search

byu-magicc / roscopter

*Under Development* - A fully-featured multirotor autopilot for ROS
http://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=2324&context=facpub
52 stars 38 forks source link

[WIP] pitch and roll mappings added #24

Closed wynn4 closed 5 years ago

wynn4 commented 7 years ago

These are the changes Michael made to the controller to include pitch and roll mapping. There are also a couple other changes but I didn't look into them in great detail. We should check with Michael before pulling these in.

plusk01 commented 6 years ago

@mmmfarrell @wynn4 what is the status of this? I noticed that in master the controller still doesn't use roll/pitch... but seems to work just fine.

superjax commented 6 years ago

It works, but not at high angles of pitch/roll. The whole control scheme is stupid though. The whole "model inverse" approach does a terrible job of handling large control inputs.

superjax commented 6 years ago

I will say, however, that this is exactly how we have done all of our flight tests in relative nav. If you don't supply large control inputs, it works.

wynn4 commented 6 years ago

I agree that the current control method seems lacking. Upping the gains (by a lot) on the 'w_PID' terms improved the overshoot from large changes in altitude significantly though. I committed new default gains a little bit ago that perform much better

mmmfarrell commented 5 years ago

Closing this because its really old and I added a new PR #38 that makes these changes obsolete