I've mostly given up tuning my nav pid... I have no telemetry yet and my ardupilot isn't very easily accessible, so trial and error is very tedious, and I've never seen D explained in a way I could understand it.
My plane flies to waypoints, but it does not come anywhere close to tracking along the centerline of my runway, so I am very afraid to try an autonomous landing. Part of my goal for my project is a reliable auto T/O and landing on a runway in a reasonable crosswind.
Does anyone have a guide to tuning the NAV PIDs? I've read the wiki, it has a very brief section on it that doesn't seem to contain a coherent procedure.
Replies
It's also hard to judge how well the plane is tracking when the track is being set from where the plane hit the waypoint radius, rather than from where the waypoint itself. The trouble I'm having implementing a fix is doing it so that it works properly when a jump command is being used.
I guess I really need to set up a simulator to tune pids and test code changes, but the only machine I have windows on is my netbook (I installed it specifically for running APM.) I also don't think there would be a flightgear or x-plane model for my e-flite apprentice.
Funny you should mention that. We just revised the manual's Nav tuning guide today. Check it out here.
It's really hard to tune PIDs without telemetry, but really easy with it. The best way by far is just to put the plane into a square waypoint pattern and play with the parameters in the Mission Planner. Without telemetry it's really time-consuming to have to land the plane every time to change anything, as you've discovered. Those Xbees are a good investment!
Ignore the D term; it's not necessary.