for linear systems modern control theory is already optimal (?) so given a linear system I'd assume an RL method would just approximate the optimal control meth

for linear systems modern control theory is already optimal (?) so given a linear system I'd assume an RL method would just approximate the optimal control method. I think the potential for RL methods is in capturing nonlinearities in actuators like artificial muscle.