They do seem to be having success in testing.
Eventually it’ll probably come, regardless of my feelings on it. But I doubt of it’ll be for a long, long time yet. I don’t expect to live to see it.
And I thought “Super Cruise” was so aircraft could stay above mach1 without using afterburner.
Google’s system uses a scanning laser on the roof to detect obstacles much like auto-guided vehicles used in manufacturing. Those things are both expensive and fragile. They have a swinging mirror inside to scan the laser side to side. If that is the technology Google wants to license, I can understand manufacturer’s reluctance. The most robust systems seem to use radar, sonar (from parking sensors) and vision systems. The vision systems use cameras to collect data and computers to identify obstacles, people, painted lines on the road, the dirt edge and more in 3D (with at least 2 cameras). Considering they are doing this dynamically while driving, it takes quite a computer.
At least one is available as an aftermarket warning device, Mobileye 560 for $849
The Eyesight system in my Legacy is pretty amazing, but I wish it was even better. I think it shouldn’t be too hard to add lane following to the lane departure warning and electric steering.
The autonomous Google car may never actually happen
I still think the autonomous Google car is primarily for their own use, providing the street-level pictures you see upon request while using Google Maps. That will be a huge money saver when it is ready for prime time.
If they can land a jet fighter on a carrier deck 100% automatically in any weather, they should be able to keep a car in a defined traffic lane without crashing into anything…You can be sure the first systems will require a driver to be seated behind the wheel, ready to take over in case of a problem or traffic situation the system can’t deal with…
That yellow line that defines the left edge of most highways could become a useful part of the guidance system… On the wide-open Western Interstates, standardizing and upgrading this line would be a low cost and perhaps useful tool to guide Super-Cruisers down the road safely…
That yellow line that defines the left edge of most highways could become a useful part of the guidance system.
Snow?
Interesting point. I wonder if any of these systems can operate in snow. I know the S-class system uses satellite mapping technology, and even prepares the suspension for varying terrain, but I don’t know how it works.
Next time you are driving, note the variations in the painted lines on the highway. Different colors, totally missing for sections, faded, covered with snow, changing from dashed to solid to double and back again…
It would take artificial intelligence for a computer to do this 100% of the time safely.
b
I was watching a TV show this afternoon. A snowplow driver was cleaning the road in a big Alaskan snowstorm, and there was a whiteout. He was able to continue driving by switching over to a GPS system that showed him the edges of the road. Just in time, too, as he approached a guard rail and would have hit the end of it head on if he had continued his original path.
Farmers now use GPS systems to program the path of their combines with amazing accuracy. It’s amazing how far technology has come.
Commercially available systems are accurate to around one meter horizontal distance. Augmented systems can be more than ten times better.
"Very true, but who gets sued when there’s an accident?"
Asked and answered.
If a pilot (or, Heck, Cap Hazelwood) puts it on autopilot and subsequently wrecks, it is his fault for not properly monitoring his systems. There seems to be no new wrinkle involved in an automotive autopilot; it’s the driver’s fault.
Automating the mechanics of driving in no way releives the “captain” of responsibility of his command. Why would it?
I didn’t think civilian GPS was allowed to have that kind of accuracy…Most if not all Marine units are only good down to about 30 feet which is usually close enough…But that’s not close enough when you are traveling down an interstate at 80 MPH…
I was out on the road today and thinking about this…I don’t think any robot can steer or control a car at high speed with the precision of a human driver…The human can respond almost instantly to changing road conditions and make supple but critical corrections. The robot can not…
GPS doesn’t have guaranteed availability to civilians: it exists for defense and they get priority. It is understood that GPS can and will be taken offline in a National emergency.
Therefore, any autpilot system would require a “plan b.”
@Caddyman, I think you’ve got it exactly backwards - robotics can (in fully mature form) respond far faster than humans possibly could - see ABS - it pumps faster than humans ever could, and new traction control can brake or not individual wheels - something a single human just could not ever control, especially in the preventing / correcting for a skid timeframe.
“Robots” already adjust in fractions of a second to road conditions for some of the suspension systems as well. “Robots” adjust timing automatically as you drive modern cars, something humans can’t do. It took a long time, but now the top of the line automatics are better at shifting than humans for better gas mileage.
I see no reason to think that eventually, “robots” or automatic systems will be better at driving than humans. And it’s incorrect to compare the best a human driver could ever be with the worst or average a “robotic” driver could be.
Robotics don’t get tired, angry or distracted, something commenters on this very forum complain about all the time. The number of drivers who don’t slow down for rain or snow and ice seem to point to the truth that while a human may be capable of responding quickly (not instantly - there’s at least a second or so lag in most cases) to changing road conditions, in real life - humans often don’t.
Automated systems can also learn from each other in a way that humans don’t. If one automated car finds a bug in the programming, the vendor can fix that bug and push it out to all cars using that system via a simple programming update. It’s a lot harder for humans to re-learn a “bad behavior”.
I think it’s as close to inevitable as any trend I’ve ever seen - humans won’t be driving most vehicles in the future. Now the question is, will it happen in 15 years or 50.
Robots can react far faster, but humans are able to interpret a far, far greater and constantly varying set of inputs.
For example, scientists are just now getting a robot to the point where it can react to being pushed in the shoulder while standing. It turns out that the act of doing so is far more highly complex than they ever realized. It took decades to even create a robot that could walk. It’ll be decades again before one can step up onto an escalator. Driving to the level that robotics could handle varying environments, especially when they’re constantly changing, is a very highly complex task. Even many humans can’t do it well.
I think we’re underestimating the complexity of driving in the real world.
When you are driving down an open road at 80 mph, there are all kinds of things going on that the driver is (should be) aware of that result in very minor steering inputs…Passengers in the car don’t even feel it, or if they do, they consider it normal and ignore it. I suspect a computer will not be able to duplicate this and the ride “feel” will be insecure, twitchy and jerky, as steering and throttle corrections are made…