Posted 08 July 2026
In this previous post, I describes my work with Grok Code to successfully demonstrate the use of vision processing with a Luxonis OAKD-Lite camera and a Raspberry Pi5 to generate ‘best clear direction’ advisories for robot motion control.
After achieving the above milestone, I started thinking about the next steps. Clearly I want to move on to achieve actual robot navigation utilizing the above ‘best direction’ information, but the current 2-wheel robot battery pack doesn’t really provide enough power for the Pi5, the OAKD-Lite camera, and the wheel motors. Consequently, I started thinking about moving the vision processing project from the 2-wheel robot onto my existing 4-wheel platform to take advantage of the latter’s larger battery pack. Looking at the pinouts for the Teensy 3.6 and Teensy 4.1 along with the 2-wheel and 4-wheel schematics, I developed the following pin assignment spreadsheet:
In the above (hopefully self-explanatory) spreadsheet there was only one significant conflict. On the current 2-wheel robot pins 29-32 are occupied by the four Hall-effect wheel encoders, and on the current 4-wheel robot these pins are used by four of the six battery charge indicator LED’s, and that function carries over into the new vision-enhanced robot. Easy enough resolution, just move the wheel encoder pins to 7-10 (actually even easier, I don’t need to move the encoders over until I actually have software/firmware that needs them, and that might be a while).
I started out thinking I would 3D print another ‘2nd deck’ plate so I wouldn’t have to disassemble the original one. Then I came to my senses and realized that I had originally constructed the 2nd deck expressly so it *could* be disassembled – duh! Here’s the mostly disassembled 2nd deck plate:
