Hello,

I am currently working on a course entitled GIS in business. To complete the course I am trying to do a paper on the prospect of a car that can drive itself:

The basic idea that I had is the heart of the system would involve the use of a GPS to steer the car and a GIS to monitor where optimal routes are available (as well as the position of the car relative to other cars on the road),

This alone would be insufficient, of course. There are always going to be some obstacles on the road that the GIS will not know about in advance. Even if a law would be passed requiring as a safety feature that every car must be autopiloted so that a GIS could theoretically monitor the location of every car in the country (which might be difficult due to groups like the ACLU), there will always be things like debris, animals, sharp objects, and most importantly, pedestrians running around. Thus, it appears evident that the car would need an alternate system to find its way around. This is why I find your invention so fascinating.

I have seen similar ideas used for the purpose of helping blind people navigate where a sonar is sent multiple times per second and is connected to a gps which then steers a wheeled stick around obstacles. I am assuming that your vehicle uses one or more comprable systems.

If I understood correctly the system can detect objects around it at speeds quicker and intervals more often much more quickly than any human driver can look and react (I would like to know a specific number of how many times per second the sonar or radar is released).

I also am trying to find information on any or all of the following topics.

1. How accurate is a GPS?
Obviously, if a GPS would be used to control a car's motion it would need to be much more accurate than your average navigation GPS which I have seen advertised for 3m accuracy starting from about $150. I have seen suggestions that a GPS can be as accurate as an inch, but I have not found out a price for them. (Again it is also important to know ot only how accurate the reading but also how frequent)

2. How feasible is it for one central computer (or network) to control the motions of thousands of vehicles?
How much information would it take up on a computer's memory to be constantly calculating the distance between the car ahead, the car behind, the car on both sides etc., and finding the optimal route to input commands to the car, for one car (and then in turn for thousands of cars).

If you know anything about any of these topics (or where to find such information), I would be very appreciative.

There would be other issues such as security, reliability, and legality, but for now I am more interested in the technical feasibility.

Thank you,
Joshua E. Pollack