I am trying to decide between actuators that exert 40 pounds load but which can move faster and actuators which can exert 120 lbs of force but move slower. There are also tradeoffs in duty cycle. Weaker actuators can stand a 40% duty cycle or higher while stronger ones might be limited to a 25% duty cycle. I can probably either calculate duty cycle or get an extra temperature probe. I had thought about mounting the actuator right on the transom, close to the pivot, using a 12" travel actuator, and the 120 pound one to be sure the rudder
would work even though I was giving up most of the mechanical advantage in mounting.
There are two fundamentally different approaches I can see taking. One would be to use the arduino. This is a relatively simple device that has a lot of support in an active user community. It is programmed in a variant of C++ and your program is always in control. If it ends it is just started again.
To program it, you need a system that can download the code using USB. The code is stable when it is loaded and can't be changed until the USB is hooked up again.
If you want your program to do multiple things, you need to write your own, possibly simple dispatcher.
The other approach would be to use something like a Raspberry Pi. This allows you to run Linux
. You can have multiple tasks running and the system dispatcher will deal with things. As an example, one task might work the autopilot, another task could monitor
, and a third might try to deal with man overboard
monitoring. You could run openCPN
in spare cycles, say, while giving the pilot program priority.
But I have read about neural network software
, and while I have never used it, it seems like you have an easy goal to conceptualize. You want minimum divergence from your planned path and, as a second goal, minimum perturbation of the boat.
Each axis could be an input to the net. You would probably want some smoothing on the output to the rudder.
You also have a wider choice of programming language, and there are high level languages that can make patching easier.
The hardware cost would be higher. $50 for the Pi, maybe, another $50 for a display. Everything else would be the same.
But the end result could be a pilot that learned your boat and how it reacts to the helm
, and after a short period could become an expert handler.
Just thinking ahead. I have a long road in front of me.
I programmed for a living for many years. This will be a large coding task in any case. While getting something that can hold a heading probably will not be hard, (famous last words) I could see it taking a year or more to get some degree of sophistication in the algorithm.
The device I am thinking of using for compass and gyroscopic attitude measurement is the MPU-9255. It has its own on-board processor that can process attitude and direction information. I will be studying product data sheets
These actuators are typically used, say, to open a window or even the lid of a box. They require a dpdt switch because the DC polarity has to be reversed to control direction. The arduino setups I have seen use pairs of dual contact relays to let the 5v from the arduino control and reverse the 12v supplied to the actuator. I was going to look for something solid state for the relays. My gut feeling is that solid state relays will be more reliable and, well, quieter.