Can anyone suggest me a technique ?
What have you explored so far?
http://www.robotshop.com/blog/en/how-to-make-a-robot-lesson-7-using-sensors-2-3683 shows almost any type of sensor you could use as well as a means to limit where it may go.
Clem
i have thought of using a line of ir sensors.........along the side of the arena............but then............it gets too much expensive and also the sensors do not give me the distance of the robot from the side of the arena
Sounds very much like a college project.
I was in a Mall the other day and there was a matt on the floor with a projector overhead. The children were stepping on colored projections of objects and their feet were being detected and reacted to by the projector. Perhaps an overhead camera viewing your arena would be able to view an LED on the Robot and a computer could process the position from the map of the image.
John
I think like John Wiltrout, the camera seems a good option.
May I ask if you have think about triangulate the position with bluetooth? One in the robot and other three as beacons.
I know it is not easy to programming, but it will be accurate.
if you can put down a floor covering, it could be made from a matrix of sensors. TI make a interface chips for capacitance or proximity or load. You could determine which cell of the matrix the robot is in. Peter
You could use an infrared strobe to trigger an ultrasonic ping on the robot.
The strobe would also start a timer on each of 2 ultrasonic receivers - when they receive the ping, the time of arrival will indicate exactly how far the robot is from that receiver. 2 known distances will uniquely locate the robot on a flat surface.
Your robot could have one ball in the center in contact with the ground and then X and Y counter wheels that pulse count its movement. Heck you could probably hack a computer mouse to do it. Then once you set it's home position, you can locate the robot using an X and Y position from the 0,0 home position. Realize that if the robot gets lifted or sufficiently bumped that all bets are off. You could consider a combination of methods, to keep your accuracy.
A top down view will probably be the most accurate method though.
Good luck!