I imagine that all of these vehicles will, as a rule, incorporate continuous data uploading to the parent company. Part will be for the mechanism of self-driving (i.e., they’ll need to be connected to other cars, to GPS, to online datasets about upcoming traffic and weather patterns), part will be for the sort of product-improvement data that many software manufacturers already collect, part will be for evidence in case of collisions. (And since cars usu. operate in a dense cellular communications environment with easy data transmittal options, they may not need the same kind of black box as an airplane, which may operate over the open ocean, well away from communications networks, for example.)
You’re not suggesting cars navigate via GPS signals are you?
Survey.
Let’s start with a couple of those initial applications I mentioned.
How much would you pay extra to have the front collision warning with automatic braking? (It’s part of a $2100 upgrade on the VolvoS60.) Would your willingness to pay more change if you had teen aged drivers?
How about for “cruise control” that can also maintain a safe distance and stay in the lane both at highway speeds and in traffic jams?
Both as part of the same package?
And when the GPS signal isn’t received? And the optics are blinded by reflected sunlight? And the traffic light goes out? And a car crosses the center line requiring a decision only a person can make? and a million other what if’s that can’t be codified into the programming?
And the human driver spills coffee in his lap? And the sun gets in his eyes? And he sneezes? And…?
Like DSeid mentioned, I’m certain the cars will use GPS as part of their system. Just as smartphone mapping devices use wifi and cell tower location data as part of their systems (in addition to GPS, and accelerometers, compasses, etc.).
Right that anything that can go wrong will. It is possible and will happen that all of the lasers will simultaneously fail at the same time as the GPS fails and the radars fail just as all the other redundant systems fail just exactly as a car crosses the center line (before the car slows to a stop as datasets completely malfunction).
Of course it is much more probable for a human to have a stroke or a heart attack or a seizure or sudden bilateral retinal detachment or just be distracted …
But honestly I’d give the autonomous system a much better chance at avoiding the collision with a car crossing the center line than a human driver, as well as a better chance at not being the cause of that car crossing the center line.
And a human brain can think where a computer can’t. All you can do with a computer is add more lines of code to correct for flaws after-the-fact.
There are components of the system that can perform better than humans such as auto braking in the event of a tandem crash.
Sure but that isn’t how most human controlled MVA’s happen. Most of them are just due to inattention, risk taking, errors of judgment and, frankly, being drunk.
The type of unusual situation you are talking about where human ingenuity is required is not at all common, and I think you overestimate human ability to react well to the unexpected anyway. Many faced with such situations will panic and do precisely the wrong thing.
Further, autonomous driving software will probably be notably cautious: for example, faced with lack of visibility it will probably be more likely to just do something sensible like stop, where a person might rashly carry on.
This is so quaint. Shades of John Henry vs the Steam Drill, or perhaps Kasporov vs Deep Blue or Watson on Jeopardy.
Driving is a task that is mostly mundane but that requires maintained high vigilance anyway because at any moment it can require extremely quick reaction time decision making. Those decisions virtually never are chess like in their complexity or all that novel in nature.
Creative and divergent thinking (the area where human brains have a huge advantage over any computer, the advantages of daydreaming) while driving is not a good thing.
I find it really weird that anyone would put human perceptions and reaction times up against a machine.
Driving isn’t painting a masterpiece or writing a hit song. It’s responding to perceptions and following rules. Those are things that machines are much, much better at than us.
If you have a few minutes watch that. The video shows high speed robotic hands working faster than a human could even dream of.
I would agree with that.
I would strongly disagree with that. There are an infinite number of variables that occur while driving and computers can’t think. I think you grossly underestimate how complex a variable task is to a computer.
Unless it’s programmed for the human factor that would be a dangerous assumption. I’ve been in a number of situations where visibility deteriorated instantly. This has happened to me on busy highways where everybody is in trail in all lanes. Everybody collectively took their foot off the accelerator and gently braked. It would work great if EVERY car has the programming not it would work in reverse if only a couple of cars had it.
All this comes down to liability. Lawsuits will allows follow money and regardless of
I mentioned before that we have the technology to track vechicles every accurately for purposes of congestion pricing, and you took issue with that then, saying that GPS isn’t accurate enough.
What you’re missing is that modern navigation doesn’t just include GPS. It includes everything from GPS to speed sensors to acceleration sensors to WIFI hotspot tracking to turn sensors and maps. It’s a redundant system that does an excellent job of determining location and which can withstand multiple failures while still maintaining situational awareness.
It’s also pretty good at determining when it can no longer determine location. An automated system could easily be devised to notify you to take over manually if the navigation components begin to lose track, well before all positional awareness is gone.
And I think you grossly underestimate how often human drivers are completely oblivious to the road. After starting to ride a motorcycle this year, I’ve become acutely aware of what people are doing in their vehicles while driving. I check on the driver every time I have to pass a vehicle, and between cell phones, texting, arguing or engaging in animated discussion with a passenger, eating food, playing with the radio, or otherwise engaging in non-driving activities while driving, I swear that half the people on the road are relying on dumb luck to keep them alive.
The bar for an automated car is pretty low. I could see them starting out on major roads that are certified for such use. The maps would be guaranteed to be accurate, there could be laser reflectors on the roadside to make it easier for the cars to ‘see’, differential GPS transmitters, whatever. Start with some of the major commuter highways that people drive every day.
I was skeptical of automatic cars until very recently, but Google has racked up a hell of a lot of miles of unassisted driving, and that’s on standard roads with no attempt to make the infrastructure work better for such vehicles. It’s been very impressive.
I do agree that liability is a major issue, though. People don’t trust automation, and complexity makes it easy for defense lawyers to construct plausible chains of causation where none exist. Look at what happened to several European car makers over ‘unintended acceleration’. They’ve paid out huge settlements despite the fact that no cause has ever been found, and likely doesn’t exist.
The first time an automated car fails and kills somebody, the lawsuits will be flying.
Situational awareness in a car is plus or minus a few feet. Aircraft are spaced out to avoid the inherent problems of wandering. GPS in aviation is the greatest thing since sliced bread but blows chunks when it quits. It cannot be used autonomously without human interaction.
This discussion revolved around the idea that cars would operate themselves. Taking control when the computer stops working is a lawsuit waiting to happen.
imagine them relying on a computer to drive for them. situational awareness will go out the window.
The first time a bunch of cars go by in close trail they will immediately be followed by a line of teenagers trying go duplicate it. After the novelty wears off it will be the general public reading magazines and watching TV while their car drives into a construction zone wall because an update failed to load.
well lets take a look at that statement. From Wiki:
" As of 2010, Google has tested several vehicles equipped with the system, driving 1,609 kilometres (1,000 mi) without any human intervention, in addition to 225,308 kilometres (140,000 mi) with occasional human intervention. Google expects that the increased accuracy of its automated driving system could help reduce the number of traffic-related injuries and deaths, while using energy and space on roadways more efficiently."
How many miles are driven everyday in the United States alone and how many “occasional human interventions” does that translate into? this is currently being done as an experiment by people tasked with babysitting the device.
I agree the system has great potential to reduce traffic related crashes but only if the system is secondary to driving such as we see know with auto braking systems and anti-yaw, anti-lock brake systems. The idea of cars operating in train mode is dangerous because it gets people use to a dangerous driving situation when the computer can’t be used.
It will be the battle of the century. Useful but flawed technology versus the lowest human denominator.
A small update. Mercedes upcoming systems incrementally advance the car taking over.