After seeing the bird defecating on their kitchen counter, I decided this would be the perfect thread to ShowHN my latest creation: a 3 axis gimbal stabilizer for those not lucky enough to have cooperative avians. It's electromechanical, not biological, so don't get your hopes up.
FYI the audio is distracting, consider blanking it out and replacing with silence.
The gymbal is pretty effective, do you use a 6DOF system to provide the feedback? I have a system that I had hacked together with a 3 axis accelerometer and never could get it critically damped to the point where you wouldn't see some oscillation in the view. I speculated that if you had a gyro as well you might be able to use the magnitude of the gyro signal to tune the feedback loop with the accelerometer such that you could keep it critically damped over a wider variety of changes in orientation.
There are other such systems as you describe using 6DOF (like Alexmos, used in many DIY projects and commercial handheld gimbals), but none get to this level of performance. In particular these systems develop a lot of horizon drift under lateral loads affecting the accelerometer.
Our system can be configured to use up to 19DoF (that's 9DoF x 2 sensors, and a barometer), which is partially the reason for the performance you see here.
One software suggestion. Have an option to put in an orientation tracking factor so you can have the PID loops slowly track the aircraft if desired. In some cases you want a perfectly level flight, but in others you want to have the feeling of flying smoothly with banks and all. It would allow you to dampen out all the craziness of quad flight, but still see majestic turns and such.
Oh yes, you're talking about the dreaded "GoPro Jello". At specific vibration frequencies, the entire image begins to move around. We're doing lots of signal processing to counteract that.
It's my middle finger to the spambots. I figured they already have workarounds against all the permutations of '- at -', but no way they've thought of that one.
Funny, this owl video shows that when blindfolded, the owl's vestibulo-ocular abilities are undiminished. I was wondering about that when the chicken video guy was speculating about blindfolding to prevent the chicken from rapidly changing targets. I guess it would actually work. Seems kinda mean though.
There's a big flaw in the NPR writeup. The research is on head tracking, but the writeup leads with an example of head + eye tracking, which is a much more difficult-to-test but more important hypothesis.
This… this is almost certainly why chickens' heads track more "smoothly" than humans. We don't need our heads to track; our eyes can with much less effort.
I wonder how this would have helped 'Mike the Headless Chicken', a chicken that survived for years without a head, being fed with a pipette and raking it in at freak shows:
I actually thought this was going to be an article about whether people with a smaller risk appetite have more stability, but nope, really is about chicken.
https://vimeo.com/81292853
The video shows a synchronous split screen of the very same flight, with an unstabilized GoPro on left half and a stabilized one on the right half.
(other than the aircraft and the camera, everything there is self developed: firmware, hardware, gimbal mechanics)
Some more videos here: https://vimeo.com/user22741569/videos