If you have invented something which can do that, and has reasonable power requirements, and can be built for anything less than a trillion bucks…
I want to know. Seriously. Because I will invest everything I own in a startup and make a bajillion dollars.
Of course, I’m betting your just talking out your arse.
Oh, ever so sorry for this fatal confusion – it should of course read “high **subsonic ** speeds”!
Small wonder no expert wanted to follow me with this last specification, as this would have defined an aircraft even the Pentagone couldn’t afford.
Yet high subsonic speeds means in excess of the 400 mph announced for Sikorsky’s X-Wing. To fly that fast, both coaxial contrarotative rotors of this concept (featuring a tail-mounted pusher propeller) would have to be brought to a complete stillstand at top speed.
Sorry once again: this was the real joke post, if any!
You’re contradicting yourself here. The hackers don’t need a centrally administered system. As you note, they just need to change the information the car’s system gets. Having the cars’ control decentralized would limit the number of cars you could hijack at once, though, and make the hack require some footwork. That’s not really a large barrier. You are aware of the traffic signs that warned of zombies this week in Austin?
It does seem that for the traffic management features mentioned in the thread, you would need at least centralized area management. It also seems to me that the system would have to know each vehicle’s destination (at least short term) for intelligent routing of traffic load. That’s the spot I would concentrate on if I were a complete jerk, and this system was real. If you can control the information the cars are using to navigate, you can probably control everything else it does, too.
The cutting edge of self navigating vehicles drive like drunken teenagers on some level, and that anyone who actually knows how to drive will turn off the currently available driving aides when they want to navigate a course in the least possible amount of time. BMW has a car that can navigate a track once you have shown it the course. The DARPA projects have just gotten to the point where enough of them can navigate the course to consider it a race to the finish. They still crash. A lot.
So, when those problems are all addressed, and they can drive like this.
Gimme a call. I imagine I will be an old man. Until then, I am not gonna put my butt in one. I have watched the internet grow and have its fits and starts. It was not made by dummies, they just did not account for every eventuality. The proposed traffic management system has a lot of the same problems to address, and adds many, many more by placing those problems in the real world. Remember, it also has a much higher price of failure.
This does not even touch some of the other technical problems that would need to be addressed. Who is gonna make these computers? Is this a published standard or open sourced program that every automaker can use, or do you have a contract with one company who makes them all? I am an open source advocate, but I don’t see the first being workable. The temptation to fork the program or the standard and add features for either prestige or profit would be too high. Additionally, do you really want to expose yourself to the possibility of being on the road next to a fool with an unauthorized AI driving? (“check it out, I got my car to treat traffic like a slalom!”) The second does not seem to be very likely, either. Getting race teams to use the same ECU is difficult enough. I don’t see the manufacturers being really happy about buying from one AI supplier. Perhaps a body to approve/regulate them would be the solution.
I am aware of it, and I not contradicting a damn thing. Use your head; a hacker is someone who uses one computer to acccess, alter, or delete data on another computer. You don’t need to be a hacker to do anything like what I proposed: all you need is a transmitter tuned to the right frequency or a powerful enough one to mess with people’s signals. Any schmoe can do that for about $50 worth of electronics parts from the nearest Radio Shack.
But you don’t need a centrally controlled computer, either, and it’s not really good for optimizing traffic load. The problem with such a system is that it creates a single point of failure and control. Secondly, it can’t respond as easily to changing road conditions. Sure, if you knew exactly when every car was going to come on, and had zero transmission delay, the thing could figure it out. But you don’t.
So what you do, is give every car a very reliable system, with preferably five or fewer basic rules, and let that drive itself, with cues from the road and other cars. Birds do it. It’s what people do. And our systems can do it yet. Maybe one day they will.
The thing is, a system like that won’t be centrally planned and managed - it will grow organically out of the market. In fact, it’s already happening. I’ve got a GPS unit which can get real-time traffic in some areas, and plan routes around it. No one else needs to know about my GPS unit, you just need some form of traffic monitoring, a way to get it out to the public, and then let the market sort it out.
What would be nice would be a ‘smart road system’. A road system that monitors traffic movement and calculates average speeds and traffic density. Then you could access the data throuh a mobile internet connection and the gadgets in your car could do whatever they wanted with it.
