IMU Not behaving correctly.

[ironman]

Making a release tag is one thing, but testing it in all possible configurations that pypilot can possibly be deployed in is another chapter. There are not enough people in the pypilot world for system testing. Don't want to start a discussion about it, but given the complexity of pypilot it's pretty amazing the master branch has been quite stable recently. On my boat I run pypilot 0.34 though and I have no intention to update.

For now, I just downloaded the latest pypilot 0.43 and was able to calibrate the imu by turning the raspberry on my desktop here, so the master branch does not seem to be your issue. I'll tell you what my setup is:

Openplotter 3 image, updated to the latest OS patches.
just a git pull, and a sudo python3 setup.py install
run pypilot at the prompt as user pi
opencpn 5.8.4 with pypilot plugin 0.40.30.0 from the master plugin catalogue on that same machine.
in opencpn, click pypilot plugin, config: connect to 10.10.10.1, calibration: first accelerometer, then alignment: level, then go to compass, turn raspberry slowly. Got a circle in the sky off the globe, then it snapped to the globe, and the age was reset (see image).

Mind you that I did not use any of the openplotter wrappers around pypilot. As brilliant as the openplotter initiative is, it is inherently lagging behind the developments of pypilot and there are not enough rainy days in a year for me to figure that all out. Take care.

[rszemeti]

When I try to run the calibration script directly I get this:

https://i.imgur.com/7FsR3sF.png

Should I try and roll back to version 0.36 or maybe 0.34?   Any ideas hiow to do this now I installed the head from GitHub?

[ironman]

That 'pyt' must be a typo, apparently introduced 3 days ago. You can either edit /usr/local/lib/python3.9/dist-packages/pypilot/ui/autopilot_calibration.py and remove the last line, or go back to a certain commit using the code below. When I do that, I get a garbled screen, but I suspect that to emanate from my VNC connection - I don't have a proper raspberry screen.

Nevertheless, please note the positive scenario I described used the pypilot plugin to opencpn.

Code:

pi@openplotter:~/pypilot $ git checkout 787ac9
Note: switching to '787ac9'.
[...]
HEAD is now at 787ac95 fix rudder range setting
pi@openplotter:~/pypilot $ sudo python3 setup.py install

It's a bit painful, but you know what they say: no pain, no gain :-D

[rszemeti]

OK, I've done that and switched to the detached head, version 0.25 ... lets see how that goes.

[seandepagnier]

I have fixed the inadvertent pyt... Sorry I made a mistake. Could you try with master branch?

[rszemeti]

OK, that seems to work again, thank you ... and to be clear, does this version fix the bug so it will now calibrate with "IMU only" selected?

I'll have a look through the code and see if I can find the display bug ... it is still present in this version.

[seandepagnier]

I think the display bug may be a graphics card issue?