I now have my Raspberry Pi based computer vision demonstration robot working! This is a big step forwards, as I can start looking at more interesting projects and competitions with it.

Specifications

• 4Tronix PiCon Zero
• Raspberry Pi 3
• V1 Camera module
• 4Tronix Initio chassis
• Logic power is a USB power bank
• Motor power is 6xAA batteries
• Raspberry Pi image created with PiBakery - and using precompiled NUmpy and OpenCV libraries
• Programmed in Python

Updates since last time

I’ve rewired to use the PiCon Zero instead of the L298n and all that point to point wiring. This has made it a lot neater, and there’s less to debug.

I’ve also separated the Raspberry Pi power out to a USB power bank. This means I can run the Pi without the motors, and I can also run the motors without the Pi. This is a good thing, as it means I can debug the motors without the Pi crashing. It also means that large current issues won’t reset the Raspberry Pi.

Video

The Skittlebot video tells the story of it’s creation, and adaptation from the Arduino robot, along with the first tests of the computer vision system.

Skittlebot was built when I found that armbot was too big to come to PyCon Uk 2016. It uses a camera and visual processing to locate coloured objects, and knock them over. See how I built it, see it running, why I made it and how the code was written.

The video mixes footage that was unreleased from multiple sessions, and includes the point where I’d shaved my head for a while!

Code

The software uses PiCamera, Python and OpenCV.

The code is at https://github.com/orionrobots/skittlebot.