What would be even better if we have a way to track every car’s position in a way that doesn’t compromise privacy. If we could do that, you could then fine-tune your roads for congestion through congestion pricing. Imagine driving home, and you program ‘cheapest route’ into your GPS. It requests current pricing data from every road between you are your destination, then does a least-cost calculation to find you the cheapest way home. As roads get more congested, the price of the road goes up.
Allowing price signals to work in the road system would instantly make it dramatically more efficient. Traffic would spread out, rush hours would spread out as people time-shift to take advantage of cheaper road times, home and business construction would be stimulated in areas where the roads are under-used (or travel distances are shorter), and discouraged in areas where there is already too much congestion. These changes would happen gradually and on the margin. As a city grows and changes, the growth would happen more efficiently.
Another happy side effect is that it would get rid of an externality - the subsidized use of roads. Ideally, this externality is captured in gas taxes, but it almost never works out that way in the real world - for example, an old lady driving 10 mph under the limit in a Honda Prius is causing way more congestion than a young guy in Hummer going at the speed limit, yet she’s paying a much smaller share of road tax.
Done right, such a system could be revenue-neutral - the money collected from congestion fees could be deducted from gasoline taxes or property and business taxes. It could be phased in slowly by starting with very small prices, to allow the economy time to adapt.
And as a side benefit, you’d have perfect traffic flow data that in-car computers could use to optimize route planning.
The only problem with that is precision. A system like that can’t be perfectly accurate, and even a 1 meter error is vastly too much.
We’re just using different definitions of hacker, then. I’d consider someone who’s building transmitters technologically advanced to be called a hacker. Either way, any system proposed so far looks pretty easy to at least DoS, if not hijack.
Birds run into things. People certainly run into things. Both also try to cram too many of themselves onto a particular route. You’re at least going to need a routing protocol for these devices to speak to each other in order to provide any benefit as far as traffic load.
That very reliable, better than human or bird system is going be the main sticking point. Again, call me when it’s better than the best drivers out there. Then it’s worth paying good money to have it drive my car. Currently, arriving at the end of a controlled course safely or otherwise is a success for AI. But if all you’re saying is that it’s theoretically possible, I am not gonna argue that. I am arguing against it being practical in the near term, or even desirable if it were. It’s not just faster computers that are necessary to make this work. We need to fundamentally advance our understanding of how our vision and decision making processes work, then translate that understanding into binary code before we can even approach the level of skill a human has at driving in the real world. Once we’ve done that, we still have to beat it somehow, because even people can’t make a car train work reliably.
Upon reviewing your other posts in this thread, why are we arguing at all at this point? We’re both saying its pretty dangerous. I just think it being less reliable than people would make it unworkable, and I don’t see it getting better than people in my lifetime.
You’re talking about my system? It can be as accurate as you want it to be. Down to the millimeter, if you want. For one thing, you could use differential GPS. Or, you could use mapping technology - my GPS unit may only be accurate to 10 meters, but combine that with map software and turn sensors, and you can get a pretty exact idea of where you are at any given time. How accurate you make the system is really just a matter of money, and not of engineering or big advancements in computers. It’s pretty straightforward.
Now, an airborne navigation system is completely different. You can’t predict where anything will be in the next minute with any kind of accuracy, because the aircraft may be dropped or pushed up hundreds of feet by air currents. All you can do is leave a big physical envelope around each aircraft to keep them safe.
Question from the sidelines: could you explain what you meant here by super-critical? Thanks.
Again you sound like an expert, but at closer look it appears that you’re not an expert at all – at least neither in aviation in general nor in avionics in particular.
I’m an self-taught expert in aerodynamics and aero-kinematics (aero-mechanical systems, e.g. the helicopter rotor), but not in avionics.
That’s why in my first post (#27), I’m asking for an expert to confirm my assumption that absolutely precise and rigid electronic tracks can be calculated in real-time for every aircraft – according to vertical and horizontal safety clearances relatively to other aircraft present within or about to enter a critical range around that aircraft – by the on-board avionics for projection onto a head-up display or a glass cockpit LCD.
I further assumed that the same information can be computed and combined with the TCAS (Traffic Alert & Collision Avoidance System) for the automatic pilot to guide the aircraft safely and precisely along this personal ad hoc aerial highway (a context in which the term highway appears to be indeed of almost premonitory character!), whichever vertical or horizontal air movements the aircraft may encounter on its flight path! (And this is of course just what any GPS-related automatic pilot can already perform pretty precisely!).
