Raspberry Pi Compute Module

We’ve talked about use of the popular Raspberry Pi computer as a robot controller before.  There’s the RAPIRO, a Japanese biped designed specifically for that; and then there’s the news that Mathematica is available on the Pi for free.  But the Pi wasn’t really designed for this sort of use; it’s bigger and more power-hungry than we would like, and doesn’t have as many general-purpose I/Os (GPIOs) as we might want.

That’s about to change.

The Raspberry Pi Foundation has announced a new board, the Raspberry Pi Compute Module.  It contains the core functionality of the Raspberry Pi, including the same BCM2835 system-on-a-chip processor and 512 MB of RAM, but in a fraction of the size.  The 68 x 30 mm board weighs just 7 grams.  In place of an SD card, which would be rather bulky at this scale, the Compute Module has a 4 GB eMMC Flash storage device built right in.  The bottom edge of the card is a standard 200-pin DDR2 SODIMM connector, which exposes far more of the BCM2835’s I/O than was available on the standard Pi.

They foundation is also designing a separate IO Board, into which the Compute Module plugs.  This combination provides much the same functionality as a Raspberry Pi, though with much more I/O.  You can see a prototype of this in the video below.

However, the real excitement here centers on using the Compute Module on a custom minimal interface board, to control (say) a robot directly.  You’ve got a 700 MHz ARM processor, 512MB of RAM, a full Linux stack, plenty of disk space, and obscene amounts of I/O on a board tiny enough to fit comfortably into just about any robot.  (And did I mention that it runs Mathematica?!?)  I predict that by mid-summer, we can expect to see all sorts of robot interface boards built around this.  But it’s simple enough that just about any hobbyist could design and implement their own solution; all the tricky bits are already done on the Compute Module itself.

The Compute Module will be available starting in June, at a price of $30 in batches of 100.  Individual units will also be available for a bit more.

[Source: Raspberry Pi Foundation]