Week 4

Progress

Our main accomplishments this previous week were as follows:

• Made and programmed pneumatic firing system proof of concept
• Purchased additional robot parts
• Setup virtualized desktop with Ubuntu and Windows to support robot arm ops
• Created automated setup scripts for DENSO apps
• Started construction of robot platform

This week we focused on designing our pneumatic control system and purchasing everything that we needed to make that work. For our main demo, we controlled a compressor to compress the air in our pressure vessel to a certain PSI using a bang bang controller. See the video below for a demo. The electronics were neatly organized in a temporary enclosure for the demo. We will be moving to a metal, earthed, enclosure for final implementation. We decided to focus on getting the backend of our actuator control working, and will mount the system to the arm in coming weeks.

We purchased many different parts for our robot, including our inflator and parts for our recovery mechanism.

We have also additionally started construction on the robot platform, as seen in the second picture below. It still needs wheels, and a lot of it will be cut down, but progress is being made.

We have encountered a number of issues with our DENSO control software, as we have an older version that cannot be initially activated over the network. To solve this, we create an AutoHotKey script that will detect error states and correct them automatically, as well as performing the initial initialization. Additionally, we spun up a hypervisor OS to run both Ubuntu and Windows on our desktop. Windows will run the DENSO software needed to activate the arm, while Ubuntu will run ROS to control the arm itself, sending information through the Windows VM.

Challenges

We have a member who probably has pneumonia, but valiantly pushed through to help us get the demo to where it needs to be.

We also determined that unfortunately, additionally because of the older controller that we have, we cannot use the ROS package for DENSO robots, so we will have to write a bit more complicated python to control the arm, and this also is what called for the hypervisor setup that we created.

We have also learned that the valve that we hoped was fast enough, turned out to be not even close. We can implement a known alternative, using a servo-actuated blowgun, to get where we want to be, though. See the video below to see our initial testing:

Plan for next week

During the next week, we plan to accomplish the following:

• Make new pressure vessel, potentially a final version?
• Proof of concept of catcher
• ROS motion control to arbitrary positions
• Initial ZED detection of ball