|
From: | Gautier Hattenberger |
Subject: | Re: [Paparazzi-devel] stabilization parameters |
Date: | Wed, 28 Nov 2012 14:24:52 +0100 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:10.0.10) Gecko/20121027 Icedove/10.0.10 |
The reference model used to filter to (auto/human)pilot inputs is a
second order equation like this: angular_acceleration + 2*damp*freq*angular_speed + freq*freq*angle = angle_input So, for each new input, we compute the angular_acceleration and integrate it twice to get the reference angle and angular speed. Moreover, the acceleration and speed are bounded. Now, to change the behavior of the model: SP_MAX: bound for angle input REF_OMEGA: natural frequency of the second order REF_ZETA: damping factor (should not be less than 0.8-0.7, maximum is 1) REF_MAX_X: bound on angular speed REF_MAX_XDOT: bound on angular acceleration The response time (95% of the final value when doing a step input): ~ 3/(freq*damp) To increase response time, you can increase the frequency, and the bounds. But if the tracking of the reference model (with the feedback gains PGAIN, DGAIN and IGAIN) is not good, it will not change that much the way it works. Also, if the sensitivity of your IMU is wrong, the estimated angles will not be correct, and the control will be either too fast or too slow. Gautier Le 28/11/2012 02:31, Eduardo lavratti a écrit :
|
[Prev in Thread] | Current Thread | [Next in Thread] |