A couple of weeks ago Tridge mentioned an idea: It would be great if droneshare scanned uploaded tlogs for parameters that are outside of the recommended range.
I've just now gotten around to adding this feature, so now any time you upload a flight via the webpage or Andropilot it will check the parameters against the recommended values in the APM source code. If it sees a parameter outside of the recommended range it will show a warning.
If you see a warning for one of your vehicles, it is probably best to ask in the appropriate DIY drones forum. It may be that the value you are using is fine and the comment in the source code is just wrong - I don't think these ranges have been used much in the past, so some errors are possible.
Here's a typical flight that has a warning.
Also, I can now happily report that a few hundred new flights are being uploaded each week.
Comments
Hi Gary,
The ranges come from recommended min/max values that have always been stored in the APM source code (but I think not used before - not sure)
@Nathan - yep at some point we (me plus other devs hopefully) will start grouping by self identified vehicle types (bixler 2 etc...)
Hi u4eake, Alas - the web UI is missing the sign in feature (still). I'll add it someday. Currently the only way to sign in is with andropilot.
how can I find my uploaded flights back ? I only seem to be able to see the last uploaded flights.
This could become a great way of sharing PID settings..
Think its a great start, next step should be to show the value ranges where a parameter is outside it..
i.e. LOITER_LON_D my value is 0.015 which showed up in red.. but why, it is too high, too low? what would be a more sensible value.
Great work :o)
That is a very cool idea, are you creating a range of user settings from the data? By that I mean something like, for 60 inch high wing trainers the range X to Y seems to work?
How long until we can hand fly something and record all the info and the system itself have a think about it and create its own settings? (fixed wing) Reverse engineering.
great, thanks Kevin!