’d rather read the SDMB than drive in Miami traffic. I’d be up for an automatic car so I could read but also travel on my schedule, not the train’s.
Overall, I’m most curious about the sociological impact they’ll have on the poor folks. Recent changes in emissions and airbags and such have reduced the supply of cheapo clunkers that poor folks can afford to buy and then drive until it won’t move anymore. In effect the floor price of a minimally drivable used car is now near $1000 versus $100 when I was in high school. Some of that is general price inflation, but much of it is specifically that clunkers are now less clunky & more expensive.
Fast forward to a time when 75% of 10 year old cars are self-driving. And they refuse to go when their sensors are failed. We will see more examples of cars becoming brittle purchases that depreciate slowly & predictably until the first major failure. And then they’re scrap. Kind of like you can total a 2-year old car now simply by firing all the airbags.
I’m not suggesting this is an obstacle to society embracing automatic cars. That will happen with no regard for its impact on the 3rd & 4th hand market for clunkers.
This. Auto automation? I really want to go there. It can’t get here soon enough for me. I haven’t seen an argument against it here yet that wasn’t grounded in ignorance. Our acceptance of traffic deaths is horrifying. If someone proposed a new system of travel today that was sure to result in 30,000 to 40,000 American deaths per year, what do you suppose the reaction would be?
Realistically I don’t expect self driving cars any time soon. There are too many people with that “but what if a storm’s a-comin’?” attitude. But it will be great when it happens.
The marketing challenge will be convincing the 95% of Americans who think they’re above-average drivers.
Yes, but I believe if a driverless car gets to a point where its software and sensors can’t come up with a safe route and it can’t figure out where the side of the road is or how to pull off, it will have no choice but to stop and prompt the driver to take over. That would be the failsafe that is least likely to cause accidents.
Long haul trucks and delivery vehicles will convert in large numbers first. Then taxis and limo’s. Along with high end cars. Then average cars. Until some areas ban all but self driving cars. After a while the only place a human will be able to drive will be a track. As more cars become self driving they’ll start talking to each other making them safer and safer.
It doesn’t matter. Drivers who want them will buy them for the convenience, not because they think the robot is better than they are (which it is.)
We went over most of this recently in GD: http://boards.straightdope.com/sdmb/showthread.php?t=742275
I don’t have the energy to go through it again so soon, but the short version of my position is:
Heavily assisted/autopilot cars for the general public aren’t too far away tech-wise, but the car that I can take a nap in without hazard is so far away that I wouldn’t really want to put a date on it. That makes me sad, because I’d love to jump in the car, say “Follow that car!”, and begin playing “Foggy Mountain Breakdown” on the banjo. I know I could do this today in a taxi, but I have some compassion for my fellow man, and don’t want to get kicked out of a taxi.
I can’t wait for it . . . but there’s no way I’m riding in a computer-controlled car unless all the other cars on the road are computer controlled too.
TIs have always been a niche product which is why their price stubbornly stays around the $100 mark. Driverless systems will be for the masses and regulated by government much the same way that OBD systems are all standardized with common faults that are easily looked up. These systems will have to have commonality such that they are plug-n-play, so overwhelmingly they will be treated by the markets as commodities, thus the price curve will be similar to PCs in the last 50 years.
I meant the monetary aspect of it. Just like today’s cellphones are not “user repairable”’ they are quite easily user-replaceable. In fact, it is so utterly ridiculous for the common joeblow to attempt to repair any sort of electronic device that if you spend just an hour researching how to do it, you’ve already lost money by not just simply replacing the system outright.