It's completely possible to run an Arduino based headless CNC machine! Here is the current setup I am using:

Hardware:
1.) SD Shield from SeeedStudio - as a side note: I hesitate to call this a "Shield" as it doesn't repeat the female headers on the top, so by using this you essentially lose access to your Arduino's V and GND ports. You need access to the GND port so you can tie the two GNDs from each Arduino together. I have circumvented this limitation by using two mini bread boards, as shown here: Also, it would be an easy enough fix to replace the current male headers with something like these, but I just have not gotten around to doing that.

It's not the most elegant solution, but it's working.... (as a side note: the USB cable in "master" is just to power the device, not communicate with it. I have since switched to a wall wart!)

Software:
Master Arduino: In my setup, I am calling the "master arduino" the one that parses the gcode. For this I am using a program called grbl. IMO, this is the holy grail of arduino based gcode parsing. you can download it here, or read a little about it here. I actually found it via the contraptor project, here. You can see grbl in action here!

Slave Arduino: The arduino that is connected to and reads the SD card. After pouring over a bunch of different fat libraries for arduino I settled on Bill.Greiman's fat16lib. It was very easy to implement and did exactly what I needed it to do.

I have had some problems overflowing the Master's buffer, but this is mainly due to the fact that I'm having a hard time setting up flow control. What I need to have happen is for the slave Arduino to Serial.print a line and then WAIT until it receives the "OK" from the master. I can't figure out how to do that reliably, so occasionally i overflow the master's buffer.....

There's something wrong with the X axis. Unlike the Y, it's very difficult to move. Well, not very difficult but more difficult than the Y. I think there may be a bit of a design flaw... stupid designer.

From what I can tell, there is some friction between the X-axis-rail-mount-plates (the ones slotted end to end) and the X-axis-cross-plate (the one that the X-axis acme nut mounts to). I'll try taking some sandpaper to it and if that doesn't work, perhaps the dremel? Someone also suggested I use wax.. i don't even know where I'd find wax though. Also, I should have boxed in the X-axis cross plates because there is flex when the nut is being driven.

I'm going to move forward and continue with the current iteration. After the new power supply is delivered and the EasyDrivers are installed, I'll see where we're at. But, I'm afraid I'll have to make some changes to the design and re-order all of the parts from ponoko. It's not the end of the world but I hate the fact that I didn't get it perfect the first time...

Though, I should be used to that by now.

On a different note: When I started this project I was certain that a computer would be running the machine. However, I've been using the arduino to test the motors and when I get the 2 additional EasyDrivers I'm planning to run them through their paces with the arduino as well. I keep thinking how awesome it would be to use the arduino permanently as the controller. Yes, replace a full PC with a tiny arduino! My machine has a very small footprint to begin with, and I keep thinking how cool it would be house *all* of the controls in a tiny matching box! I don't even know if that's possible, but wouldn't it be sweet!

I'm imagining something like this: Obviously, do the CAD/CAM on a workstation but save the G-code to an SD card. Then, insert the SD card into the arduino, turn the arduino on, it would then look for a specific text file. Parse the file and run the machine! I know it's a lot harder than that and I'm doubting there would be enough room on the arduino to store the code to both control the reading of the SD card and parse the file and run the machine.  Either way, a guy can dream right?