Progress Report – November 2013

Alan showed off our new arm at the October HBRC meeting.  This arm was designed by Matthew and assembled by Matthew and Alan.  It features a velo gripper, 2 MX-64s and 2 MX-106s.
At HBRC, Alan welcomed interested members to join us on Saturdays.  Alex and his son Maxim joined Alan and Frank the last few weeks and have made an immediate impact.

We’ve begun putting PR2Lite on the ground again.  PR2Lite has gained a lot of weight since the last ground runs, and the pivoting wheels have not been straightening properly.  Maxim added springs to encourage the wheels to complete their rotation.  The springs have enabled PR2Lite to drive again, but we want to further increase the reliability.

The linear actuator in the base is responsible for rotating all four wheels in unison to be in one of 3 states: forward, rotation, sideways.  The base was designed assuming the full 4 inches of linear actuator extension.  The linear actuator provides 3-wire feedback to determine its extension.  Unfortunately, the feedback is off the min/max by about 10%.  Frank is modifying the higher level code to compensate for this imprecision to get full extension.  With the springs and full extension, it seems likely that PR2Lite’s floor navigation will be satisfactory.

Next Alex and Frank are going to get PR2Lite navigating on the floor using SLAM.  PR2Lite’s software supports ROS slam, including the accelerometer.  The software appears to be working on a pedestal, but it’s time to test on the ground.  As a point of comparison, Alex and Maxim have been getting their Neato to navigate using SLAM.   Maxim presented his Neato progress at the November HBRC.

Image

We’ve now installed the new right arm on PR2Lite and reprogrammed the servo IDs.  Based on preliminary testing, the new arm looks good and strong!  Next, we will do full integration of the new arm into the ROS configuration.

Image

Alan has made steady progress on the new laptop upgrading PR2Lite to Moveit and Hydro.  Based on advice by Mike Ferguson, Alan skipped trying to use Open Rave – which wasn’t going well.  With Moveit’s new support for “mimic joints”, PR2Lite’s arms and the velo gripper are mostly natively supported.  Alan has been modifying Moveit code to overcome migration hurdles.  With the improved momentum, Alan is getting more optimistic about the migration.

Matthew now has to focus on his college applications!

Advertisements

Progress Report: October 27, 2013

I went to my friend’s house to help build 3 new chassis for our school’s FTC team.
Alan and Frank set up PR2Lite to have its own wifi hub and connected the pr2lite_rosbridge GUI to it and drove Pr2Lkite around the garage for the first time in a long while.  Based on this test, we’ve decided to upgrade the parallax encoders to the 36-Position Quadrature Encoder Set so that we can improve the pid accuracy and hopefully get wheel speeds more synchronized.
Alan has been trying to upgrade PR2Lite to run on Ubunto 13,04, Hydro, and MoveIt.  Progress is painfully slow as each step taken fails out-of-the-box and requires extensive deugging.

Rush Week for AHS Robotics Club

Unbeknownst to most readers, I am the president of the Robotics Club at my High School.  Every year we have one crazy, busy week to encourage people to join clubs.  This is called Rush Week, and I’m proud of this year’s stand.  We brought in our own table for every of the 5 days, and built 4 example robots in less than a week’s time.  All in all it was successful, with about 75 signups.  Here are some pictures from that week:

Out VEX Robot, built and programmed in 3 days autonomously moved around on the table, and also was driven around on the rotunda floor.  A servo waved around a picture of a robot.

Rush Week VEX Robot

Here is an additional picture of our booth at Rush Week, complete with line follower (bottom), tri-fold, and decade-old trophies (and quite a few new ones as well).  It’s probably the best Rush Week we’ve had in years.

Rush Week

Progress Update: September 28, 2013

The Velo Gripper wrist is nearly complete!  Here is a separate update for just the Velo Gripper wrist.

We have also detached the right arm to make new drill holes to accommodate the MX64 and MX106 servos we’re putting in.  The shoulder’s MX64 servo (not pictured) fit in with minimal effort, which is nice.

Removed Arm

We were having troubles with the long cables that controlled the linear actuators and wheels; the signals would get too many collisions because the wires were too long.  Frank had trouble converting the cables into smaller cables, DB9 connectors, and RS485 signals originally, but has since gotten them to work.  Version 1 did not work, but Version 2 did.  Both are pictured below.

Franks boards

We also have a new Ubuntu System 76 laptop (quadcore i7 with SSD) which will replace our current computer in the robot, and open up some space.  We’re also moving to the Hydro ROS release and moveit! arm control sometime in the future.  Here’s a picture of the laptop:

New Laptop2

Velo2G Gripper Progress Update: September 28, 2013

We’ve made a large step in our integration with Willow Garage’s Velo2G Gripper.  I completed my design on AutoDesk Inventor 2013 of the wrist, which controls the Velo Gripper and rotates.  I also designed other parts which will complete our upgraded lower arm.  I had to change holes and parts to accommodate for larger servos like the MX106 and MX64.  Here is a screenshot of the final drawing.

Velo big update1

We ordered the pieces to be cut from Mr. Plastics in San Leandro, got them about a week later, and started putting the parts together.  The pieces fit in snugly, so glue is not entirely necessary.  Here is a picture of the wrist attached to the Velo2G before glue.

Velo big update 2.1