I have a picaxe and an Arduino, I'm trying to make a robot that scans for a golf ball, picks it up, then returns it to me, I found the CMUcam, but it is very expensive, is there an easy way to do this task? Am I overthinking this?
thanks :)
Can you describe where the
Can you describe where the golf balls are? On a nice flat surface with nothing else around? ANd how the robot is configured? If so, then some Sharp GP2D12 sensors could be panned across the area, and see the golf balls as close objects. If panning and there is a short, golf ball wide, short range indication, then that would be a ball, and the robot should head that direction to pick it up.
THe CMUcam would only work if the golf balls were of one bright color that could be discerned from the background easily, usually a flourescent orange or green.
Im not sure how much the
Im not sure how much the CMUcam costs, but you can use tracking software with almost any video imput feed to my knowledge.
@robologist The golf balls
@robologist The golf balls would either be on a grey carpet, tile floor, or green astroturf, I’m leaning towards using the astroturf so they would be easier to spot. I have a sharp GP2D120, I’ll have to hook it up to an oscilloscope and see what it gives as an output when it sees a ball as opposed to something else. I think it may be difficult because the ball is a round object and when the waves hit the round object they dissipate slightly before returning.
@icon the cmucam is usually 100-150 dollars 
Still looking for sensor
Still looking for sensor suggestions, I’ve pretty much hit a brick wall, thanks 
Any sensor that can "look"
Any sensor that can “look” ahead and direct the robot to the item will likely be costly as it requires a computer to process images. You could have the robot roam around until it runs into something and then check out the item to see if it is a ball. If you are trying to stay cheap then looking ahead to “see” a ball and drive to it will be very difficult to do on a limited budget.
I am working on a similar project and I am just having it drive around until it finds the object then it scoops it up and determines the color then it either keeps or rejects the item.
Do you want it to function
Do you want it to function like this: http://gizmodo.com/gadgets/gadgets/golf-ball-finder-goes-hitech-169088.php
Or you could try to make an imbedded chip like this: http://www.radargolf.com/
you might be able to scan through these filters: http://www.gbfgolfballfinder.com/
I was debating using
I was debating using something like this http://www.sparkfun.com/commerce/product_info.php?products_id=8924 and just painting the ball a certain color
This sensor has to be almost
This sensor has to be almost right on top of the object to see what color it is. The same thing can be achieved by using 3 LEDs (1 each of red, green, and blue) and a photoresistor. You shine one LED color on the object and measure the resistance of the photoresistor. You do that for all 3 LEDs and the results will tell you approx what color it is. If the object is red the resistance with the red LED will be much different than the other 2 since the object is bouncing back more red light than any other color.
I don’t think the ball is the problem…
I am working on a system to simply let my robot know what room he is in, let alone how to get from there to anywhere else. This would seem quite an easy thing to do at first thought…
You could easily make a non-brain ball pick-up unit and do a sweep of the room to get what ever it runs into. The major problem here is the “return to me” part. It took me over a month of work everynight to simply get my robot to find and center itself between two IR beacons!