Our main accomplishments this previous week were as follows:
We have almost finished the CAD for the arm that fully encapsulates the end effector and the base station.
You can see a render of the almost final CAD to the right. Manufacturing has begun on the base (as you can see on the right).
Electrically, we designed a PCB to accomplish all of the tasks of our electronics system that are custom (e.g. not taken care of by the DENSO system). We designed an Arduino Shield that will mount on top of a SparkFun Redboard and will communicate both with our computer, and the Robot Controller. You can see a render of the PCB to the right. At "press time", the board was 50% manufactured having been submitted for manufacture last night.
Ball detection is proving challenging.
Ball found with simplistic PCL clustering methods, but hard to separate from background, noise, and other objects
Using PCL SACSegmentation framework, ball can be detected, but only with processing time taking 2-5 minutes. Not viable for us
We looked into deep learning options.
Naturally, computer vision has been proving challenging for us this past week. Mechanically and Electrically great progress has been made. From a vision perspective, however, we have not yet found a solution that will work for us.
Mechanical
Electrical
Software
You can see the report we presented during our checkoff below for a few more pieces of information: