If driverless cars cannot detect and avoid moving and non-moving objects in the road ahead of them, or even moving objects ahead of them that are not in the road yet but might be soon and be prepared to avoid them, the whole technology is a no-go. I suspect this has already been attended to in the programming of current driverless cars.
That’s why I said long car trips. I go frequently from Silicon Valley to Disneyland for a conference and meetings, and I almost always drive. Going to Santa Barbara is about as fast driving. But those aren’t long trips. One Thanksgiving we drove from the Bay Area to St. Louis to give a car to one of our kids - it was a fun trip, we were lucky that the weather was great, but it took a long time.
I seem to have gotten on TSA pre-check so it isn’t even humiliating.
Calculus is a complex skill that fewer people can do than driving. So is language translation. But computers can do both of these things already.
Driving is not that hard a skill to start with. How long does it take for a kid to learn? Drivers ed classes are not that long, and half of them consist of scaring kids to try to keep them from doing things. Adults are usually better drivers than kids because we get programmed with the situations to expect and what to do to avoid them. Just like a computer. Except that one car’s learning can be broadcast to all cars.
Please, just because you don’t see how it can be coded does not mean it can’t be done.
Right. I don’t deny that the technology is possible. It just would seem prohibitively expensive except in small and contained areas. Even if we are talking about a 100 mile trip, we would have to make sure that every road was equipped for any and every anomaly. Further, on these exits off the expressway, we would have to ensure that every gas station, restaurant, liquor store, and strip joint were likewise “upgraded” or else these cars would be driving in circles and crashing into each other.
What happens if a guy has a heart attack while jaywalking on a one way street and ambulances are blocking everything? How long does my “smart car” sit there? Forever? 1 hour? What if there is a way around through a private parking lot that the cops are using to divert traffic? Is the car instantly programmed? Even if I am in Nowheresfuckville, WV?
Will the upgrades for public businesses and their parking lots be mandated? Paid for by the government or public funds? Does my home office qualify for public funds?
jtgain, the car is NOT just driving blindly according to some map it has. It’s looking out the windows, understanding what it sees, and reacting accordingly. Yes it has a map for strategic route planning, but so do you.
So far they’re not great at totally novel situations, like an accident scene. But the basics of driving in city traffic, reacting to lane closures, illegally parked cars, jaywalkers, etc. is all well understood and is being done live for real every day.
The comment about learning is the idea that as the software gets better, improved versions can be sent to existing cars. Thereby making the entire fleet smarter as more service experience is gained and more situations get added to the repertoire.
There are 2 ways to approach driver-less cars, smart roads or smart cars. Your arguments mainly focus on the smart road. With that approach the technology to guide the cars lies in the road itself. I’ve seen a prototype that had magnets embedded every few feet in the road so the cars could follow their lanes. As you have probably already guessed because of the questions you’re asking, smart road technology is not going to be the solution - it’s far too expensive and prone to error.
Smart Car technology puts all of the driving into the car itself. The car is able to identify the road and obstacles and responds to them in real time. There are already cars like this being tested including the Google Car. However, the Google Car still makes use of mapping technology that limits the cars range to a very narrow area of operation. The key piece of technology that’s missing is a type of AI with strong visual recognition of the world we live in. Good systems already exist but not to the level that a smart car will need to have. That kind of technology would be so valuable to so many applications that it’s practically guaranteed to be developed.
That’s not to say there’s no use for smart roads to some extent. We’d likely have a central system that would recognize congestion and reroute traffic. Cameras on highways like we have now could recognize obstacles and warn cars well ahead of time so they could move at a safe speed.
Another point you’ve made above about autonomous cars is that you seem to picture a completely closed box. The car would still need to respond to inputs from passengers, even if it’s just by voice. Passengers could direct a car where to park, to pull over for gas, or whatever other things spark their fancy. There’s no need to upgrade gas stations to handle autonomous cars. If the car pulls up to the pump there’s no reason at all that a passenger can’t just hop out and fill the tank. From a passenger point of view the experience will be almost exactly the same as being in a car with another driver. They can talk, interact, and tell the driver where to go - it’s just that in a smart car the driver is a computer.
Let’s talk when I can get home while jerking off to porn.
Well you don’t need smart cars to do … er … um … I have no opinion on this matter!
Though Google has shown cars without steering wheels I doubt very much they are going to be the first ones for sale. No one is going to reject a smart car which does 99% of the driving because they’d have to pull into a parking space with just the auto-parking assist.
Not that this is a difficult problem. Each spot could have a very cheap sensor that tells if it is occupied. (ISTR that some parking meters do this now.) When they sense a car pull into the lot they start transmitting a signal saying that this slot is available, and the car pulls into it. Hell, when I turn on my Roomba to vacuum my living room it automatically returns to its docking station for recharging now.
What happens if the one lane one way street doesn’t have any parking lots next to it, just driveways? Today I bet people just sit. The car can be programmed, if this ever happens anywhere in the country more than once a year, to talk to cars behind, tell them the road is blocked, and have them all make u-turns in driveways one by one.
That is what a lot of software maintenance is - modifying code to take into account stuff you never thought would happen. We do it all the time. My code if full of modifications to fix things that never were going to happen but which did.
Smart cars can already read a sign to know what the speed limit is, where to stop, whether a person is crossing the street, if a deer is 30 feet away from the road and headed your way. It can identify a traffic lamp and tell what color it is. It can see all that and more.