Here you seem to get confused by the wrong popular belief of the “crowded airspace”, which in fact is a virtual reality and moreover only so in the vicinity of the major hubs within a range where airliners are set on final approach slopes or initial take-off ramps, i.e. flying at minimum speeds close to stall and therefore not capable of any kind of collision avoidance manoeuvre!
Hence the big mathematical (not physical!) envelope around these huge cattle-carriers of the skies, just before they reach and after they quit their cattle-shed!
… No, it isn’t. GPS signals, even with multiple ones, are neither reliable nor precise enough for that kind of work. They will have no effect or utility for driving cars whatsoever, except to maybe compare to a map and think, “Oh, guesss I’ll have to turn somewhere up there.” Seriously, you have no idea what your talking about. What you’re demanding is vastly beyond any technology we currently have. No signal we can generate, short of high-energy radiation, is capable of the tasks you seem to think are simple.
Here, you go do it, if it’s so bloody easy. I’ll be waiting, but I won’t be waitin up.
Again, the goal can be modest and the technology to achieve the modest goal is simple.
The first goal is merely to allow ad hoc platooning - use communication to allow vehicles in highway situations to travel closely enough to both gain fuel economy from drafting if nothing else and to increase the traffic flow at higher traffic volumes. Such an outcome would benefit fuel efficiency and traffic times even for those not participating in the platoon.
For that all that is needed is[ol]
[li]Ceding control to the platoon until predetermined safe release conditions.[/li][li]Communication of upcoming traffic and hazard conditions so the entire platoon can adjust velocity accordingly.[/li][li]Intervehicle communication of more immediate changes in conditions and adjusting vehicle spacing and velocity accordingly.[/li][li]Proximity detectors within vehicles that verify the distance between vehicles and that have a hard wired minimum distance between vehicles at a various speeds and which automatically decrease velocity such that that intervehicle distance is never breeched.[/li][/ol]
One initial target could be interstate trucking.
Another would be commuters on highways switching current express and HOV lanes into ad hoc platoon only lanes.
The hacking risk in this system is minimal as minimal safe performance is hardwired into each individual vehicle. The technology is already extant and some aspects are already part of some current vehicles although for different applications.
No, this plan does not put cars on autopilot on arterials or side streets. But the limited ambition would still have substantial gains.
Almost two years later, an update.
Not much of an update but its nice to see methodical progress being made.
And yet, no one has contacted me to offer royalty payments…
Seriously, this is good news, This is exactly the kind of bottom-up solution that makes sense - retain the features of the road system while providing the efficiency of trains. I like it.
A warrent is issued (real or trumped-up fascist charge) for your arrest. Big Brother’s IntelliRoad knows where your vehicle is - and it then dierscts it to the nearest police station and blows the horn (after sealing up the vehicle, of course.
The moment that scenario becomes a possibility, there will be tons of brand-new IntelliRoad controllers clogging the rivers
This is also one of my pet topics at parties.
A few thoughts that I would have on the system.
On that point…
There are plenty of new towns springing up around the world. Such towns could have roads designed with self-driving cars in mind, possibly with tags or electronic devices at certain points to make the task easier and safer for the AI.
I’m sure it’s feasible to have a self-driving network on purpose-built roads, even if they have to share with conventional traffic.
Cars can be tracked for the purpose of road allocation, congestion pricing and train organization without compromising privacy.
Cars could communicate their location but not their identity in many different ways. There are already private traffic monitoring services providing traffic data to GPS receivers. They work by having cell phones equipped with GPS and mapping software broadcast their locations to private servers, which integrate the data and rebroadcast it as traffic location services, with no identification of the traffic in the data.
For the purpose of congestion pricing, the cars don’t even have to transmit at all. They can simply download the road price data and real-time updates, record their road use, and upload the final cost to the government servers along with payment. All they’d ever know is how much road you used up, but not where and when. Those systems can be made tamper-proof. They’d be the modern equivalent of a water meter on a house, which reports how much water you used, but not who used it and when and for what.
So… no gyroscopic jet packs? 
Not even jumpsuits for the military?