We're currently doing an evaluation of some POB Technologies products, and are trying to decide which particular project would be best to demonstrate their power and capability. We have lots of different robotic platforms, some wheeled, some with tracks, and quite a number of humanoids, so the best project would involve one, or more, of our existing robots. And it should be something of practical use - something that can be applied to autonomous operation, tracking, or even just a simple fun project as long as it's not trivial or meaningless.
In going through the specs, manuals, downloads, and forums on the POB site, we ran across fairly detailed documentation showing a mash-up that uses the POB-EYE and related modules to give an original Robosapien robot vision and the ability to track objects.
Basically, the POB/Robosapien hack involves decapitating the robot to gain access to the robot's control input line, modifying some of the plastic body shells to mount the POB boards, and installing some mounting frames.
Although we do have one of the original Robosapiens, it's become more like an old friend and kind of a collectors item at this point, so we really don't want to chop its head off or drill holes in its yellowing white plastic body.
What we're considering, and would like your feedback on, is to build a small backpack that would fit on either Gulliver (KHR-1/RCB3), Mondai-Noid (Manoi AT01), or our new RoboPhilo, then mount the camera/eye module on the head servo. The POB to RCB3 interface could be a bit tricky, but the RCB3/H2H3 could certainly handle the commands coming from the POB dynamically. In contrast, interfacing it to the RoboPhilo's ATmega32 16 controller should be pretty simple. The downside would be having RoboPhilo handle the additional mass of the backpack, and the need to do some non-trivial C programming on the robot side to make it all play.
The final test, the proof of the pudding, would be for the POB-EYE equipped humanoid to track, and attack, another robot identified by a large symbol on its chest. Here's a video of the POB/Robosapien hack in action to give you the idea:
So, what do you think? Is it interesting enough, and challenging enough, to make a good evaluation project? Let us know. Or, if you have any other ideas for a good robotics project using the POB modules, we'd love to hear about them.