I'm trying to use Flightmare(home) simulator to create a control model using reinforcement learning.
The simulator is great, but its environment accepts step command with four values of rotor thrusts in the range [-1:1].
env.step(float t1, float t2, float t3, float t4)
And the RL model output them too.
The problem is, that I can't find the way to "convert" those thrusts values to Yaw/Pitch/Roll/Throttle commands so I can send them with SBUS to Quadrotor flight controller.
I've googled for a solution for two days already and getting desperate.
I there a way/code/lib to do it? Could you point me in the right direction?