We all know that using IMU to improve GPS pos hold is already working in various controllers and has a great impact on accurateness of PH .
The idee is that, long befor the gps can sense the positioning error, the accelerometer can sense if a lateral (X-Y) acceleration occur and counteract it by leaning in opposite direction.
This add-on inertial function complements the actual GPS position hold.
The function should filter X-Y acceleration projections on horizontal plane with 50-100 ms sampleing period or so.
Then using a P-I controller, create a counter-acceleration in opposite direction.
That's all.
Inertial position hold improvment.
Re: Inertial position hold improvment.
start was here viewtopic.php?f=8&t=2503
Some weeks ago, I got X,Y acceleration vectors independent from inclination (by using matrix rotation). And tried to integrate this to main PID regulator and even fly with this
The idea is working (when you push copter horizontaly it tries to compensate and return it back) but accumulate errors and it becomes uncontrollable quickly... these X,Y vectors should be rotated to global system coordinates to correct by GPS and find bias...
Some weeks ago, I got X,Y acceleration vectors independent from inclination (by using matrix rotation). And tried to integrate this to main PID regulator and even fly with this
The idea is working (when you push copter horizontaly it tries to compensate and return it back) but accumulate errors and it becomes uncontrollable quickly... these X,Y vectors should be rotated to global system coordinates to correct by GPS and find bias...
Re: Inertial position hold improvment.
Hey, wellcome back. I don't have complete expertize in this field but some ideeas :
If we integrate acceleration, we get speed (You remembered me that one from alt hold ) . So we do another PID controller witch tries to keep speed = 0. If the PID parrameters are well tuned, it should not oscillate. Also there should be some fusion of responses between GPS and ACC integrator. Anyway, i saw a large topic about that, i was thinking the ideea died, but i hope it still presents interest.
If we integrate acceleration, we get speed (You remembered me that one from alt hold ) . So we do another PID controller witch tries to keep speed = 0. If the PID parrameters are well tuned, it should not oscillate. Also there should be some fusion of responses between GPS and ACC integrator. Anyway, i saw a large topic about that, i was thinking the ideea died, but i hope it still presents interest.