How hard would it be to write a program that takes input from 4 wireless high-speed digital video cameras and use it to be able to drive a car? My idea is to write this program, put sensors on the manual controls, and use RC servos to control controls. Use the sensors slaved to another comuter in case the main one crashes. How hard would this be? Am I foolish to think it can be done with today's technology? Should I wait another 10 years before this is freasible? Came up with idea in 1995, at age 11.

I know the feeling , coming up with an idea and then finding that it is already out there. I thought of the Greenie wirenut years before it actually came out. Not exactly robot cars but it was something.

If you can write this program , you can get a good job at the Jet Propulsion Labs . The link explains about the software used to control remote vehicles.

And there is a site from Georgia Tech that offers free opensource code for a remote control package.

I am sure this technology is being used by military and exploration more these days, and is not ready for inexpensive commercial applications.

I'm sure it can be done but it won't be easy (not even close). The whole problem would be processing the images from the cameras.

I don't think you'll be able to do this on your own. This is the kind of thing that takes large research teams years and years to do. Doing it on your own, in your free time is impossible.

Hi,
There is also the possibility that your idea already has a patent. The Network News Organizations have already reported that a vehicle is being tested for "driving itself". No driver. No wiring to the highway. A passenger tells the instrumentation where he/she wants to go, and it drives you to the place!
The presentation showed the vehicle making it's own turns on city streets, parking lots, etc.

Well, lets look at this in parts. Controlling an rc car's remote controller, using servos, is easy, very easy, in fact it's a simple robotics question. On the other hand, writing the software to take four independent images and from that determine what makes up the area its in, where it can travel, where it cant... is a difficult problem, and then on top of that you have an element of pathfinding, programmers spend hours and hours, a single programmer could spend as much as 3 months or more (depending on experience) tackling the problem of artificial intelligence pathfinding, and that's all AFTER the hardest part. To interpret 4 separate video feeds is tough. If you can program a simple text interpreter (takes a picture containing text, and from that can read the text) then you're part of the way there. Computer visual recognition is a developing science, and because of that you'd be totally on your own.

Now on a different note, do you have any experience with programming? If not, learn, you can't even dream about actually accomplishing this without being an expert programmer (or have some working for you).

Cheers (I hate to be a killjoy, but visual recognition is... hard... at best)
-Dan