Why physics get unsteady at high angular velocity?
I have a problem with the accuracy of the physics simulation. Given a link with unit mass and unit moment of inertia at zero position, I apply constant external force (10 H) and torque (10 H*m) for 3 seconds. With increasing angular velocity my acceleration starts to rise! even after the force was disabled. Angular velocity behaves well. With a lower simulation time step this effect is less prominent but still present. But with even lower time step simulation is very slow. What should I do, any advice?
I use a gazebo garden and default physics simulator (dartsim).
Here is the results (upper acceleration, lower angular velocity) with max_step_size>0.0001</max_step_size>
and (10 times lower) max_step_size>0.00001</max_step_size>