Progress Update: February 23, 2013

Working with Patrick Goebel to fix some issues with non-Chrome browsers, we now have Pi Robot’s HTML browser working for PR2Lite on an iPad, Firefox, and Chrome.  It didn’t work on an android 4.0.3 Asus Transformer TF101.  Give it a try!
Patrick has renamed the package to pi_remote and it is available at: http://pi-robot-ros-pkg.googlecode.com/svn/trunk/experimental/pi_remote

Note that there are now two interfaces, the old one with the “trackpads” and a new one without the trackpads but with arrows instead.  To pan/tilt the head in the new interface, you click on a spot on the video image and the robot moves its head to look there.

To bring up the old interface, go to: http://localhost:8181/pi_remote_trackpads.html
and to get the new interface go to: http://localhost:8181/pi_remote.html

2-23-13This is a screenshot with from PR2Lite using the old interface.

Advertisements

Progress Update: February 16, 2013

Matthew experimented with the torque, speed, and limits of the 2 AX-12 servos that control the velo gripper. The gripper can lift an impressive amount of weight.

The 6dof markers in the Warehouse Viewer for Arm Navigation suddenly stopped working. Looking at past ROS Answers, the suggestion was “network problems.”  After a ton of debugging, it looks more related to concurrency and resource contention bugs than a configuration issue.  Similar problems were identified by other ROS users. Alan backported some rosgraph fixes (ROS ticket 3946 and ticket 3947) for similar symptoms, but unfortunately the problem persists.  The only other bug along these lines is that roswtf has a bug with the graph processing.

Frank recently fixed a bug and now the Linear Actuators return their status very frequently.  These messages would be processed and pushed down a chain of processes and messages: datagrams to packets to linactuator to jointstates which were processed by the environmental service, fake joint publisher, and follow trajectory.  The effect of the large increase of low level messages was multiplied by the processing chain.  We’ve now collapsed the chain and throttled the JointState messages being produced.  Unfortunately, this reduction of load and concurrent processing didn’t fix the problem with the warehouse viewer.

Other Note Include:

  • Our git repository has been renamed to : https://github.com/rqou/prlite-pc
  • Frank has been dubugging interupt processing that results in a deadlock withing the firmware
  • Frank has been trying to reflash one of the wheel microprocessors.
  • improved pr2lite urdf
  • the imu has been integrated
  • Need to send in Maker Faire appliction.
  • Started the GUI based on the one for Pi Robot
  • Started the move to the PR2 Tabletop Manipulation stack.

Our First Velo Gripper Prototype

We got our very basic, first prototype for the Velo Gripper running today.

Alan got the joystick to work for the AX12s on our gripper, so those rotate depending on how far you move the joystick on a controller forward.  We have to make sure that the servos do not pull farther than they should when grasping objects (it could mess up our setup/servos), so for now we are just being careful about how far we move the joystick.

I drilled some holes and nutted some bolts to get two AX12’s to work in tandem.  This rotates a wooden axle, winding a wire to pull the gripper shut.  From our first tests, we can control the size and pressure of the grasp by how much the wire is wound up by the servos.  We will need to work more on the joystick programming so we can get the most accuracy from the controller.


Our own first prototype in action!

Thanks to Matei Ciocarlie and Willow Garage for the Velo 2G Gripper!

On Thursday, we received one of Willow Garage’s 2G “Velo” Gripper, a tendon-driven gripper which grasps around objects and can run off of one actuator.  We are going to get it to work with PR2Lite, using it in replacement of one of our hands.  We look forward to mounting it and testing with it.

Thanks to Matei Ciocarlie and Willow Garage for the Velo 2G Gripper!Here is the Velo we got (no Ethercat, but look at that snazzy handle).

First off, I will be working on making a mount that will work with our wrist and be able to pull the gripper closed.  Currently, we are hacking together a very basic prototype just to make sure that two AX12 servos can close the gripper.  It is surprisingly difficult to pull the tendon, and according to Willow Garage, it requires 17mm of travel and ~150N of pull to hold a 1Kg object.  We do not have an Ethercat, and are going to make our own thing to pull the gripper closed.  We’ll keep updating through the weeks as we get that working and mounted.

 Thanks to Matei Ciocarlie and Willow Garage for the Velo 2G Gripper! 2Here is the Velo Gripper next to our current gripper.  As you can see, the Velo is already larger than the current gripper, and it will be even bigger and heavier with servos.

Thanks again to Willow Garage!