What about a camera with fisheye for 180° vision and a transmitter with IR light? This would give you a very accurate direction.
Only drawback is that you can’t have obstacles between transmitter and camera.
@ WouterH - Would the camera thing work? If it would, that is definitely something I can evaluate. I am not too much worried about obstacles as I expect a clear line of sight. Do have any recommendations for where I can get such components?
I also thought about the digital compass. But, ruled it out as I too felt it wouldn’t work.
Thanks.
Depends on how fast the object moves, what is the environment, area size etc… Do you have something specific in mind?
@ WouterH - The closest I can think of is the Roomba analogy. I just want the receiver to mark the path (for now, just the direction) that roomba took. How many lefts, how many rights, how long in each direction, how many back-and-forths, etc. And I need to aggregate this information, which I can do as long as I can ‘trace’ the roomba.
Thanks.
What about wheel encoders, perhaps in combination with a compass and/or distance sensor(s)?
That was the first thought I had in reading Prashanth’s reply. I’m working on a similar problem, figuring out the bot’s position in relation to a “know” point based on GPS, wheel encoder measuring distance traveled along a direction provided by digital compass - basically trying to keep the rover on track between two waypoints and have itself know where it is in relation to the corners of the polygon.
I plan to have some decent amount of work put in next week on my library, so will be glad to share if it will help. I realize GPS isn’t going to do you much indoors, but the encoders and compass logic may be of use.
You could put omni-directional IR transmitters on the moving unit and then have a direction IR receiver on the station. You could have directional blocking vanes on the receiver kind of like they have on some stoplights so that light can only pass within a few degrees of straight on. You could then have either multiple receivers at various angles, or have one receiver on a servo that sweeps back and forth. That would give you a pretty accurate direction to target. With one on each axis it is very simple to draw the intersecting lines to figure out where it is.
If your target is moving quickly you could use the sweep method with 3 sensors so that as the target moves to one side the outer IR will pick it up and you can sweep to that side to keep it on center.
as an extension to PrintSize.Me, you can have a single IR receiver with a black, round plastic cap over it. The plastic cap contains a small rectangular hole. The cap is rotated by a servo with encoder or stepper motor. The position in which the IR receiver received the largest signal is the angle to the target.
It wont help here but interesting non the less, watch the the 2nd video BAE's Striker helmet gives fighter pilots 'X-ray vision' - BBC News
They use a sequence of LED’s on the helmet to track it’s orientation.
There is this:
http://www.seeedstudio.com/depot/omnidirectional-sensor-p-1215.html?cPath=144
But who is willing to spend that kind of cash, just for the lens…
@ GMod(Errol) - reminds me of the iPhone 360° lens:
But a hemispherical lens should work fine too. That’s how Mobotix does it in this product: Products
But for the price of the seeed lense you can buy an Iphone, or two, AND a Dot… 
@ stevepresley & @ mhectorgatoHero - Using wheel encoders, though a good idea, might not work for the application I have in mind.
I am doing this as a fun project and I agree with the rest here … who is willing to spend that kind of cash!
@ PrintSize.Me & @ WouterHKing, you guys have a good idea here. I can take it up if the cantenna thing doesn’t live up.
Thanks!