autopilot route plugin

[rgleason]

I've cloned sean's version and made a comparison.
A number of src files have changed and several functions removed like intercept and boundary avoidance which you mentioned somewhere.
It won't be difficult to catch this repos up to those changes.

Also Sean's plugin is using API111 and this one is using 116.

Also I have used standard libs used in OpenCPN for nmea0183 and plugingl which are slightly different. I do not know how much of a difference that will make in course keeping. Perhaps Sean can advise.

Marco, will you be available to test the plugin for us?


Also I think that perhaps I should check and compare pypilot plugins? What do you think?

[ironman]

Originally Posted by rgleason

Marco, I am going to include your document in the Plugin Documentation as a pdf attachment.

Feel totally free to do so, but I might yet make a few more iterations. I'll keep you posted. Be aware that it is quite an involved thing to test -both at the desk and on the boat- so I'm learning as I go and I might still be wrong at some points.

I have deliberately not yet raised the subject that this plugin seems to have been tested with pypilot only. I do have a raymarine st2000+ I could test this with, but I'd like to have at least the confirmation that there is one more person in the world who would be interested to use this for real.

[ironman]

Originally Posted by rgleason

Marco, will you be available to test the plugin for us?

Also I think that perhaps I should check and compare pypilot plugins? What do you think?

Yes will be happy to test.

At the boat I use the pypilot plugin straight from github, but using the PIM/Windows version occasionally at home I have not mentioned any major discrepancies yet.

[bcn]

Originally Posted by ironman

I have deliberately not yet raised the subject that this plugin seems to have been tested with pypilot only. I do have a raymarine st2000+ I could test this with, but I'd like to have at least the confirmation that there is one more person in the world who would be interested to use this for real.

Ironman...


a recent post that might be in the scope where you are searching:
https://www.cruisersforum.com/forums...ml#post3537499

[Dave Wright]

...I'm late to this party, but very much interested in using and improving the RouteAutopilot plugin..!

I've been working for quite a while on a totally home-brew AP as a framework to push myself up the Python learning curve (PyPilot came along after I started, I looked at the code and set it aside until I could grok it). At this point, an RPi is processing input from an AHRS (stabilized magnetic heading and attitude), running SimplePID, and controlling a vintage Neco drive to the needed rudder angle to keep us on heading. All good there...

NMEA strings are generated from the AHRS and sent via AdHoc wifi to a MacBookPro 10.14.6 and OCPN 5.2.4+6b314e6 which displays heading, rudder angle, etc...

Just this week I got my RPi heading servo to accept the APB from OCPN, and did some testing on the river. Initial results were very encouraging, until I arranged the test Route across the current...

As you can imagine, I got the expected 'fishhook' tracks as we fell below the route but doggedly maintained heading.

So I started looking closely at how OCPN computed heading-to-steer, and thinking about how to solve this. And today I discover APRoute. Just installed and enabled, and looking forward to testing on the water.

So yes gentlemen, you are doing good and needed work, carry on..!

[ironman]

Good story, interesting project - would be intesting to see some more details.

Originally Posted by Dave Wright

... until I arranged the test Route across the current... As you can imagine, I got the expected 'fishhook' tracks as we fell below the route but doggedly maintained heading

To correct an increasing XTE resulting from a cross current would ideally require a separate PID loop just to stay on track. To tune such a loop would be a demanding job. Even pypilot just internally uses a simple P gain only to calculate a correction angle from an XTE. That's why this Route Position Bearing principle is so elegant - it requires one parameter only, the guide radius. Still, with a typical radius between 25 and 100m, it can not be recommended for negotiating a narrow creek in cross currents - if ever.

[rgleason]

Marco, last night I started updating my repos with Sean's more recent code, and ran into this trying to build

pidc.obj : error LNK2019: unresolved external symbol __imp__gluDeleteTess@4 referenced in function "public: void __thiscall piDC:rawPolygonTessellated

I had to make a number of changes to the plugingl files, not sure if they are correct. Will be working on it...

[ironman]

Originally Posted by rgleason

Marco, last night I started updating my repos with Sean's more recent code, and ran into this trying to build. I had to make a number of changes to the plugingl files, not sure if they are correct. Will be working on it...

Was out for some days. This is beyond my knowledge - can not help you I'm afraid.

[ironman]

Originally Posted by ironman

I made a bit of a braindump on the pypilot forum; this might give some insights as to usage is concerned. https://forum.openmarine.net/attachment.php?aid=1636

New version of this document available at https://github.com/pypilot/workbook/...CPN.plugin.pdf with some incremental insights.

[Dave Wright]

I'm going to be doing on-water testing of the APRoute in the coming days. Presently running macOS 10.14.6, OCPN 5.6.0 plugin API 117, and APR 0.4.4.0. I've noticed some bugs in behavior that may be already dealt with but not updated and pushed to PIM. Also unclear on how PIM handles updating an installed plugin. So before I start compiling a buglist for my particular installation please let me know if/when a new version of APR is coming to PIM, and how best to help. New to the forum, so learning the way... Thanks.

[ironman]

For now I would think you would be better off compiling version 0.5 straight from https://github.com/seandepagnier/autopilot_route_pi, as this version has a few issues already resolved. In this github repository you will also find a couple of still openstanding issues documented. Have fun!

[Dave Wright]

Don't have the toolset to compile, or that level of competence at this point. Will check the buglist and see if I have anything to add, but will have to await a push to PIM before I can test...

[ironman]

rm -Rf autopilot_route_pi/
git clone git://github.com/seandepagnier/autopilot_route_pi.git
cd autopilot_route_pi/
mkdir build
cd build
cmake ..
make
sudo make install

[ironman]

Never compiled anything on Mac, but I was able to get it compiled on my wife's macbook, yet I cannot get it to register with opencpn:

• On Mac, installed git, cmake
• Installed gettext with brew install gettext (first installed brew for this)
• Installed wxwidgets by downloading the tarball wxWidgets-3.1.5.tar.bz2, unzipping it, then configure and make. Didn't do make install.
• Finally compiled using the previous post, but changed the cmake

  cmake .. -DwxWidgets_CONFIG_EXECUTABLE=/path/to/wx-config

But as said, still no cigar. To be quite honest, I feel like a dilettante on macos, and I think you are better off waiting for @rgleason's PIM release, as you suggested. But you will find the issue I put in github inhibiting. https://github.com/rgleason/autopilot_route_pi/issues/6

[Dave Wright]

Appreciate your effort..! I have so many learning curves open at the moment that creating yet another development environment would have been too much. Understand that when I transitioned to hardware in the previous century I had been coding in 80C49 assembler, 68HC11 FORTH, and a little bit of C. Now, this old dog looks at Python and the head spins...

I just upped to OCPN 5.6, and some stuff is broken (the old route), so there may be some additional interactions going on, or the install / remove of APRoute may have stepped on something.

Thanks.

We will proceed as the way opens...