I realize that Principle 3 is often related to control theory applications, however do we have any examples of matching a given PID system to neurons? I know that in nengo_benchmarks we have adaptive control examples, but I couldn’t quite figure out where in the code a neural version of PID was being used. @brent didn’t you have a neural PID controller implemented somewhere?
If it doesn’t exist, I can create my own, but it would nice to have prior art to avoid duplicate work.
@tcstewar put some examples of PD, PID, and adaptive control here: https://github.com/tcstewar/nengo_learning_examples/tree/master/control. This nicely shows how all three are related.
What do you mean by “matching a given PID system to neurons”?
What do you mean by “matching a given PID system to neurons”?
I meant matching equations for the various components of a PID system to synapses.
Looking at the code now, I just realised that the gains on the connections aren’t mapped to synapses at all, they’re just gains on the transforms. Not sure what I was thinking.