PID control during planning

Original report by Guillermo Zaragoza Prous (Bitbucket: GZaragoza, GitHub: GZaragoza).

Hi,

I am trying to implement a DirectedControlSampler derived class with a PID control inside it, so the control planning algorithm uses it during the execution. As it releases different controls for short periods of time (in order to reach the destination state), I cannot store it in a control class (not even in a compound class I think) because I need to save also the number of steps for each of the controls.

I have started to add one dimension to the control class, so I can store the time for each control within the compound control object, but it won't fix the issue because when the algorithms, such as RRT or KPIECE1, call the propagationWhileValid method, its implementation is not designed for it.

So, my question is if there is anything thought for the use of PID control within the library or should I implement a new propagationWhileValid method. My ultimate goal is to compare the algorithms' performance with random controls generation and deterministic controls generation.

Thank you in advance.

Start by reading DirectedControlSampler, the control and compound control classes, and propagationWhileValid, then trace how RRT and KPIECE1 invoke propagation. The issue does not identify files or tests, and the desired API for variable-duration PID controls is unresolved; completion would require an agreed design and validation of PID control during planning.