This is used for Beaker's "high level" control, in particular any neural network control algorithms are stored here.
This is used for Beaker's "low level" control. It can take commands from the "high level" Raspberry Pi. It handles all the details of Beaker's physical components, like input sensors, telemetry transmission, and motor control.
Since Beaker utilizes both an Arduino and a Raspberry Pi, code is broken up into two main directories named (drum roll...) arduino and raspberry_pi.
The bulk of Beaker's Arduino code is housed in the arduino/lib
directory. Here you will find literally all functionality wrapped in classes. Each lib has a header comment explaining the libs role.
The controllers realize balance and basic functionality. There are some oldies but goodies, like pid_control. Each controller
has a header-comment explaining what it does. Of note is the rPI_control
, which is the controller that acts as a middle-man
between the Raspberry Pi and Beaker's body. This is the controller used for any neural-network control.
When working with any of the non-neural network controllers, such as PID, you can easily interact with Beaker wirelessly via the
raspberry_pi/terminal/arduino.py
script. Once the arduino app is loaded and running, ssh into the pi, run the script and type
H
to see a list of options for how to interact (and tune) the controller you are playing with.
The system checks directory holds a bunch of arduino scripts that allow the developer to interactively test out the functionality of Beaker as well as the low-level libraries that makes Beaker work. For example, there is an 'encoders' script that lets the developer turn the wheels (by hand) and see in the serial output that the encoders are working. It's important that the underlying code works before attempting to lay on more complicated algorithms. The README in System_checks has some suggestions about the ordering in which someone performs the checks. Note that these are not needed to run before each use. They are there in case things go wrong, and you want to find out where exactly in hardware/software things are failing.
This directory holds the python-based raspberry pi scripts. These are primarily Tensorflow-based neural-networks
that usilize the rPI_control
arduino app mentioned above.