robots A-Zrobot timelinegamestv & moviesfeaturescontactyoutuberss feed
bg logo logobump

The DARwIn-OP Might Be A Game Changer

Some new details about the DARwIn-OP have emerged from the Humanoids conference, and I must say that Aldebaran Robotics had better be worried about this bad boy.  This is an open source, hackable robot which is not only thousands of dollars cheaper, but has comparable if not better specs and Robotis’ reputation behind it.  Users will be able to modify it to their liking while still taking advantage of a larger community’s progress.

It’s an interesting project; a collaboration between several U.S. universities (RoMeLa at Virginia Tech, Purdue University, University of Pennsylvania) and South Korea’s Robotis.  Originally there were going to be two completely different models of the robot; the DARwIn-HP (High Performance) based on RoMeLa’s DARwIn 4 RoboCup platform, and DARwIn-LC (Low Cost) based on Mini HUBO (pictured).  Mini HUBO was created so that researchers could test walking algorithms on a small, inexpensive platform before trying them out on the full-scale Jaemi HUBO.  However, in the end they decided to design a new robot from scratch, and since its hardware and software was going to be open source they changed its suffix to OP (Open Platform).

Although the DARwIn-HP is not for sale and will not be open, the DARwIn-OP is fully open source and hackable.  All software is based on a generic platform, so you do not have to use Robotis’ software that comes with it (C++, Lua, LabVIEW, Matlab, ROS, Windows or Linux, anything goes).  You’ll be able to download free CAD files for the parts to make them on a 3D printer, and detailed assembly instructions to put it all together.  Robotis will be selling the parts separately, so anything you can’t make yourself you’ll be able to get from them.  And since it’s modular and you can put it together yourself, you won’t need to send it away if a new component becomes available or you need to replace a part.  You can easily add whatever components you’d like such as sensors or grippers.

Most exciting of all, the new RX-28M Dynamixel servo is an exceptional actuator with new non-contact magnetic encoders for feedback, which significantly improves its life while increasing the resolution by 4 times (4096 compared to the previous 1024)! The M stands for metal gears, which are more durable and long-lasting, and it sports user-modifiable PID gain and TTL.  Robotis servos are already the gold standard for RoboCup teams around the world – and the RX-28M is more reliable than ever before.  And it’s powerful: the DARwIn-OP can move at 24cm/sec, and (unofficially) is able to run – where both feet were observed off the ground simultaneously.  And this is before the gait has been optimized.  It can stand up in less than 3 seconds when facing down, and less than 4 seconds when facing up.  Despite supplying two CPUs and the power-hungry servos, the battery lasts 30 minutes and has a standby mode for lower consumption. Less expensive educational robots, like PALRO or Vstone’s Robovie-PC, cannot hope to match its performance.

Artificial intelligence aside, the dream of owning a QRIO-like robot may be just around the corner.  Personally, I’d like to see the DARwIn-OP get a more complete enclosure; by which I mean I’d like to see the servos completely hidden.  Give it some Crafthouse hands for grippers, and maybe stereo cameras instead of the one.  It could happen.  Its adaptability is a big reason to get excited about it, which (incidentally) would make its namesake proud.

[source: DARwIn-OP @ Sourceforge]

  • Do you have any ideas on a good way to handle the hip/ankle joints for a complete enclosure? Dual servos at these joints are just HUGE!

    • Robotbling

      That’s definitely a problem area. Particularly with the Robotis servos which are pretty bulky compared to some other brands. But there are other areas that might be refined.