Meeting Notes from 7/27 and Week of 7/30

Meeting Notes

• Discussed weight distribution (drive wheel skidding). Determined that it may be possible to slide the frame backward to improve the weight distribution. Otherwise, I can cap the acceleration or add weights to the back of the robot.
• Miles will be giving a Gazebo presentation on August 13.
• Shoot for a demo for the end of August to show Ford/ABB.
  • Should be a video of Abby that demonstrates coordinated drivetrain and arm motion.
  • At minimum, this should include driving point to point, and performing an arm motion (such as raising and lowering).
  • Ideally would involve picking an item from a shelf and placing it at a workstation.
• Needed for this demo:
  • Path following (precision probably not needed, but is a plus)
  • Arm motion (pre-choreographed at least)
  • Gripper
  • Shelf
  • Standard carrier (box) and shelf for full demo
• Wyatt recommends I look at Jason Harper's thesis on localization using image processing of floor tiles (Hough transform), but Jesse Fish's thesis downplayed the usefulness of this technique.
• Steph's work is on recognition of traversable surfaces (floors) using point-cloud data from the Kinect.
• Chad experimented with optical flow for odometry with Otto. Wyatt isn't sure why he didn't choose to use it. I am considering optical flow to avoid the problems of wheel slip, but if Chad had difficulty with it, I won't bother.

This Week

Got RRI files used in 109 from Russell. These included RRI configuration files for the robot (XML files in /HOME/GSI directory), RAPID code, and extensive C++ code for the PC side. Unfortunately, none of it is really simple enough to use as a unit test. I spent most of Monday reading through the code, and couldn’t find any smoking gun as to why it works and mine doesn’t. I spent a good chunk of Thursday playing around with trying to get mine working.

Notes/Things I've Tried

• For all of these tests, I am attempting to connect to a simple UDP listener server written in Python, running on my Ubuntu Desktop
• I hadn't been capitalizing the names of my XML configuration files (though this shouldn't matter on a Windows filesystem). I corrected this, but it didn't help. Still getting error 112600 (ERR_COMM_INIT) when I call SiConnect
• My Settings XML file was missing an empty <Servers> tag. I added this, but no change in error message.
• I tried using the example files posted in Henrik Berlin's post here, with no success. Even using this supposedly tested and working code (I modified it to use the correct IP address), I still get nothing.
• I tried the above example code using the service port, wired directly to my Ubuntu desktop. I modified my UDP listener to listen on the IP address assigned by the IRC5's DHCP. Still nothing.
• I've tried rebuilding and redeploying the system, as suggested in another thread on the forum linked above. This also didn't work.
• I'm beginning to suspect my UDP listener, but it can receive messages from a simple UDP sender. Perhaps the IRC5 is expecting a response?
• Using Wireshark, listened for traffic from the IRC5 when it is trying to connect. Nothing. This suggests bad config files, bad network config (although I can connect to the IRC5 just fine with RobotStudio), or RRI is somehow broken.