I'm attempting to build a 9kg hex that can drop 6 beers on parachutes to people floating between houseboats on a lake.
I need your ideas: How can I adjust the tuning parameters in-flight to account for the loss of 6 x 400g beers, which is 25% of the AUW?
The beers are released by the rotary latch release pictured above. I can mix (copy) this to channel 6 and use the channel to vary something like CH6_RATE_KP along the range of the drops, but this doesn't let me also adjust other valuable params such as stabilization P.
Safety is a primary concern, and I won't fly over my floating friends unless it's rock solid. Safety prompted the heavy structure that is now my challenge. I have a large blade guard superstructure that sits on the floats and serves as a frame for a protective net of thin UHMWPE thread.
It's been tough recognizing the right PIDs no matter how many videos I watch of other people tuning, or how much I practice on Jason's PID sim. My thrust-to-weight ratio is 2.0 with four beers; the flight above was with two.
Compassmot result is <1% and vibration looks OK, right?
I'll skip detailing my design until it's really working, but thanks to everyone for the ideas and inspiration, particularly Robert's offset motor build thread.
I have three days left, so I'm looking for software tuning ideas. I don't have time to reduce the moment of inertia or re-engineer a better frame. I'll leave it at home if it's not ready for safe flight.
Thanks for your help and ideas!
Comments
I am polish so my advice is forget about beer - use vodka instead. Same stopping power with less volume and mass. One 0.5L bottle of vodka is more or less equal to your six cans of beer in terms of alcohol content.
APM only works with scissors and WATER bottles ask Marco about that :) .
Ilya - Love the I term idea. I'll experiment with that. Certainly it should help with COG and throttle, but I'm guessing it might not help with wobbles.
earthpatrol & John - The parachute deployment has been really successful actually! 10 releases, 10 deploys, no snags. The beers are in neoprene "koozies" and the chutes slow them to about 10 ft/s. They take about 20 ft to open. Given that our default is throwing a naked beer at them from the third story of the houseboat, the parachute part is a huge safety improvement :)
There's a downward-facing cam for the "bomber" to use FPV goggles. More testing definitely needed. I'll be tuning as fast as I can charge batteries!
Although a cold one arriving by parachute does sound pretty awesome.
Sure wish I had an idea for your PID issues but watching the video and given your time constraints, I would opt for more testing before unleashing payloads on your friends in the next few days. If a shoot fails on the beers, and say you drop them from 30m/100ft. That's about 120 Joules of kinetic energy per can "beer"ing down on them at about 45 mph just prior to impact. Maybe instead of dropping them, you could land, flip your switch to release two beers, and then fly away leaving two beers behind? Good luck and looking forward to seeing the design evolve.
Here's an idea... The integral term is supposed to take care of an imbalance that is there for an extended amount of time. The default integral terms are probably way too low to handle the weight imbalance created by dropping a beer can. A quick hack might be to try upping I on the rate controller as well as the max value of the integral term, but it could lead to other weird behaviors, and might not react quickly enough.