Thursday, May 20, 2010

Robogames Wrap-Up

Well, it's almost a month since Robogames, but it's been hectic with the end of the semester. The semester is now over though, so it's robot time again.

Issy did terrible in Mech Warfare. He walked like a champ, but had numerous other issues. On Friday afternoon his gun literally exploded inside. Thankfully, I found a Sports Authority around the corner and purchased an extra gun(have I mentioned how much more convienient of a venue San Mateo is over the old place?). Apparently though the Trendnet camera decided that 6V was just too much - it stopped working on Saturday, and Issy was dunn for. He's already been torn down, and retired from Mech Warfare.

He'll be rebuilt over the course of the next month or so, this time with a Fit-PC2, stereo camera head, and a tail. He's going to be running ROS, which ought to be cool. Here's a teaser shot of what he will (hopefully) look like:

On other fronts, I'll be posting some other robot goodies over the coming days. I'm currently working on some finishing software touches for our first "social" robot at Albany, Nelson. I'm also working on a robowaiter entry for next April.



  1. You should have done the walker challenge, it was a lot easier than MechWarfare : )

    Seriously though, Issy is an awesome robot. The new version should be very cool. Good luck.

    George Collins

  2. That sounds like an awesome plan. Id be very interested in how easy ROS is to use. The willow garage dev looks so cool, but if its easy to get running its bound to become a standard platform which is very exciting.

    Good luck

  3. George,
    Thanks for the comments. I'm hoping that Issy3 will actually be competing in the walker challenge next year -- his stereo vision should allow some interesting navigation.

    ROS is not exactly *easy* to get running. It is getting easier every day though, but it does assume a solid background in python or C++ (and preferably, both). I'll be posting a lot of stuff about ROS over the next few weeks as I start unveiling some other projects that are nearing completion.


  4. Mike,

    What is this project going to use for communication to main brain/controller? ArbotiX?

    Aslo, are you going to work on this project all by yourself?

    I think several people at Trossen forum will be very interested in this project. It looks like several of them have most hardware and alltogether we might progress faster with this project.

  5. Alex,

    Yep, the ArbotiX acts as a coprocessor and does all the real-time leg manipulation (gait generation, leg IK). It'll also do the foot sensor sampling (creating a tight closed loop outside of the PC).

    This design then leaves the PC free to do stereo vision, path planning, etc. It can send either high level trajectory, or footstep paths (on tough terrain) to the ArbotiX.

    At the moment, yes, I'm working on this by myself. However, the ROS driver is about 90% ready for an 0.1 release (hoping to do that this weekend). The ROS driver is actually being used on several projects I'm working on, and I think others will get a huge step forward with it (I'll probably post my thoughts on ArbotiX/ROS tomorrow as a full blog post).


  6. > the ArbotiX acts as a coprocessor
    So the equivalent for it in the nature would be something like reflexes - fast and automatic, no thinking envolved.
    Then PC will be like a brain - providing higher level thinking.

    > I think others will get a huge step forward with it
    I think you are right. It looks like there are quite a few people by now who played enough with Robotis microcontroller and are now ready for something better and more exciting.

  7. How are AirbotiX and PC going to communicate?
    Are you thinking about making it completely autonomous, or hybrid of autonomous and RC ("assisted autonomy"? :) )?
    If hybrid, wouldn't Xbee channel needed to communicate with the controller (Airbotix Commander)?

  8. I think we need to move these thread to Trossen for others to benefit/participate :)

  9. Please see