My idea would be to use a GPS module on an Arduino-based 'bot, and then carry around a small second GPS module, and have the robot use its own location data and "my" location data, and follow me.
Feasibility? I have a pretty good idea on the 'bot portion, but not really the second. Ideas/suggestions?
An interesting idea, and certainly possible. The biggest issue you're going to face is probably accuracy. While in a car it's pretty adequate, GPS has a pretty wide margin of error. They tend to be accurate to a couple of meters/5-6 feet. While that may seem pretty tolerable, if you take a turn your robot could take the turn way too early or way too late. So you'd likely need to augment your location with something in addition to the GPS coordinates. Also, if we suppose the coordinates are 100% accurate, you're going to have to implement a sort of "Follow me" type procedure. This would ensure that the robot would follow you exactly. Otherwise if you're 10 feet north and 5 feet west of the robot, it'll try to go diagonally toward you, and might run into a wall, or into a pond,or something.
That out of the way, your second device is the simpler. It just has to receive the coordinates from the GPS and send them to the robot. You can get a simple transmitter/receiver pair for pretty cheap at Sparkfun.
So, a possible solution would be to use a line-of-sight thing like... Using a PC to get the coordinates of an IR light from a Wiimote, and have an IR LED on my shoe or something.
While this would be really easy, I still don't really get how the GPS would work, if someone could give me a psuedo-code-like block diagram idea or at least an "idea flow" of what would be going on between GPS receivers and the Arduino. Then I could understand how to code... :\
How close is it going to follow? Would there be a delay?
On a good day GPS is accurate up to 12 feet. On average with a day that isn't too cloudy with direct line of sight to the sky 15-20 feet is typical.
I wonder if some type of radio signal on the leader and a signal strength detector on on follower. It would be like those automatic solar panels that use two photo sensors. It keeps rotating until the sensors read the same amount to keep the panel looking directly at the sun.
Ever seen RoboCup? The standard soccer game uses an IR ball that is detected by the robots, this may be a good starting point.
@Sirpoonga: I'm thinking a following distance of approximately 3 feet. (~1m, for the rest of the world)
@fliggygeek: I'll look into that. Luckily at my college there is a robocup team. (Carnegie Mellon University) I'll see if I can get some responses.
So as more people respond, it seems GPS is going to be useless, since 12ft deviation is way too much. Especially if it loses sight of (both?) IR lights. Any other suggestions for a non-line-of-sight tracking method?