Sunday, April 6, 2014

I am excited about this Pixy Vision (CMUCam5) sensor that stared on Kickstarter a while back. The project started shipping mid-March 2014, I was an early backer and received 'award' last week.

Out of the box it comes with decent work-in-progress documentation with a enough basics to start playing with it. Given this is a Kickstarter effort and not a full commercial product, I don't expect it to be flawless or perfectly documented. However the group did really a nice job with packaging and providing the quick start info!

I was able to get the hello_world with pan & title rig running immediately with no microcontroller. Next I hooked up an Arduino to demo the microcontroller SPI interface option. I am rusty on C/C++ so there were a few things that I was easily confused on. However, I am now running on the SimpleIDE though... I did try some of the Arduino propeller C libraries, no go. Need to speed a lot more time on Prop C first!

Not having much experience writing an interface to sensor from scratch I struggled on basic stuff here. However using Spin and Propeller tool I was able to generate similar Pixy SPI data output on the Propeller as in the Hello_World.ino example.

My attempt was a learning exercise on a couple different levels... I couldn't get the SPI_Spin object to work correctly. I ended up using parts of the Parallax Semiconductor 'Application Note AN012' to make this work. Leveraged a USBee LA also to measure and replicate the Arduino communication. Again, because of my rusty C/C++ skills getting through classes, template classes, etc. This time I ended up reinventing the wheel... Success is in what I learned along the way.

In the picture: TV running Propeller tv_text.spin object. PixyMon - PC based monitor/configuration tool for Pixy. USBee suite running the LA. The data on the tv represents SPI data output from the Pixy. The green box is the object Pixy is 'trained' to report on.

NEXT STEPS: Do something with the data besides look at it. I have the Zumo or Sting Ray that will likely be the target. Perhaps phase 3 of the Table Top challenge...

(Thanks to FIRST FRC Team 701 - Robovikes for the green ball, works great for tracking testing)