Finding an alternate route due to unexpected heavy traffic is also possible today using existing apps.
How large will these “trains” of pods get and how do we facilitate pulling out cars mid-trip? This is why travelling by bus is annoying having to stop all the time to let people on and off.
Make that more like 5 years. What rebuilding of infrastructure, the existing prototypes are designed to drive in present day infrastructure. Do you realize that the Google Prius/lexus prototypes have run up to Lake tahoe and such. They are not just working within a tiny fraction of the bay area.
The average driver sucks, plenty of people that think they are great drivers are shitty drivers. The features in the next couple model years like novelty bobble mentioned are already here and in production in fairly common cars like Toyota Camry, Ford Fusion, and Honda CRV.
I am hunting for the videos that show how googles cars see the environment around them and they are aware of pedestrians, dogs, bicycles, etc. They can also see, read, and obey most street signs.
Found it
Driving around London can be a PITA precisely because that already kind’a happens. Around, not in; I’m talking about the beltways and not the streets.
You’re chugging along, in busy but fluid traffic, following Mz Nice’s directions. Then Mz Nice says “oh no! Heavy traffic ahead! Quick, do you want me to reroute you to a way that will take 3 minutes less?” and you don’t just have to click “no”, you have to click “No”, “I said no” and “Hell fuck NO”.
And she keeps doing that for something like an hour… damnit, the plane isn’t going to leave without me for three damned minutes, just go on!
Note that all which would be needed is a setting for “minimum delta-time which will be considered relevant before rerouting is offered”. The logic is not particularly difficult to define, what’s needed is someone to think about it and a good interface (and this last one can be the hardest part).
Cars will be able to “pull out” at will, just the same way any other car can…
The thing is, these days technology-based products go pretty rapidly from being a niche product for rich people and tech junkies, to being something that everyone uses.
It’s hard to see why that won’t be true with cars. Once some Mitt Romney equivalent has a self-driving car that can handle city and highway traffic, it’ll be less than 10 years until Joe and Jane Sixpack are buying a self-driving car.
The reason, of course, is that however expensive it is to develop a piece of software, every copy of it can be created almost for free. And the additional hardware for a self-driving car will be fairly cheap: a few hundred bucks worth of cameras and other sensors feeding data into the computer.
Either infrastructure rebuilding won’t be necessary, or self-driving cars won’t happen.
There will be no more appetite to do massive speculative infrastructure investments to make self-driving cars a reality down the road, then there has been to make non-speculative (everyone else has it already!) high-speed rail a reality in the U.S.
One thing I don’t understand is why some people think these ‘trains’ of car pods are an important part of the idea.
First of all, communication between self-driving cars makes self-driving cars inherently hackable, which sounds to me like a bad idea. If you hack my laptop and it becomes unusable, I’ve lost a laptop. If you hack my car and it becomes unusable, if I’m riding in it at the time, it might come to a stop by running into another car or some solid object at high speed. For that reason, I wouldn’t want my self-driving car communicating with other cars.
Second, while there’s surely some energy savings involved, how much exactly? ISTM that the big energy savings from self-driving cars is that instead of needing a car that seats 4 or more people just because once in awhile you need that much capacity, you’ll only own as much car as you need all the time, and the rest of the time you’ll rent whatever additional car capacity you need. So most self-driving cars will be just big and powerful enough for one person, which means they’ll be able to weigh far less than today’s cars.
Third, in terms of highway capacity, one-person cars will mean that existing roads will have room for a lot more cars, since traffic lanes will only need to be half as wide. That should save enough room so that ‘trains’ won’t be needed to increase highway capacity.
So ‘trains,’ assuming they can work without any risk of cars’ computers being hacked, might add some icing to the cake of self-driving cars, but that’s about it. And IMHO, that’s a big assumption.
I do not anticipate an increase in single passenger cars because they’re autonomous. They’ll still be used for the same things as today - hauling all your stuff, your kids, whatever.
And I really don’t think a massive overhaul of infrastructure is necessary. Cars are already very aware of their surroundings with RADAR, LIDAR, stereo vision, GPS, etc.
I have a better reason. Once corporations seem proof that self driving commercial vehicles are safer and more cost effective than paying a human driver, you’re going to see thousands of them integrated seamlessly with already mostly automated containerized shipping ports and rail yards.
Im curious how long it will take before things like drivers licences would become irrelevant. Once the cars are settled into the culture… have to be at work at 7am and kid to school at 8… no problem toss kiddo in and send the car on its way. Car comes back to you at work.
Kids, drunks, blind, neurologic impairments, amputees all more able to drive.
I am excited because I am one of the people who drives because I must. If my car could do my driving for me while I answer phone calls or emails! I am a happy boy. Point my car at las vegas and read or nap a bit…sign me up. I have even heard grumbling of concern in car insurance and auto body circles that this may slowly cripple if not destroy that market.
car accidents are one of the more common causes of death in sub 30 age ranges. how many times do we see families losing a kid or a spouse to a car accident. Most of us know someone who has been seriously injured or killed in a car accident.
To the naysayers
This is exiting #livinginthefuture kind of stuff. We have heard these arguments before…seatbelts were going to trap you in burning cars, helmets were going to limit hearing and peripheral vision, and lots of other “bad things”. It was fear of change then as much as it is now…and you will look just as ignorant and closed minded 5-10 years from now when the streets are filled with them.