"I have plans to design an autopilot for my boat and have read this post with great interest. Before starting I should want to ask you two questions that has puzzled me:
- I have seen that some boat autopilots does not need a rudder feed-back. I suppose that all PID regulators do need some sort of feed-back. Do you have any idea how they can do this?
- Some autopilot claims that they are adaptive. They can learn the characteristics of the boat. I know you cannot give us an adaptive controller, you told us that, but can you hint to me how this is done. Is it still a PID regulator that is used with some other program tuning the parameters or do they use some other type of controller? "
Well all closed loop controllers need some kind of feedback measurement(s) to function, otherwise it is 'open loop' control, kind of like a steering wheel, where it's up to the human operator to know where the car is aimed at and make adjustments or not.
In the Classic PID controller the feedback is called the 'PV' for process variable. It is the actual measured temperature in a temperature controller, or the actual level measurement in a level controller. For you autopilot steering the feedback would either have to be something that can measure actual heading like a electronic compass or a GPS heading, etc. Measuring the actual rubber angle would work to a degree but wind and current variables could cause a change in heading regardless of the rudder angle. I'm not sure what the standard measurement used in for boat auto pilot controllers. Interesting applications and I would favor GPS but autopilots have been around longer then GPS so there must be other effective measurements.
Lefty