[question] Math behind converting attitude errors to body rates

[ayhamalharbat]

Hi @Jaeyoung-Lim ,

I have a couple of questions:

1. What is the logic/math behind converting the non-linear attitude errors to angular velocity setpoints (as shown below)? is there a physical meaning to this conversion and to the constant attctrl_tau_?
   

   mavros_controllers/geometric_controller/src/nonlinear_geometric_control.cpp

   Line 63 in 8b3fff0

   desired_rate_ = (2.0 / attctrl_tau_) * error_att;

   
   and
   

   mavros_controllers/geometric_controller/src/nonlinear_attitude_control.cpp

   Lines 56 to 58 in 8b3fff0

   	desired_rate_(0) = (2.0 / attctrl_tau_) * std::copysign(1.0, qe(0)) * qe(1);
   	desired_rate_(1) = (2.0 / attctrl_tau_) * std::copysign(1.0, qe(0)) * qe(2);
   	desired_rate_(2) = (2.0 / attctrl_tau_) * std::copysign(1.0, qe(0)) * qe(3);

2. Do you have any practical recommendations on how to get an accurate and proper normalization of the thrust command, i.e. how to estimate the norm_thrust_const_ and norm_thrust_offset_?
   

   mavros_controllers/geometric_controller/src/geometric_controller.cpp

   Lines 408 to 410 in 8b3fff0

   	bodyrate_cmd(3) =
   	std::max(0.0, std::min(1.0, norm_thrust_const_ * thrust_command +
   	norm_thrust_offset_)); // Calculate thrustcontroller_->getDesiredThrust()(3);

Thanks in advance!

[cillian-bao]

I encounter the same problem，if the code written in this way, I dont't think it is a geometric control approach. @Jaeyoung-Lim

[Jaeyoung-Lim]

@cillian-bao I have explained why this is the case in #230